Home

Version 1.10 System Manager`s Guide

1. Update NIS database by var yp make on the NIS master server Run SetupClient Figure 2 10 Configure client information on the server using SetupClient NetServer Configuration Commands a 2 27 Do You Need to Run SetupClient SetupClient is required only if your installation includes diskless SunOS client workstations that boot from the server Client workstations with a local disk simply mount the exported file systems they need to access What Does SetupClient Do SetupClient adds or removes one or multiple diskless SunOS clients If the NetServer is running NIS SetupClient updates the NIS host s etc hosts etc ethers and etc bootparams databases If the NetServer is an NIS master SetupClient updates the local NIS database however it does not propagate the local update to other NIS servers If the NetServer runs NIS but is not an NIS master SetupClient issues a warning to indicate that the database is out of date Note Do not run this command unless you have already run SetupExec to i install the architecture dependent executables for the diskless SunOS client machine The executable directory for the client s application architecture must be present on the NetServer If this directory is absent SetupClient generates an error SetupClient updates the etc exports file to allow root access to each client s root file system It exports the client s swap and dump partitio
2. ust file system Figure 4 6 shows the usr file system Although most of the files in usr are SunOS files treat them as Auspex specific software Because the files in usr are not intended to run on workstations usr is defined in etc fstab as type 4 2 and cannot be exported to client machines usr 5bin 5include 5lib acrobat adm gt var adm auspex bin boot gt kvm boot contrib diag dict etc games hosts include kvm lIddrv lib local lost found man gt share man mdec gt kvm mdec net gt var net nserve gt etc nserve old openwin pub gt share lib pub sccs lib share man spool gt var spool sys gt kvm sys src gt share src stand gt kvm stand sys gt kvm sys tmp gt var tmp ucb ucbinclude gt include ucblib gt lib xpg2bin xpg2include xpg2lib Figure 4 6 The Auspex usr file system Disk Partitioning a 4 15 AUSPEX D usr openwin file system The usr openwin file system contains most of the OpenWindows 3 0 files but not the X11 NeWS server Consider these files like the ones in usr Auspex specific software they are not to be mounted and run by diskless workstations on the network Diskless Sun or Sun compatible clients can obtain SunOS and OpenWindows from the export file system where the software must be installed from distrib
3. Auspex drives are set at the factory to the highest capacity listed in Table 8 2 To operate your drive at the lower capacity name your drive following the naming conventions described in Table 4 2 and Table 4 3 on page 4 3 Note If you want to dump data to a 4 mm tape without data compression D for example if you specify dev rast1lo in dump make sure there is no compressed data prerecorded on the tape Trying to back up uncompressed data to a 4 mm tape containing compressed data generates a tape write error Always check the LED on the tape drive before issuing the dump command If the tape inside the drive contains compressed data the LED is green if the tape contains low density data the LED is amber 8 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The Siz To erase prerecorded data on the tape use the mt command For example mt f dev rastllo erase e Density and Blocking Factor Options for Dump Table 8 3 lists suggested arguments to use with the dump and restore commands for the Auspex tapes listed in Table 8 2 A Caution In the dump command do not use a blocking factor greater than 126 Doing so may cause data to be unreadable when you try to restore it For example suppose you specify the blocking factor to be 512 which means that you try to write blocks of 262144 bytes 512 bytes x 512 to tape Before the data is written each block of 262 144 bytes is broken up into smaller blocks
4. Nov 17 17 47 36 server_name vmunix adl0 fatal error Permanent drive error on write Unrecoverable Disk or File System Problems in Multiuser Mode a 6 9 Note If an unrecovered error occurs on a drive that is mirrored the SP can i complete the disk I O request on the other member of the mirrored partition In this case no error messages are generated from the application program the FP or vmunix The SP however prints the following warning message Warning a mirrored partition has a dirty or damaged component vp The SP also displays other error messages to indicate why the virtual partition is dirty If you see an error message from an application program the FP or vmunix but not the SP it is likely that a file system problem not a disk error occurred If an unrecoverable error happens to an LFS mounted file system the FP takes the file system offline so you can fix the problem The file system isolation feature is discussed in the next section File System Isolation A file system is considered isolated when the FP takes it offline after detecting hardware or software problems Isolating an unreachable or corrupted file system gives you time to repair it while allowing clients of other file systems to continue using the server unaffected Only an LFS mounted file system a file system managed by the FP not HP can be isolated The error conditions that trigger file system isolation are described in
5. restore imvf dev nrast4c Tape File 1 Tape File 2 Tape File 3 k a a End of file End of tape Figure 8 3 1 5 1 tape encoding Tape File 1 Tape File 2 Tape File 3 End of file End of tape Figure 8 4 1 9 tape encoding 8 18 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Common Tape Drive Errors This section describes common errors you might encounter when using a tape drive to back up or restore a file system The errors are categorized into the following types a Operator errors a Media errors a Tape drive errors This section gives either a sample command that causes the error or describes the scenario in which a tape drive problem is detected It also lists the error messages or symptoms and provides suggestions for fixing the problem Operator Errors This section discusses errors caused by the person using the tape O Error Command Error messages Analysis Solution Write Error Command Error messages Analysis Solution Blank Tape Error Command Error messages Analysis Solution tar cvf dev rast2 vmunix ast2 tar no tape loaded or drive offline dev rast2 I O error Tape drive does not exist or the tape is not loaded Specify the correct tape drive or load tape in the drive tar cvf dev rast2 vmunix ast2 tar tape is write protected dev rast2 Permission den
6. When enabled and in the event of a system crash the corescreen sh 8 command prepares and mails a subset of Processor board cores to Auspex Customer Service When enabled mailmessages sh 8 prepares and mails reboot messages to Auspex Customer Service The ax_admmsg 8 command searches for Auspex error messages and generates an error report The ax_config 8 command accesses configuration date and generates a configuration report The ax_load_flash 8 command downloads Flash PROM firmware to specified processor boards Man pages for getcores sh reporter sh corescreen sh mailmessages sh ax_admmsg ax_config Flash PROM Download Utility on page 10 7 Other commands unique to Auspex systems rc 8 is modified to download and start the Auspex processors The ax_write_cache 8 command provides the system manager with a tool for managing and controlling the write cache To use the write cache feature you must install the Auspex Write Accelerator on the Storage Processor dump 8 is modified to allow you to specify the time and file system name to be recorded in etc dumpdates This change ensures that the appropriate information is recorded in etc dumpdates when you dump a partition created by ax_clonefs halt 8 and reboot 8 are modified to include the s option which causes the system not to run sync 2 at reboot With this change reboot can complete even if the NetServer has a file system mounted from
7. 1108 1 DMA channel d transfer error d 1 ufs_getdirattrs dir reclen not word aligned 1113 2 ae d resetting lance 1114 2 ae d No carrier transceiver cable problem 1115 2 ae d Transmit retried more than 16 times 1116 2 ae d Transmit late collision net problem 1117 2 xdr_long FAILED 1118 2 xdr_u_long FAILED 1119 2 xdr_u_short decode FAILED 1120 2 xdr_u_ short bad op FAILED 1121 2 xdr_bool decode FAILED 1122 2 xdr_bool bad op FAILED 1123 2 xdr_opaque decode FAILED 1124 2 xdr_opaque encode FAILED 1125 2 xdr_opaque bad op FAILED 1126 2 xdr_bytes size FAILED 1127 2 xdr_bytes bad size FAILED 1128 2 xdr_bytes out of memory 1129 2 xdr_bytes bad op FAILED 1130 2 xdr_enum dscmp FAILED 1131 2 xdr_string size FAILED 1132 2 xdr_string bad size FAILED 1133 2 xdr_string out of memory 1134 2 xdr_string bad op FAILED 1135 2 xdr_array size FAILED 1136 2 xdr_array bad size FAILED 1137 2 xdr_array out of memory E 12 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1138 2 xdr_mbuf long crosses mbufs 1139 2 xdr_mbuf putlong long crosses mbufs 1140 2 xdrmbuf_getmbuf failed 1141 2 bad auth_len gid d str d auth d 1142 2 ku_fastsend error s dest IP address s 1143 2 svckudp_send xdr_replymsg failed 1144 3 s fs dirty fsck or mount read only 1145 1 mbuf_to_fc_sattr x_handy lt nfss
8. Number of NFS version 2 calls over UDP Number of NFS version 2 calls over TCP Number of NFS version 3 calls over UDP Number of NFS version 3 calls over TCP Number of puts over FTP Number of gets over FTP Number of put bytes over FTP Number of get bytes over FIP Number of FIP workers Number of UDP dgrams received Number of UDP dgrams sent out Number of UDP checksum errors Number of TCP segments received Number of TCP segments sent out Number of TCP segments with errors Number of active connections Number of active open TCP Number of passive open TCP Number of TCP workers Displaying Performance Data in Histograms a 9 41 NP ip TCP pkts in NP ip TCP pkts out NP ip TCP pkts w error NP ip TCP pkts forwarded NP e packets sec NP e bytes sec NP e incoming bytes sec NP e outgoing bytes sec NP e incoming errors NP e outgoing errors NP e outgoing collisions NP e outgoing discards NP e outgoing error percentage NP fddi Incoming Outgoing Frames sec NP fddi Incoming Outgoing Bytes sec NP fddi Incoming Frames sec NP fddi Incoming Bytes sec NP fddi Incoming Frame Errors sec NP fddi Outgoing Frames sec NP fddi Outgoing Bytes sec NP fddi Outgoing Frames Errors sec NP fddi Max transmit frame size mtu NP fddi Outgoing error percentage Number of TCP packets received Number of TCP packets sent out Number of TCP packets with errors Number of TCP packets forwarded
9. Number of packets processed by the Ethernet interface Number of bytes processed by the Ethernet interface Number of bytes received by the Ethernet interface Number of bytes sent by the Ethernet interface Number of errors received by the Ethernet interface Number of errors sent by the Ethernet interface Number of collisions on the Ethernet interface Number of outgoing packets discarded by the Ethernet interface Percentage of errors sent by the Ethernet interface Total number of packets received and sent on the FDDI interface Total number of bytes received and sent on the FDDI interface Number of bytes received on the FDDI interface Number of frames and errors received by the FDDI interface Number of frames sent by the FDDI interface Number of bytes sent by the FDDI interface Number of frames and errors sent by the FDDI interface The MTU value on the FDDI interface Percentage of errors sent on the FDDI interface Number of packets processed by the Ethernet interface Number of bytes processed by the Ethernet interface NP E packets sec NP E bytes sec 9 42 a Version 1 10 System Manager s Guide Number of frames received on the FDDI interface AUSPEX W W AUSPEX NP E incoming errors NP E outgoing errors NP E outgoing collisions NP E outgoing discards NP E outgoing error percentage NP atm packets sec NP atm bytes sec NP atm incoming bytes sec NP a
10. Tape media errors are harmless but they slow down write operations They also reduce the tape capacity Whether or not you need to take corrective action when ECC errors occur depends on the number of errors indicated in the vmunix message If fewer than 20 errors occur per megabyte of data transferred you can ignore the error message If more errors occur and other error messages appear we recommend using a new tape or cleaning the drive heads If you continue to use the tape remember that the actual tape capacity is reduced because of ECC errors and you probably need to change the backup script to ensure that the tape does not reach the end before the backup completes Recommendation Because tape media errors are more common when the tape drive heads are dirty clean the 8 mm drive heads once a month or every 30 hours of use the 4 mm drive heads every 8 to 10 hours of use once a week at most and the DLT4000 when the Use Cleaning Tape LED is lit For the 1 4 inch drives clean the heads every 8 hours of use Also keep the tape drive door closed when not in use to keep out dust Refer to the NetServer s hardware manual for more information on the type of cleaning cartridge for your drive 8 20 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Write Error Command dump Oucbsdf 100 6700 65000 dev rast2 dev ad2c Error messages vmunix ast2 22 recoverable errors per MB transferred vmunix ast2 the tape may be
11. ael flags 43 lt UP BROADCAST RUNNING gt E 4 a Version 1 10 System Manager s Guide cache ON Ys cache CON A art virtual download boards and start daemons lfs file systems partitions start more auspex daemons allowing local netmask ffffff00 broadcast 123 45 678 0 AUSPEX W W AUSPEX inet 123 45 678 1 netmask ffffff00 broadcast 123 45 6784 0 ether 0 0 3c 0 a7 91 ae2 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 ether 0 0 3c 0 a7 92 ae3 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 ether 0 0 3c 0 a7 93 ae4 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 9 ether 0 0 3c 0 a7 94 ae5 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 ether 0 0 3c 0 a7 9 ae6 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 ether 0 0 3c 0 a6 b0 ae7 flags 43 lt UP BROADCAST RUNNING gt inet 123 45 678 1 netmask ffffff00 broadcast 123 45 678 0 ether 0 0 3c 0 a6 bl running routing daemon Duo running SNMP agent mount vat nfs starting additional services biod starting system logger starting local daemons auditd sendmail xntpd ax_lfsd
12. on page 5 18 ax_mdetach 8 detaches a member partition from a mirrored virtual partition that currently has two members It is typically used for removing a member that was added with ax_mattach for taking a snapshot of the partition The change is automatically recorded in etc vpartab Note Before version 1 8 software the new member was manually removed i in etc vpartab Now that it is done automatically do not manually edit etc vpartab Editing the etc vpartab file to remove a member partition such as vp6 from the mirrored partition vp41 and then running ax_mdetach returns a warning message similar to the following WARNING etc vpartab out of sync vp6 missing from vp41 verify vp41 state with ax_vpstat and update etc vpartab When this warning occurs ax_mdetach detaches the member but does not edit etc vpartab ax_mrestore 8 updates a member of a mirror after a disk or media failure affected the member This command runs automatically if the system detects that the two members are not identical at boot time or when ax_mattach is run You may occasionally run ax_mrestore manually from either to back up a mirrored partition to another drive or to update a damaged member Be aware of the following when you use ax_mrestore ax_mrestore may delay the server s responses to other commands For example an Is or find command may take longer if ax_mrestore is in progress The longer response time is more obvious when mu
13. s Guide W AUSPEX FORMAT MENU disk select a disk type select define a disk type partition select define a partition table current describe the current disk format format and analyze the disk repair repair a defective sector show translate a disk address label write label to the disk analyze surface analysis defect defect list management backup search for backup labels quit format gt t The format program usually automatically recognizes the type of the disk selected The exception is when format fails to recognize a valid label on the drive In this example ad4 is a 4 GB disk For a complete description of the options in the format menu see Sun s System and Network Administration The following sections describe only the options that display menus different from those on a Sun server Caution Before using the format command to change the size of a physical A partition or format a disk verify the partition or disk is not a member of a virtual partition The format command may not issue a warning message when you attempt to format a disk that is part of a virtual partition Defining a Disk Type The type option in the format menu displays a list of drive types from which you choose one appropriate to the disk to be formatted While the menu lists disks from various vendors Auspex only supports Hewlett Packard drives Seagate drives and Micropolis drives The type option displays follo
14. 96MB memory installed Hostid ffffffff Part Number 10 0107 Rev CD Serial 3 AUSPEX VERSION 2 9_HP SUNMON 09 12 96 10 58 Initialize 96 MB of Memory Completed 4 Test 0 MB of Memory Address Test Completed E 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX ps EPROM boot sp status 0 7 sp found in Boot device slot 0x4000 bootblock loaded sp status f0 sp status f0 root on ad0a fstype 4 2 Boot vmunix Size 1122304 229176 85512 bytes ml6u_npages_dvma 358 ml 6 u_net_interfaces 15 VAC ENABLED in COPYBACK mode Auspex 1 9 SunOS 4 1 4 AUSPEX1 1 Tue Nov 5 04 41 40 PDT 1996 Copyright 1989 1996 Auspex Systems Inc Copyright 1983 1994 Sun Microsystems Inc cpu ASPX Auspex SPARC Processor mod0 Ross RT625 mid 2 mem 98304K 0x10000000 avail mem 88797184 dmaQ at SBus slot 1 0x400000 espO at SBus slot 1 0x800000 pri 3 e zs0 at obio 0x200000 pri 12 SPO at Auspex VME slot 12 VME address 0x0 SP1 at Auspex VME slot 13 VME address 0x0 SP2 at Auspex VME slot 14 VME address 0x0 IOPO at Auspex VME slot 5 VME address 0x10000000 memory 256 MB IOP1 at Auspex VME slot 6 VME address 0x20000000 memory 256 MB IOP2 at Auspex VME slot 7 VME address 0x30000000 memory 256 MB ad0 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt acdl lt TOSHIBA CD ROM XM 3301TA 0272 gt ad2 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad3 lt Seagate
15. Also if a station from another vendor cannot handle the maximum MTU size set MTU to 1500 FDDI Offnet MTU X Console NIS service type Domain Name Size of the maximum transmission unit for FDDI traffic addressed to a network not directly attached to the server This value is tunable for each FDDI interface on the server Specifies whether an X terminal is used as the system console Type of NIS service the NetServer is providing to clients if any NIS domain name This field appears only if type of NIS service is Master Client or Slave If type of NIS service is NONE this field does not appear 1500 or 4352 default Because some FDDI to Ethernet routers cannot handle packet fragmentation effectively if this value is set to 4352 these routers might drop packets that originate from the NetServer Setting FDDI Offnet MTU to 1500 allows these routers to route packets between FDDI and Ethernet networks without fragmentation while achieving the highest possible speed for FDDI to FDDI traffic Yes or No default Master Slave Client or NONE default NetServer Configuration Commands a 2 15 Table 2 3 The NSconfig form fields Continued Field Description Possible values Create Default Route Run RouteD System Contact System Location Change Date and Time Change Time Zone Notify Auspex on reboot Set up a default route Yes displays a field in which to specify a gatew
16. Architecture application 2 22 FMP 1 2 hardware 1 5 kernel 2 22 software 1 10 Sun system 2 22 Architecture dependent executables in export exec 4 11 4 17 installation 2 21 location of 2 28 arp command 10 15 ARP table displaying or flushing 10 14 ask me option in boot command 3 6 astn 1 13 ATM 1 8 Attaching a member to a mirrored partition 5 10 Auspex environment availability 1 11 client traffic 1 11 consolidation 1 11 expandability 1 11 features 1 11 auspex manual page 1 16 Auspex Performance Monitor 9 1 to 9 48 Auspex processor error messages E 8 Automatic file system mapping 4 21 ax_add_device command 1 13 4 27 5 11 6 4 6 5 ax_admmsg command 1 15 ax_arp command 1 12 10 14 ax_clonefs command 1 14 4 27 4 32 4 33 8 2 8 14 8 15 ax_config command 1 15 10 2 how to extract configuration information 10 2 interactive and noninteractive modes 10 4 ax_diskconf command 1 13 4 26 4 31 5 10 5 23 6 1 8 13 ax_errd daemon 1 12 ax_expand command 1 14 5 10 5 15 5 18 ax_fddistat command 1 12 ax_fsutil command 1 14 6 14 ax_hot_plug command 1 13 4 27 4 32 4 33 6 4 6 5 6 6 ax_isolated command 6 12 ax_isolated daemon 6 12 ax_keyenvoyd daemon 1 12 ax_kill command 1 14 6 15 6 16 ax_label command 1 13 4 7 4 11 4 36 4 41 ax_lfsd daemon 1 12 ax_load_flash command 1 15 10 7 ax_loadvpar command function of 1 13 5 10 invoked by ax_mconvert 5 20 ax_lslabel command 1 13 4 42 ax_mattach co
17. CPU utilization on the SP Number of I O operations NVRAM overrides NVRAM coalesces Number of page size 8K buffers used in NVRAM Number of page size 512B buffers used in NVRAM Total operations per second in the virtual partition Number of sectors read on a selected disk Number of sectors written on a selected disk Number of read operations on a selected disk Number of write operations on a selected disk Total number of operations on a selected disk Total amount of time spent on each operation on a selected disk It is the sum of queue time and service time Time an operation request waits to be serviced Amount of time needed to service an operation request Percentage of utilization on the drive Number of sectors read on all disks Number of sectors written on all disks Number of read operations on all disks Number of write operations on all disks Total number of operations on all disks Displaying Performance Data in Histograms a 9 45 SP all disks total_time_per_op Total amount of time spent on operations on all disks It is the sum of queue time and service time on all disks SP all disks queue_time_per_op Amount of time that operation requests for all disks wait to be serviced SP all disks service_time_per_op Amount of time taken to service operation requests on all disks SP all disks drive_utilization Percentage of utilization of all disks Starting ax_perfhist with Options and
18. For more detailed information about the tables see the commented text in etc format dat O a N wo D gt o D N o Size in MB x 1000 Figure 4 1 Partitioning a 9 GB disk using the default partition table O f gt os a N w D gt o D N o Size in MB x 1000 Figure 4 2 Partitioning a 9 GB disk using the stripe partition table U D a N a i DA 1 2 3 Size in MB x 1000 O Figure 4 3 Partitioning a 4 GB disk using the root partition table Disk Partitioning a 4 9 AUSPEX W Distribution of File Systems on Default Partitions All disks from the factory have default partitions when you receive the disks Default File Systems on the Root Drive The factory default partitioning scheme on the root drive places file systems in ad0a ad0g ad0d and adOf These file systems are mounted onto the directories usr usr openwin and var respectively may remain in var On smaller root drives var crash should be mounted from another file system G Note The file system var crash can be mounted from another file system or it Default File Systems on Default Drives All other drives are formatted at the factory and labeled with the default partition table File systems are initialized using newfs on partitions a g and h but these file systems are not mounted to any directories If the partitions defined on default drives at the factory do not meet your needs use ax_l
19. The drive logic is confused because the drive is not properly seated in the drive rack or the drive hardware is damaged Remove the tape drive and replace it in the drive slot to reset the tape drive logic After the drive executes the self test repeat the write operation If the error happens more than three times the drive hardware probably is damaged and you need to replace the unit 8 22 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Measuring the NetServer s Performance About This Chapter This chapter describes the Auspex Performance Monitor which gathers performance statistics about your server You can display the statistics in real time on the screen of any workstation connected to the server or you can save the statistics and view them later When playing back the statistics you can display them in histograms or graphs for easy interpretation The following commands collect display and convert server statistics ax_perfmon Collects server statistics displays them in real time or saves the data to a file ax_perfhist Displays statistics collected by ax_perfmon in histograms or graphs A variety of processor names appear on screens generated by the Auspex Performance Monitor You may not have all the processor boards mentioned in this chapter or on the screens The following list describes some of the ax_perfmon processor naming conventions 4 NPrefers to a dual CPU processor board
20. The following example illustrates how to display samples from a histogram 1 Load statistics to a histogram in the Data Display Window 2 Click the middle mouse button on a sample displayed on the left side of the histogram A marker which is a vertical dotted line appears This line marks the first sample to be displayed after you zoom in 3 To specify the last sample of the histogram after you zoom in click the middle mouse button at the sample to the right of the first selected sample Another marker appears as shown in Figure 9 20 To change or remove a marker follow these steps a Point anywhere in the histogram and hold down the right mouse button to display a menu which is shown in Figure 9 20 b Select Reset Markers from the menu Both markers in the histogram disappear To redraw a marker repeat step 2 The menu contains several other options which are explained in Menu for Zooming In and Zooming Out on page 9 34 c If two markers exist and you want to redraw the first one simply click the middle mouse button at the new location For example if the first marker is at 11 32 00 and the second at 11 33 00 you can click to move the first marker to 11 34 00 If you click at another location again the marker that is initially at 11 33 00 is moved 9 32 a Version 1 10 System Manager s Guide W AUSPEX v Auspex Systems Data Display Tool File v Properties v Print 7 Qty Hists amp 1 Clear
21. a option is in either command ax_config displays information from var adm config report in addition to the probe data when you choose the d command in the ax_config main menu Non interactive mode The following commands invoke ax_config non interactively ax_config o filename dva ax_config i filename wv The first command form is described in detail in Non interactive Mode on page 10 5 Use the command to copy the contents of var adm config report to another file The second command form copies the configuration information from a file that you saved or edited into config report it is used only by Auspex authorized personnel Interactive Mode To run ax_config interactively enter ax_config or ax_config v The v option means verbose mode which displays status messages while ax_config probes the software and hardware The following are sample messages that appear if you execute ax_config v Loading Hardware Data Performing Software Probe Performing Hardware Probe Main Menu ax_config Compare Hardware Data to Hardware Probe Display Configuration Data Edit Hardware Data Initialize Hardware Data Load Hardware Data from EEPROM Write Hardware Data to file Quit a S H He Da enter Choice gt The following section describes how to display configuration data with the d command in the main menu Other commands in the menu are for modifying the configuration report only Auspex
22. because the value 262 144 exceeds the maximum block size supported by the server The actual size of data blocks written to tape is equal to the maximum block size supported by the tape driver which depends on the HP model In this example if you use restore with the same blocking factor 512 to restore data that has been backed up on an HP IV based server the data cannot be read because the block size restore expects to see 65 532 bytes is different from the actual block size on the tape 64 512 bytes Table 8 3 Suggested dump and restore arguments Drive Type Tape capacity Option Argument 4 mm 60 meter 2 6 GB size 6300 density 110400 blocking factor 100 1 3 GB size 4100 density 55000 blocking factor 100 4 mm 90 meter 3 9 GB size 9600 density 110400 blocking factor 100 1 95 GB size 6100 density 55000 blocking factor 100 8 mm 112 meter Exabyte 2 3 GB size 6700 8200 density 65000 blocking factor 126 8 mm 112 meter Exabyte 4 7 GB size 11500 8500 density 141000 blocking factor 126 2 3 GB size 6700 density 65000 blocking factor 126 Tape Drive Capacity a 8 7 Table 8 3 Suggested dump and restore arguments Continued Drive Type Tape capacity Option Argument 8 mm 112 meter Exabyte 9 8 GB size 11500 8505 density 141000 blocking factor 126 4 9 GB size 11500 density 141000 blocking factor 126 2 3 GB size 6700 density 65000 blocking factor 126 8 mm 160 meter Exabyte 14 0 GB size 16282 8505XL density 141
23. different from SunOS 1 15 overview 8 2 syntax 8 2 with the T and m options 8 15 Dump level 8 2 dumpfs command 4 2 4 22 Dynamic Table Format 4 22 E ECC Error Checking and Correcting error 8 20 Echo mode 10 13 eject command 1 13 4 2 4 6 Index 6 a Version 1 10 System Manager s Guide AUSPEX WD W AUSPEX Email to Auspex 2 7 2 16 3 10 Enabling the Write Accelerator 7 8 End of tape error 8 20 Enlarging a striped or concatenated virtual partition 5 10 5 15 5 16 Error Auspex processor E 8 detected by LFS 6 10 file system problems 6 9 logging 10 11 media 6 2 6 23 message from the FP 6 9 6 25 message from the NP 10 11 message from the SP 6 3 6 8 6 23 6 24 message from vmunix 6 9 notification E 8 recovered E 8 recovery on mirrored partitions Error conditions for file system isolation 6 10 Errors detected by UFS 6 10 Ethernet 100Base T 1 8 2 14 2 32 address 2 30 analyzing data 9 47 interfaces 2 14 ports 1 4 Examples ax_diskconf display with p option 5 25 disk partitioning among multiple disks 4 12 Executables architecture dependent 2 21 2 28 4 11 Expanding a striped or concatenated virtual partition 5 10 5 15 5 16 exportfs command 4 30 exports file 1 14 Extended diagnostics 3 3 5 13 F Fast File System FFS file system format 4 22 fastboot command 3 2 3 11 fasthalt command 3 2 3 11 Fat Fast File System FFFS file system format 4 22 FDDI interfaces 2 14 FFFS Fat
24. enter the following command to allow the remote host to display histograms on your display xhost hostname where hostname is the name of the NetServer on which you start ax_perfhist To display server statistics in histograms simply enter ax_perfhist without any arguments An empty graph is displayed The window shown in Figure 9 9 is the main window for viewing statistics For a description of the full syntax of ax_perfhist refer to Starting ax_perfhist with Options and Arguments on page 9 46 When using ax_perfhist the mouse button functions are consistent with those in your windows environment For example to activate a command on the screen press the left mouse button to see a pull down menu hold down the right mouse button while pointing to a command This guide assumes that you know how to close open move and resize windows Displaying Performance Data in Histograms a 9 19 Auspex Systems Data Display Tool Qty mists 7 Figure 9 9 Data Display Window Figure 9 10 shows the top portion of the Data Display Window and describes the general functions of the buttons used in the window Detailed information about the windows opened by these buttons is provided in the sections following the figure Select a data file or Quit the ax_perfhist program configuration file to load Select the number of Print current hist gram histograms in the window or set up print properties
25. for example vp3 consisting of the group of physical partitions to which you want to move the contents of vp1 2 Use ax_mattach f to attach vp3 to the mirrored partition 3 Use ax_mdetach to detach the original member vp2 from the mirrored partition The contents of vp1 are now moved to vp3 Moving a Two Membered Mirrored Partition With No Unmount This section provides a procedure for moving a two membered mirrored partition In the following procedure the mirrored partition is vp1 and its members are vp2 and vp3 1 Create two new virtual partitions for example vp4 and vp5 each of which map to the group of physical partitions to which you want to move the contents of vp1 Use ax_mdetach to detach vp2 from vp1 Use ax_mattach f to attach vp4 to vp1 Use ax_mdetach to detach vp3 from vp1 OW pee 202 Use ax_mattach f to attach vp5 to vp1 Note After each ax_mdetach command in the previous procedure the mirrored partition has only one member As a result no mirroring function can take place before the subsequent ax_mattach command is complete 5 22 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Displaying Disk Configuration Information You can display information about the current configuration of all the disks in the NetServer using the ax_diskconf command This command lets you see at a glance how your file systems are arranged on the available partitions and which drives have unused part
26. it refers to the CPU that performs a specific processing function whether or not the CPU shares a board with another CPU The following lists the major hardware subsystems in the NetServer a FMP processors Host Network and Storage Processors a High performance disk subsystem a CD ROM and tape drives Note NetServer Version 1 10 software supports only servers with HP VII and i HP VIII NP III and NP IV and SP IV and SP V boards As more users join the network processor and storage capacity can be added to accommodate the increased work load The following sections discuss the FMP processors the shared cache memory and the Enhanced VME Bus over which the processors communicate with each other Host Processor HP The Host Processor performs these functions a Booting the system 4 Downloading software to the other processors a Running UNIX SunOS and applications a Exporting the standard Sun ONC services a Running the NetServer system utilities and diagnostics The HP manages NFS Version 2 services for requests to file systems mounted on the HP Most NFS requests for Auspex Local File Systems LFS are processed by the NP The NetServer s SPARC based HP is binary compatible with SPARC application environments and provides a compute capability independent of NFS loading Caution The architecture of the HP VIII is aushp Loading kernel modules A compiled for an architecture different from that of the HP might
27. oc 0 0 0 SP3 OXI oc 0 0 0 SP4 OZI oc 0 0 0 0 Total DISK OPs s ill Figure 9 2 Sample System Summary screen for an NS 7000 600 Series NetServer The System Summary screen contains the following information a The first column describes CPU utilization for NPs and SPs CPU utilization is expressed both in percentage and in a bar graph made up of equal signs If a dash is displayed instead of a percentage the processor is not installed Note The NP s CPU statistics pertain to the network processor only even i though each NP board has two CPUs Utilization statistics for the file processor CPU also located on the NP are listed by FPs Capturing and Displaying Real Time Performance Data A 9 7 a The upper center statistics show HP utilization The utilization statistics for the HP which show how busy the HP CPU is are divided into three types User utilization represented as usr on the screen shows the percentage of CPU time in user mode System utilization represented as sys on the screen shows the percentage of CPU time used in system mode Idle represents the percentage of time when the CPU is neither in user nor system mode User utilization system utilization and idle time add up to 100 percent a The rest of the screen displays statistics on operations processed by network interfaces FPs SPs and statistics on disk activity For the network interfaces the num
28. of tasks such as the system installation procedure and some system management procedures require manual booting using the PROM monitor boot facility 3 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX exportable file systems If a file system mount fails because of a problem witha virtual partition or a bad disk the system boots to single user mode with error messages fy Note A system on its way to booting in multiuser mode attempts to mount all The procedure for booting in single user mode can be considered a modification of the usual multiuser boot procedure that is appending the s flag to the boot command line causes the system to boot in single user mode The procedures for booting in each mode are described in the following sections For more detailed information about system booting and shutdown procedures see Chapter 5 in Sun s System and Network Administration Accessing the PROM Monitor The PROM monitor program resides in PROM on the HP board This program provides a set of functions that control the system before the operating system is booted These control functions include 4 Bootstrap operations a Low level system configuration a Power on diagnostics a Extended diagnostics a Resetting the subsystem controllers Use the PROM monitor program whenever you want to use one of the functions listed above This section discusses only how to access the PROM monitor so that you can boot the
29. on page 4 20 Adding or Replacing a Drive a 4 29 AUSPEX D Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 6 Mount the file system For example to mount the file system created in step 4 enter the following command mount home src Note If the new file system is an LFS file system you must enter the mount i command with only one argument This format causes mount to read etc fstab If you enter two arguments that is if you specify both device and mount point in the command the file system is mounted as file system type 4 2 and all operations on this file system are controlled by the HP not the FP 7 Enter the mount command again without arguments This form of the command generates a list of all mounted file systems Check the output to verify the new file system was mounted properly 8 Add an entry to etc exports to export the newly created file system For example home sre 9 Export the file system by entering the exportfs command as follows exportfs a 10 Enter the exportfs command again without arguments This form of the command generates a l
30. on the new raw virtual partition on which a file system will be mounted that is do not run newfs on virtual partitions that are members of a mirrored partition For example if vp1 and vp3 are members of a mirrored partition named vp2 enter the following command to create a file system on vp2 usr etc newfs dev rvp2 5 14 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Caution You must run newfs on any raw defined virtual partition before A attempting to mount the partition You do not need to run newfs on the underlying physical partitions 7 Use the mkdir command to create a directory for the file system 8 Edit etc fstab to define a mount point for the file system for individual partitions that are members of a mirrored striped or concatenated virtual partition Only the mirrored striped or concatenated virtual partition itself should have an entry in etc fstab fy Note When you modify etc fstab to define mount points do not add entries For example if vp1 and vp3 are members of a mirrored virtual partition named vp2 only vp2 has an entry in etc fstab vp1 and vp3 do not Similarly if ad5h and ad6 h are members of a concatenated virtual partition named vp4 only vp4 has an entry in etc fstab ad5h and ad6h do not 9 Use the mount command to mount the file system on the partition 10 Restore the file system that you backed up to tape in step 2 11 Add an entry to etc exports to make t
31. 0 3c 0 39 1d hostl el 144 48 252 4 deleted hostl e2 144 48 252 6 at 8 0 3c 0 30 10 hostl e2 144 48 252 6 deleted Flush the ARP cache if the IP to physical address binding for a host on the network changed and you want to update the cache quickly In the previous example suppose the physical address of host1 e1 changed and you flush the cache When host0 e1 tries to reach 10 14 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX host1 e1 because the entry for host1 e1 is missing host0 e1 sends an ARP request for the new address binding information without waiting for the ARP entry to time out When host0 ael receives the new information it updates the cache immediately If you do not flush the cache after the address change host0 e1 uses the out of date information for host1 e1 until the entry times out before it sends an ARP request In this case updating the cache might take up to several minutes Auspex has not modified arp 8C which is part of SunOS and only pertains to the ARP table maintained by the Host Processor vmunix Because ax_arp allows you to specify an interface number use ax_arp not arp to query the ARP table used by a selected interface Displaying Network Statistics for a Network Processor To display the contents of various network related data structures on an Auspex network processor use the ax_netstat command The ax_netstat command differs from ax_perfmon in that statistics display
32. 1 023 MB 720 MB 500 MB 420 MB 322 MB 238 MB Table 4 7 Root drive partition table in MBs Microp Partition 4 GB 3 GB HP 2 1 GB 2 GB HP2 GB 1 76 GB 1 35 GB 1 GB a 100 100 50 50 50 50 50 50 b 600 650 650 650 650 500 450 400 c 4 094 2 882 2031 2 000 1909 1 682 1 291 954 d 200 150 120 120 120 120 120 120 e 1000 300 90 80 40 70 30 10 f 1000 1000 800 800 800 650 450 200 g 250 250 220 220 220 220 160 160 h 944 432 98 80 29 70 31 14 Based on the partition sizes defined in the selected partition table default stripe or root you can partition the disk in different ways For example you can partition a 9 GB stripe drive into a b g and h partitions or you can partition it into e and f partitions See Figure 4 1 through Figure 4 3 for the different ways to partition a 9 GB disk Partitions for other size drives have identical proportions In the figures each bar represents a partition scheme in which the size of each partition is defined by a partition table The numbers shown along the bottom of the diagram indicate the partition sizes 4 8 a Version 1 10 System Manager s Guide W AUSPEX To display the partition table for a disk use the dkinfo or ax_lslabel command The following are examples of these commands dkinfo adn ax_lslabel adn where n is the disk drive number a number in the range of available drive slots Each partition table is defined as an entry in etc format dat the data file used by the format command
33. 13 4 26 4 31 5 10 5 23 6 1 8 13 ax_expand 1 14 5 10 5 15 5 18 ax_fsutil 1 14 6 14 ax_hot_plug 1 13 4 27 4 32 4 33 6 4 6 5 6 6 ax_isolated 6 12 ax_kill 1 14 6 15 6 16 ax_label 1 13 4 7 4 11 4 36 4 41 ax_load_flash 1 15 10 7 ax_loadvpar 1 13 5 10 5 20 ax_Islabel 1 13 4 42 ax_mattach 1 13 5 10 ax_mconvert 1 13 5 11 5 20 5 21 ax_mdetach 1 13 5 11 ax_mrestore 4 27 4 32 4 33 5 11 5 13 ax_netstat 1 12 10 14 10 15 ax_perfhist 1 12 9 1 9 18 to 9 46 ax_perfmon 1 12 9 1 9 2 to 9 17 ax_remove_device 1 13 4 33 5 11 ax_sputil 6 16 6 27 ax_tapestats 4 27 4 32 ax_vpstat 1 13 5 11 ax_write_cache 1 15 7 6 7 8 b boot 1 13 3 2 c continue 10 12 crash 10 10 cron 8 10 dd 4 38 6 26 df 4 22 5 3 dkinfo 1 13 4 2 4 9 4 28 5 24 6 27 dump 1 15 8 2 dumpfs 4 2 4 22 Index 4 a Version 1 10 System Manager s Guide eject 1 13 4 2 4 6 exportfs 4 30 fastboot 3 2 3 11 fasthalt 3 2 3 11 fddistat 1 12 format 1 13 4 7 4 36 5 7 6 26 fsck 1 14 3 8 4 22 6 26 10 10 getcores sh 1 15 go goto 10 11 halt 1 15 3 2 3 10 10 12 init 1 12 2 19 installboot 6 26 6 28 MAKEDEV auspex 1 13 4 4 mkdir 4 29 mount 4 2 4 22 4 30 mt 8 7 8 22 ncheck 8 17 newfs 4 10 4 29 5 7 6 26 6 28 nfsstat 1 12 10 14 10 19 NSconfig 2 10 2 12 2 17 NSinstall 6 26 passwd 2 3 rc 1 15 reboot 1 15 3 2 3 8 3 10 reporter sh 2 7 restore 8 2 8 3 8 9 SetupClient 2 27 SetupExec 2 22 shu
34. 2 0 If Superblock Is Unreadable If a file system superblock cannot be read during system reboot fsck stops and the system cannot boot to multiuser mode If the superblock problem happens during normal server operation an error message reports that the superblock is unreadable In either case the SP generates an error message showing the number of the sector that caused the media error Follow this procedure to reassign the superblock 1 Use the ax_sputil command to reassign the superblock The following is an example Suppose the SP generates the following media error message Nov 17 17 47 36 host_name SP1 1196 Disk Media Error slot 22 drive 1 sense key 3 sense code 11 sector 16 Enter the ax_sputil as follows to reassign the superblock ax_sputil reassign 22 16 This command reassigns sector 16 of disk 22 to a new location However because the superblock is larger than one sector you may need to reassign more than one sector 2 Use the fsck command to restore information from the alternate superblock to the new superblock fsck b 32 rad22a To see a complete list of backup superblocks before running fsck use the newfs N command Block 32 is always the first alternate superblock For example to list the superblocks for ad0a enter the following command newfs N dev rad0a 3 If you are already in multiuser mode mount and export the file system If you are in single user mode use fsck p to check the fil
35. 6 15 L Label disk 4 7 4 28 4 40 5 7 5 8 6 23 6 24 option in format command 6 24 LED on the tape drive 8 6 8 22 Left arrow key 2 10 Level 0 backup 6 26 6 28 LFS changes to df mount and umount commands 4 22 default format 4 22 file system isolation 4 18 filesystem type 1 14 4 18 4 30 5 11 mapping file systems to FPs 4 20 Listing backup superblocks 6 24 Listing file systems on a drive 5 23 Load Hist button 9 24 Loading statistics to a histogram 9 21 Local drive location 2 25 Local terminal 2 19 LRU buffer statistics 9 13 Mail Recipients 2 7 mailmessages sh script 2 7 2 16 Maintaining tape drives 8 1 MAKEDEV auspex command 1 13 4 4 Making anew NIS map 4 14 Man pages OpenWindows 4 16 Managing virtual partitions 5 14 Mapping LFS file systems to FPs Mapping physical and virtual partitions 6 1 Marker in a histogram 9 32 Master server 2 15 Maximum number of virtual partitions on anSP 5 7 Maximum Transmit Unit MTU off the FDDI network 2 15 specifying in NSconfig form 2 15 Maximum virtual partition size 5 3 Maxusers kernel parameter 2 6 Media errors definition 6 2 messages for 6 23 Memory used by net interfaces 2 7 4 20 A Index 9 Menu format command 4 38 partition 4 39 Message Auspex processor errors E 1 E 8 error E 1 kernel panics 10 11 logging 10 11 panic E 8 recovered errors E 8 standard boot E 1 startup errors E 1 text E 8 Meta cache 9 12 Mirrored partition adding anew me
36. 9 32 for information on defining which samples to zoom Zoom All If multiple histograms are displayed zoom in on the samples in all the Histograms histograms ax_perfhist selects the samples based on the markers set in the Expand View by 10 Expand View to Full Reset Markers Match Y Scale on All Hists Reset Y Scale on Hist Continuous Plot Discrete Plot Remove Hist current histogram This option is useful if you want to compare statistics that were collected in the same period of time Display 10 more samples on either side of those samples that are currently zoomed For example if the histogram contains a total of 20 samples this option displays two more samples next to each marker Display all the samples in the current histogram Remove the markers currently in the histogram Standardize the Y axes in all the histograms by using the same scale minimum value and maximum value The scale and values in the current histogram are copied to other histograms in the Data Display Window For example if histogram 1 uses 0 0 and 28 1 as the minimum and maximum values respectively and if you click in histogram 1 to select the menu option the corresponding values in the other histograms change to match the ones in histogram 1 Restore the scale minimum value and maximum value on the Y axis to the original settings before you used Match Y Scale on All Hists Display the statistics in a continuous graph This o
37. AUSPEX Maintenance Commands MAKEDEV auspex 8 NSconfig 8 NSinstall 8 NSupdate 8 SetupClient 8 SetupExec 8 SetupTty 8 auspex 8 ax_add_device 8 ax_admmsg 8 ax_admmsg check 8 ax_arp 8 ax_checklicense 8 ax_chkdrive 8 ax_clonefs 8 ax_config 8 ax_crash 8 ax_decyrpt 8 ax_diag 8 ax_diskconf 8 ax_docs 8 ax_drives 8 ax_drstat 8 Make Auspex special files Configure the software on an Auspex network server Install the software on an Auspex network server Auspex utility to install new releases Add a diskless SunOS client to the Auspex network server or remove a client from the server Install architecture dependent executables on the server Configure a tty device for a modem or terminal Introduction to the Auspex network servers Utility to allow disk drives to be added or replaced while the NetServer is running UNIX level routine to extract Auspex error messages from the message files in usr adm UNIXx level routine that mails an error message report to a designated list of people Display and flush address resolution table on an Auspex network interface Checks for valid optional software product licensing Auspex drive initializer Auspex command to create and configure a virtual partition that duplicates data in another virtual partition Auspex utility to display and modify system configuration information stored in EEPROM System crash analysis Decrypt optional produ
38. Acceleration Virtual Partitions RAID 0 1 RAID Device RAID5 Parallel l SCSI Channels Internal SCSI Disks and Devices Figure 1 2 NetServer Functional Multi processing architecture 1 10 A Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Advantages of the Auspex Environment The Auspex environment is distinguished by its multiple processors drives and networks centralized in a single server This feature provides several system management advantages Consolidation The NetServer s high performance storage subsystems reduce the need for maintaining and backing up multiple copies of files and programs spread across multiple conventional servers Balanced client traffic You can distribute client traffic more evenly for these reasons a Client data can be distributed among disk drives to provide maximum I O efficiency a The virtual partition capability allows you to distribute logical disk partitions over multiple physical disk drives thus improving file access times to the file system mounted on the partition This feature provides a dramatic performance improvement on heavily used file systems a The client load can be balanced among multiple networks Distributing traffic on several networks reduces delays and increases efficiency Expandability If the number of clients outgrows the server s current capacity the server can be expanded to support more clients or networks
39. All Histl Fri Feb 3 15 25 57 1995 Servertacdelco File acdelco dat 25 0 24 34 23 54 22 84 22 14 21 34 20 64 19 94 19 14 18 44 17 64 16 94 16 24 15 44 14 74 14 04 13 24 12 54 11 84 Reset Markers 11 04 10 34 Match Y Scale on All Hists Reset Y Scale on Hist F eo Continuous Plot 8 Discrete Plot tir i 6 Remove Hist i 5 a a 5 4 3 2 2 L 0 0 Fri 15 6326 Fri 15326349 Fri 15327310 Fri 15327333 Fri 15327357 NPO cpu busy Ave 9 9 I A Zoom All Histograms Expand View by 10 Expand View to Full SCnNoOnonfrrPruonrrrRo nh Figure 9 20 Pull down menu for zooming samples between the markers 4 To zoom in click the right mouse button anywhere in the histogram When the pull down menu appears click on the Zoom option Figure 9 21 on page 9 34 illustrates the histogram after the zooming After you zoom in it is easier to examine the statistics The bars are now wider and the X axis has more space to show the collection times for the samples 5 If multiple histograms are displayed and you want to zoom in all the samples collected in the time period defined by the markers in one of the histograms follow these steps a Point anywhere in the histogram that contains the markers b Press the right mouse button to display a pull down menu Click on the Zoom All Histograms option Displaying Performance Da
40. Arguments The full command syntax for ax_perfhist is as follows Refer to Table 9 8 for explanations for the options and arguments used in the syntax ax_perfhist cf configfilename df datafilename pf print_to_filename l1p qt Table 9 8 Options and arguments used in the ax_perfhist command Option and argument Function cf configfilename Specifies the configuration file that determines the type of statistics in each histogram the appearance of each histogram the window layout and so on For more information on saving a configuration file refer to Configuring the layout of the Data Display Window on page 9 28 Always define a configuration file before using the ax_perfhist command to print the histograms to ensure that the output contains the types of statistics you want and prints in the proper format df datafilename Specifies the name of the file containing the statistics recorded by ax_perfmon The statistics in this file are loaded to the histograms pf print_to_filename Specifies the name of the file to which the histograms are saved You can save the output of ax_perfhist and print them out later Ip Prints the histograms once ax_perfhist loads the configuration and data files qt Quits ax_perfhist after loading the configuration and data files and printing The following are examples of the ax_perfhist command ax_perfhist cf 6histogram cfg df host1 dat This command displays the Data Di
41. CSEXPORT exec kvm ARCH sunos VERS____J Path to usr share files CtEXPORT share sunos VERS Distribution medium CDROM Tape gt gt CDROM Drive location Local Remote gt gt Local CD ROM device name acd1__ e g acdl acd2 acd209 Mount point of CD ROM cdrom Ike TAB CR or arrow keys to move among fields when done hit CTRL F to execute HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VAL Figure 2 7 Sample SetupExec form for a local CD ROM device NetServer Configuration Commands a 2 23 AUSPEX W Auspex NetServer SetupExec Form Fri Feb 2 15303 Architecture type SARCH SunOS version ERS FileServer partition EXPORT export1 Path to executables EXPORT exec ARCH sunos VERS Path to kernel executables EXPORT exec kvm ARCH sunos VERS____ Path to usr share files EXPORT share sunos VERS Distribution medium CDROM Tape gt gt Tape Drive location Local Remote Neanta Tape device name rmtO__ e 9 rstO rst8 rmt0 rxtd Select the location of the tape drive HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VAL Figure 2 8 Sample SetupExec form for a remote tape device Auspex NetServer SetupExec Form Fri Feb 2 15 00 Architecture type ARCH SunOS version VERS FileServer partition EXPORT export1 Path to executables C EXPORT exec ARCH sunos ERS Path to kernel executables C EXPORT
42. E athe aN a iA EE Ghatak GE EE 8 20 End f tape Errors Sioi eso chk Gu ounce a E E a Gomes 8 20 Tape M dia Errors e E i E EE E A eee E AA E E ARN 8 20 Write BrtOris ick oaea a a a sy ea dara Rial Nae E AE A A de 8 21 Read Errom sageeo ne a A E AE a S I gaia E EEE Se he 8 21 Write Faure Error c Siso ahai a o a E Gas ES d a aAa 8 21 Tape Jam EfTOr S ida e isk tis ace Sas slats dia anh eed hin aon alae 8 21 Tape Drive Errors orenera ekta a e e vad eis dose ite ane Geek Wane a idl g 8 22 Write Errotia sic sha E seers ede aA oe td ake eas 8 22 Online ERO hg weep reg a e sy Baas elds nokta tuasy E a Sey g 8 22 Chapter 9 Measuring the NetServer s Performance About This Chapter rers Menaes corks ds E a hed eda a RE heed 9 1 Capturing and Displaying Real Time Performance Data 9 2 How ax perfmon Works oseca eerie Ga ict Seay mre er eae eS 9 2 Starting the Performance Monitor sn fie ytdwe wap porta 9 3 Interactive Performance Monitor Commands 00 000 eee 9 5 Interpreting Statistics Screens tu ovat eeu hanie yes AUR tate Cacennes 9 6 Screen LIGA Ch cinta estuen Rieck a o a aes teen EEE 9 6 System Summary Screen axa pig Oh ache Rend tute asa Pade Chee octane 9 6 INP Statistics Screen ai roeien eg ben PEE E Rican hd ade eases 9 8 NP Protocol OChOCN yeu renee Rid Ruch ora cate GTER uaa Para 9 10 Contents A ix BP Statistics Sereen cig oa rai eaa haar ae aana A Ratan VE Ea 9 12 SP Statistics Sereer sos
43. ELVIS Wisse eesti ts A E tes ed MIE RATING el tata asin ed REE aa A ng i B 5 Printing a Documentaire a e eee eee B 7 Appendix C_ Time Zones World Tim Zones A ceed nok eeanclcecetitne id a ea ody ated an eae leat ieee oN C 1 Appendix D Configuration Worksheets About This Appendix si sauce meamea her pene ee ey ae Oa ee pa oa ea a D 1 Host Information Worksheet 0 0 0 0 cece ccc cece eee e eee eeee D 2 Client Information Worksheet 0 0 0 0 cece eee eee eee eee e eens D 4 Disk Drive Information Worksheet 0 0 0 0 ccc ec e eee e eee e eee D 6 Virtual Partition Information Worksheet 0 00 0 cece eee ee ees D 7 Appendix E Messages Abo t This Appendix erritar enoni ien hr AN teen eth Mant EA Canasta E 1 Standard Boot Messages oc iat tiete danas uted ba teal oie d heals Caren eae E 2 System Startup Error Messages inca 6 Pinte acte tastes Pe Seb errre rrene E 7 Auspex Processor Error Messages 0 6 6 0 c cece cece eee ee E 8 Index Contents a xi W AUSPEX Figures Chapter 1 Figure 1 1 Figure 1 2 Chapter 2 Figure 2 1 Figure 2 2 Figure 2 3 Figure 2 4 Figure 2 5 Figure 2 6 Figure 2 7 Figure 2 8 Figure 2 9 Figure 2 10 Figure 2 11 Chapier 3 Chapter 4 Figure 4 1 Figure 4 2 Figure 4 3 Figure 4 4 Figure 4 5 Figure 4 6 Figure 4 7 Figure 4 8 Figure 4 9 Figure 4 10 Chapter 5 Figure 5 1 Figure 5 2 Figure 5 3 Figure 5 4 Figure 5 5 Figure 5 6 Figur
44. Install Software From CD ROM on page 6 29 6 28 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX Using NSinstall to Install Software From CD ROM Normally installing the operating system is not required because each server is shipped with the NetServer software installed However there are situations where you want to copy the operating system from the distribution CD ROM to a disk a When the root disk or a file system on the root disk is damaged and must be re created a When you want to create a spare root drive a When you want to update to a new operating system release See the Software Release Note included with a your new release package for the use of NSupdate 8 NSinstall installs file systems from the distribution CD ROM to a specified disk Its syntax is as follows NSinstall cdrom slot The slot is the slot where the CD is inserted Slot numbering starts from 0 Without an argument NSinstall uses the CD in slot 1 NSinstall can run in either single user mode or multiuser mode Note NSinstall assumes that the drive to which the file systems are installed li is formatted For information on how to format a drive refer to Using the Format Command on page 4 36 Formatting a disk erases all data that existed on the disk before the format operation The following subsections describe the steps for installing file systems from the CD ROM Starting NSinstall If you are installin
45. One CPU is for network processing and one CPU is for file processing With SBus adapters each NP board can have three FDDI interfaces three 100Base T Ethernet interfaces high speed interfaces or three ATM interfaces Note ATM LANE is available as an optional product Contact your i authorized Auspex representative for more information a FP refers to the File Processor on an NP Measuring the NetServer s Performance A 9 1 Capturing and Displaying Real Time Performance Data This section describes ax_perfmon an Auspex Performance Monitor command for collecting and displaying NetServer performance data From a single workstation you can use ax_perfmon to display performance data from several Auspex servers The concise display of NetServer performance data is a valuable tool for system management decisions enabling detailed investigation of site and application specific NetServer performance Note If you upgraded your NetServer software from any previous version D notice that ax_perfmon rearranged data in some of its screen displays to accommodate hardware features introduced with this software release The information you are looking for may be in a different location on the screen How ax_perfmon Works ax_perfmon gathers data from each processor in the system and provides both real time display and fast motion replays of significant NetServer performance events The statistics gathered include 4 CPU utilization for
46. Options in pull down menu in Data Display Window 9 35 Table 9 6 Pull down menu options displayed in the input file list 9 37 Table 9 7 Differences between xwd and screendump 00000 cee eeeee 9 40 Table 9 8 Options and arguments used in the ax_perfhist command 9 46 Chapter 10 Special Maintenance Appendix A UNIX Manual Pages Appendix B Online Documentation Appendix C Time Zones Table C 1 Time zone listing s4 0s borg tata t harsher gewtadis alan eres t C 1 Table C 2 Additional time zone listings 2 e260 cece ke or ee ee C 4 Appendix D Configuration Worksheets Appendix E Messages Table E 1 NetServer boot event explanation 0 cece cece eee eee E 6 Table E 2 Auspex Processor Error Messages st ct is faded ta Moe ne dae ae 8 E 9 xvi A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Preface About This Guide This guide gives you detailed information about how to operate and manage an Auspex NetServer It offers procedures and recommendations on how to run the NetServer software efficiently to ensure reliable and fast file service at your site Applicable Documentation Because Auspex NetServers run the SunOS 4 1 4 operating system and are fully compatible with ONC NFS many Sun Microsystems documents are relevant to NetServer operations This guide documents differences from SunOS and does not attempt to document SunOS or system administ
47. Removing aCD ROM 4 6 adrive 4 31 7 4 a failing drive 6 7 ahistogram 9 24 a member from a mirrored partition 5 11 a Statistics file 9 37 Repairing the root drive using fsck 6 27 Replacing a drive containing multiple partitions in a one member mirrored partition 6 22 in multi user mode 6 6 in single user mode 6 5 procedure for 4 25 with mirrored partitions 6 20 with unmirrored partitions 6 19 Replacing an inaccessible drive at 6 24 7 10 6 22 Index 12 A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX reboot 6 4 Replacing an invisible disk on a running system 6 19 reporter sh command 2 7 reporter sh script 2 7 restore command overview 8 2 syntax 8 3 Restoring a file from 1 5 1 produced dump tapes 8 18 with damaged inode 8 17 Restoring status virtual partition 5 12 Restrictions on concatenated and striped partitions 5 7 on mirrored partitions 5 8 on virtual partitions 5 7 Return key 2 10 Rewind option for tape drives 4 2 Right arrow key 2 10 Root damaged partition 6 26 directory 2 28 4 7 4 11 filesystem 4 13 installing using NSinstall 6 32 Root disk backup 6 26 8 11 8 14 booting from backup 6 26 copying 6 27 damaged 6 26 drive label 6 27 recovering from failures 6 26 replacing 4 26 Route daemon 2 11 2 16 Runt partitions combining 5 2 S savecore command 10 11 Saving ax_config information to a file 10 6 Saving performance statistics to a file 9 19 Screen captures in ax_perf
48. Root Disk 9444 lt i cciaiias oxs aya w oe Meares 6 26 Booting From the CD ROM in Single User Mode 5555 6 27 Using NSinstall to Install Software From CD ROM 00004 6 29 Starting NSinstall dy hess tinea dag heat a beatae a bean ga wea aad 6 29 NSinstall FOrmi sya oa bas heme yar ay a ee mee eye aes 6 30 Executing the NSinstall Ports iis a4 s eee iteie ahaa Ne Ea 6 33 Chapter 7 Write Acceleration About This Chapter 2 wcihs atte ectdies GG Astin hee aod E aE A 7 1 Write Acceleration Overview 000 c cect e e aie E E 7 2 How the Write Accelerator Improves Throughput 00 7 3 viii a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX How the SP Maintains Drive Information 00 0c eee eee 7 4 Write Accelerator Restrictions 0 000 cece teen tne eens 7 5 Write Cache States and Commands 00 cc eee eee eee 7 6 Enabling the Write Accelerator uu susunu nrnna rreren 7 8 Disabling the Write Accelerator i413 sta asin na nurun r runn r rnnr 7 9 Managing the Write Accelerator 0 0 0 cece cee eee ees 7 9 Recovering From Write Cache BADCHECKSUM or DIRTY Errors 7 9 Recovering From a Disk Media Error DIRTY State 7 10 Recovering From a Disk Hardware Error 00 7 10 Recovering From a Drive Configuration Change 5 7 11 Chapter 8 Backing Up the NetServer About This Chap
49. The NetServer s performance increases proportionately with the increased number of drives and functional processors supported by the server Availability The NetServer offers uninterrupted service to network users when performing system administration procedures such as backing up file systems adding removing or replacing drives and reconfiguring or enlarging a virtual partition The file system isolation capability ensures that a corrupted file system does not affect the operation of the entire server the File Processor takes the file system offline so that you can repair it and make it available to users again Backup speed Backup operations can take place entirely between SP boards and the backup devices giving increased speed when backing up large amounts of data Backup devices can be tape drives tape jukeboxes or tape stackers In administering the NetServer be aware that some standard administrative practices are modified to take into account the number of devices and networks that exist in the centralized NetServer environment Some procedures are altered to reflect the unique features of the NetServer environment However system administration procedures in many respects remain identical to those used for the SunOS operating system by Sun Microsystems on which the Auspex operating system is based Refer to the following section SunOS Enhancements for information about Auspex changes to SunOS Advantages of the Auspe
50. Version 1 9 Sy D Contents D Figures D Tables D Preface DO Overview of N gt D Software Conti gt D Booting Up an gt D NetServer Stor gt D Virtual Partitio DD Recovering Fr DO Write Accelera gt D Backing Up the gt D Measuring the gt D Special Mainte DD Time Zones bD Configuration DDO Messages bD Index Printing a Document Version 1 10 System Manager s Guide Figure B 2 Bookmarks For your convenience you can print individual pages or the entire contents of the online manual Bring down the File menu from the top of any of the windows and choose the print option The Premier Software Series CD also has a directory named psdocs that holds PostScript files for each document You can print the documents directly from psdocs To print documents in A4 format use the files with the psA4 suffix Online Documentation a B 7 W AUSPEX Time Zones World Time Zones This appendix contains a listing of world time zones A time zone name from this table is required during the NetServer configuration process Table C 1 Time zone listing Time zone name Time zone area North America U S Eastern U S Central U S Mountain U S Pacific U S Pacific New U S Yukon U S East Indiana U S Arizona U S Hawaii Navajo Canada Newfoundland Canada Atlantic Canada Eastern Canada Central Canada East Saskatchewan Canada Mountain Canada Pacific Canada Yukon Mexico BajaNorte Mexic
51. Yes to restrict rsh access to the etc hosts equiv _ etc hosts equiv server and thereby make the server more secure Partition The physical partition that you want to Default is e reserve for export Set up home Whether to set up the home file system Yes when you install software for the first time NO when you reinstall software This option is automatically disabled if the target disk doesn t have the h partition partition The physical partition that you want to Default is h reserve for home Executing the NSinstall Form After filling out the NSinstall form execute it by following these steps 1 Execute the form by typing Ctrl F NSinstall displays a number of messages describing the operations performed for the installation When the installation is finished the following message is displayed Scratch install DONE If this is a partial install please check the etc fstab file on the target disk to make sure that is has the desired information Partial install is used for partition repair only so the fstab file might not have the desired information You might want to check the log file tmp NSinstall log on current root disk or NSinstall log on the target disk NSinstall session ended on date Partial installation means that some partitions were not selected The messages are also saved to NSinstall log on the target drive and tmp NSinstall log on the current root drive If you use
52. a yes to the query However NFS Version 3 on an Auspex NetServer does not work with UFS file systems Any UFS file system exported to and mounted by a client must now be explicitly mounted on the client with NFS Version 2 Some client operating systems such as a PC with NT do not have the option to explicitly mount NFS versions Such systems query for NFS Version 3 then appear to mount UFS file systems from the NetServer but access to the file systems fails See your client machine s documentation for mount options that specify NFS protocols NFS Version 3 and UFS Limitations a 4 23 AUSPEX W Specifying a Host Name in etc fstab For a NetServer to access an exported file system with root privilege the fstab entry for the file system must use the name of the appropriate network interface of the NetServer that exports the file system The entry must not use the name for the entire NetServer For example you want to export home engineering from netserver1 to netserver2 with root permission If netserverl e5 is the Ethernet interface attached to the network that is also attached to netserver2 the fstab entry for home engineering on netserver2 must include the following text netserverl e5 home engineering home engineering If you specify netserver instead of netserver1 e5 in the entry netserver2 cannot access the file system The host name used in the entry is the name that maps to the IP address of the network interface on the
53. an unreachable machine halt 8 reboot 8 and fastboot 8 now include the c option which instructs the system to perform a cold boot in which all diagnostics are performed shutdown 8 is modified with the I option to send warning messages only to logged in users The option speeds up the shutdown process if the list of machines mounting files is not up to date or if machines mounting files do not have rwalld 8C support SunOS Enhancements Man pages for rc shutdown rwalld Chapter 7 Write Acceleration Dumping a Clone Partition to Tape on page 8 15 halt reboot and fasthalt and fastboot on page 3 10 A 1 15 Table 1 2 Changes to SunOS Continued Types of changes Description For further information see NeTservices commands The asu 8 command allows an administrator to manually stop NeTservices start NeTservices or reset the daemons with a restart stop shuts down NeTservices on all interfaces start enables and starts NeTservices on all interfaces restart stops NeTservices on all interfaces then immediately starts NeTservices again This option is useful as a troubleshooting tool If NeTservices is not functioning correctly a restart may clear up problems help gives information about using the asu command NeTservices Product Guide NeTservices Administration Guide and NeTservices Concept and Planning Guide For a complete list of fil
54. and NetServer throughput by reducing the response time involved in writing data to disk The write cache on an SP V with the optional Write Accelerator III and current PROM Flash code contains 8 MB of NVRAM with battery backup to protect data when a power failure occurs I O Cache Memory The shared I O cache memory which resides on all NP boards is used as a large disk cache No processor instructions are stored in or fetched from this memory Memory and backplane bandwidth are devoted exclusively to I O Enhanced VME Bus The Enhanced VME Bus implementation provides 100 MB per second block transfers between the NP and SP boards 1 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Optional Products Auspex provides a system for installing optional software products on the NetServer Software packages are immediately accessible through the acquisition of a decrypting key and license from Auspex Optional product packages are located on the Auspex Premier Software Series CD ROM The packages must be installed with pkgadd 1M onto the NetServer system software to run Installation instructions are in the CD ROM booklet and in the Version 1 10 Software Release Note Documentation for all the software packages is provided on the Auspex Premier Software Series CD ROM Installation and printing instructions for documentation are in the CD ROM booklet and in Appendix B of this manual The following optional software produ
55. as ae0 and afddi1 Please refer to the nfsstat man page for complete nfsstat syntax The reporter sh 8 script when enabled automatically mails site reports to Auspex Customer Service The script calls nfsstat 8C with the z option to reinitialize nfsstat statistics If you do not want the statistics fields to automatically reinitialize edit the script usr auspex reporter sh Comment out the nfsstat command that uses the z option and uncomment the nfsstat command that does not use the z option Network Changes Whenever you add a new network interface or change an IP address the router daemon does not automatically pick up the change The daemon complains with an error message similar to packet from unknown router After additions or changes always restart the daemon in routed See the in routed 8C man page for further information Managing Network Interfaces a 10 19 W AUSPEX UNIX Manual Pages About This Appendix This appendix lists the online manual pages for the commands and utilities added or modified in SunOS by Auspex Some man pages are modified because changes have been made to the software and some are modified to follow Auspex conventions UNIX Manual Pages a A 1 User Commands a2p 1 arch 1 ax_dbx 1 ax_sparc_dbx 1 ax_tapestats 1 bar 1 cpio 1 dd 1 eject 1 installf 1 mach 1 md 5 mt 1 nohup 1V pax 1 perl 1 pkgadd 1 pkgask 1 pkgchk 1 pkginfo 1 pkgmk 1 pkg
56. as optional products Refer to the appropriate documentation on the Auspex Premier Software Series CD ROM for more information An IP address in decimal separated octets format for example 192 9 200 100 Yes or No AUSPEX W W AUSPEX Table 2 3 The NSconfig form fields Continued Field Description Possible values Subnet Mask Broadcast Type FDDI MTU Subnet mask for each network interface Select 0 if the host part of the broadcast address is all Os select 1 if itis all 1s The broadcast address is generated based on the broadcast value the subnet mask and the Internet address Size of the maximum transmission unit in traffic sent over the FDDI network directly attached to the server The MTU limits the size of a frame that can be sent by an FDDI interface This value is tunable for each FDDI interface on the server Null the decimal separated octets of the subnet mask such as 255 255 255 0 or an 8 character hexadecimal string such as ffffffO0 If you do not use subnets delete all characters in this field Default is ffffff00 0 default or 1 1500 or 4352 default It must be equal to or greater than offnet MTU for a given FDDI interface Note Set MTU to 1500 if you use an internetwork bridge to connect an FDDI D network and Ethernet network Ethernet supports a smaller MTU The smaller value allows all traffic to be forwarded between the FDDI and Ethernet networks
57. as ttya and ttyb To configure the TTY ports use the SetupTty command The flowchart in Figure 2 3 describes what you need to do when configuring the TTY ports Note The ttya port is preconfigured to support a system console Do not use i SetupTty to change the configuration of this port Use serial port B Do nothing Run SetupTty No Use terminal on port B Set up configure and attach modem Yes Modify terminal type in etc ttytab No Modify modem type in etc remote Hayes compatible modem Yes Do nothing Figure 2 3 Flowchart for configuring a TTY port The following describes the functions of the two serial ports a The first serial port ttya is preconfigured as a console port You must configure the console terminal to match the configuration of ttya a The second serial port ttyb can support a modem or a terminal Use SetupTty to configure ttyb for the device you intend to attach to the second serial port Note If you specify ttyb as a modem port SetupTty configures it to support a i default modem type of hayes If you plan to connect a Hayes compatible modem no configuration of this port is required If you plan to connect a non Hayes compatible modem edit etc remote to change the default settings manually 2 18 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX SetupTty e
58. authorized personnel should use the other options 10 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Displaying Configuration Data A dentered in the ax_config main menu prints a header listing basic system configuration information such as customer name the system s serial number installation date date of most recent hardware upgrade and name of field service representative Auspex authorized personnel review the header information and edit it if necessary when making hardware or software changes to the system The software probe data and hardware probe data follow the header If the ax_config command uses the a option hardware data from var adm config report also displays The following list describes special symbols used in the configuration report x s in the part number field When the part number is xx xxxx the line is used as a delimiter to separate the devices that can store Field Replaceable Unit FRU identifiers from those that cannot When the revision number or serial number cannot be determined a question mark is printed Numbers in parentheses If the number is in an entry for a drive or board the number is the firmware revision number If it is for a memory module the number is the size of the memory module After a hardware upgrade the information stored in var adm config report may become obsolete The Auspex authorized engineer who upgrades the hardware for your server
59. board Procedure for recovering from a system crash Network related issues including commands for improving the server s connectivity and availability such as ax_arp ax_netstat and nfsstat Note For information on optional software products including installation and documentation refer to the Auspex Premier Software Series CD ROM Special Maintenance a _ 10 1 Configuration Management The ax_config command is a valuable aid in configuration management It displays and modifies system configuration data which is stored in the var adm config report file Only Auspex authorized personnel should modify the data The data is particularly useful when you troubleshoot the server and try to determine whether you need to replace a hardware unit ax_config extracts system configuration information from three sources a var adm config report which contains information reflecting the system configuration the last time ax_config modified var adm config report Software modules which ax_config probes to obtain the software version numbers Hardware devices which ax_config probes to obtain hardware part numbers Caution The data format in var adm config report is critical for ax_config Do A not manually use a text editor to modify this file When you invoke ax_config in verbose mode that is when you enter ax_config v messages similar to the following appear indicating the types of information being extracted Loadi
60. board is seated properly If this fails to resolve the problem contact your Auspex representative for assistance FAIL There is a pro ED IN ERNAL FAI ED IN ERNAL DIAGNOSTIC CHECK LOOPBACK CHECK blem with the SP Contact your Auspex representative for assistance The HP cannot communicate with the SP Check to be sure the board is seated properly If this fails to resolve the problem contact your Auspex representative for assistance BOOT DRIVE READ LABEL FAILED The SP was unable to read the label from the boot drive Check the SCSI cables and connectors Try replacing the SCSI cable connecting the SP to the boot drive Check the boot drive Try starting up the system with a different boot drive label found attempting boot anyway The label on the boot disk was read correctly but contained bad information Possibly a problem exists with the installation procedure Try reinstalling the software on the boot drive Try replacing the boot drive Corrupt label The label on the boot disk was read correctly but contained bad information A problem may exist with the installation procedure Try reinstalling the software on the boot drive Try replacing the boot drive The following messages may also appear These messages give trained service personnel more detailed information about the fault that occurred sp hard reset sp status X X X X sp_loopbac
61. boards the file names are core3iop_npn m and core3iop_fpn m As in the file names for other core dumps the n in the filename is the number of the board the first NP in the system is numbered 0 the m is the suffix of the core dump file Attempts to reboot Runs fsck on the file systems when rebooting If no inconsistencies are found during fsck the NetServer resumes multi user operations Note To examine var adm messages or var crash hostname in single user mode first mount var it is not mounted by default in single user mode Crash dumps can accumulate in var crash hostname until they fill up available disk space Save the files to tape if appropriate and delete periodically You can also disable the saving of crash dumps by commenting out the following lines in etc rc local mkdir p var crash hostname echo n checking for core dump intr savecore var crash hostname echo To enable the saving of crash dumps to a specified directory leave the lines uncommented 10 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Sometimes the system hangs without crashing In this situation you may want to force a crash dump for debugging purposes To do this first abort to the PROM monitor prompt by pressing the Break key Then enter HP gt go 0 Note Remember to type a space between the command name and its argument This forces the crash dump to be saved in t
62. described in the examples are unrecoverable errors because only unrecovered errors cause file system isolation File System Isolation Caused by Media Errors That Cannot Be Fixed The following sample procedure shows how to recover from a file system that was isolated because of multiple media errors 1 Enter the following command to determine which file system has been isolated grep isolate var adm messages Nov 19 17 05 02 hostl syslog ax_isolated Filesystem dev vp80 mounted on export root is isolated 2 Determine which SP handles the file system with the error In this example the error occurred in vp80 which is supported by the first SP that is SP0 Examine the SP error messages in var adm messages by entering the following command grep SPO var adm messages Nov 19 17 04 50 hostl1 SPO 1196 Disk Media Error slot 36 drive 1 sense key 3 sense code 12 sector 3884624 Unrecoverable Disk or File System Problems in Multiuser Mode a 6 15 AUSPEX W Nov 19 17 04 51 hostl1 SPO 1196 Disk Media Error slot 36 drive 1 sense key 3 sense code 12 sector 3884625 Nov 19 17 05 51 host1 SPO 1196 Disk Media Error slot 36 drive 1 sense key 3 sense code 12 sector 3884632 Nov 19 17 05 52 host1 SPO 1196 Disk Media Error slot 36 drive 1 sense key 3 sense code 12 sector 3884633 These messages show that multiple media errors occurred in sectors that are close to one another which may be the result of a faulty re
63. drive is not recognized by the SP simply remove the drive without using ax_hot_plug or ax_remove_device Caution Do not remove a disk while it is spinning down Removing a disk drive while it is still spinning can cause permanent damage to the drive 3 If you have modified etc vpartab in step 1 run ax_loadvpar to load the virtual partition table 4 Run fsck p to check all the file systems and then press Ctrl D to go to multiuser mode The system boots successfully without the failing drive but file systems on that drive are no longer available to users Disk Problems at System Reboot a 6 7 Recoverable Disk Problems in Multiuser Mode A disk may be visible when the system reboots but inaccessible when the system tries to read from or write to it in multiuser mode When this happens the SP tries to reach the disk up to 10 times If one of the retries succeeds the error is considered recoverable and the SP generates a message stating that an error was recovered For example Recovered write drive fault slot 3 drive 1 The SP always generates an error message after each failed retry The following example shows an SP error message for a failed attempt to access the drive 1193 Drive Fault slot 10 drive 1 sense key 4 sense code 44 Note The slot number not the drive number in the SP error message shows li the location of the drive The drive number shows which drive in a particular slot is causing the error It
64. each processor a network and disk I O rates and I O types a file system operation loads mixes and rates 4 SP loads write accelerator usage and write operation rates a cache usage and age distributions for network file and disk subsystems ax_perfmon gathers data over user selected intervals and integrates the data over that interval That is if the interval is five minutes ax_perfmon measures total activity for each parameter for five minutes and then divides the results by 300 seconds to give the average rate for the parameter Gathered this way the data reflects average demand performance To obtain data that reflects peak demand performance select a very short interval seconds or a fraction of a second 9 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The data displays in multiple screens each containing different types of performance related data System summary Summarizes CPU utilization for each processor traffic over the network interfaces LFS file system operations data cache activity and disk activity for disks supported by SPO which are the drives in the NetServer base cabinet If more than one SP is installed in your system pressing the S key displays a System Summary screen with disk statistics from the next SP NP statistics Displays CPU utilization information about the network processor on an NP Also displays the statistics for frames received and sent by the network interface
65. field unchanged NetServer Configuration Commands a 2 9 Input fields are enclosed by brackets These fields require you to type an explicit value or to accept the default value a Below the dotted line at the bottom of the screen the commands for controlling the form appear for example commands for resetting the form and accepting the form This area is also a message area displaying explanatory messages when the cursor is on an input field and error messages when user input is invalid All configuration forms can be scrolled If part of the form is invisible because the form is larger than the screen or window press the Tab Down Arrow or Right Arrow key until the desired field appears The string MORE followed by a number on the dotted line at the top and bottom of the screen indicates the number of lines currently not displayed Table 2 2 lists the commands and keys that control form functions Table 2 2 Keys for editing configuration forms Key Function Selects the HELP function which displays a help screen Ctrl C Cancels this form and returns to the shell prompt Ctrl F Executes the form Ctrl H Backspace or Delete Ctrl L Ctrl R Ctrl U Ctrl W Home Page Up Page Down Space Bar Tab or Right Arrow Return Down Arrow Left Arrow Up Arrow NSconfig Deletes the previous character Repaints the screen for X terminals only Resets each field to the default value Deletes the
66. files on this file system are ignored In fstab entries for file systems containing client root directories be sure to specify the nosuid option This option minimizes security violations because it prevents users on the clients systems who have root permission on their root directories from creating setuid programs in their root file systems and executing them on the server Of course the client s etc fstab must be configured without the nosuid option so the client has root Warning active dev ad0a dev ad0b dev add0g dev ad0f dev ad0d dev ad0e dev ad0h dev acdl dev ad2a dev ad2g dev ad2f dev ad2d dev ad4a dev ad4g dev ad5a dev addg dev vpl dev vp5 note by convention the pass count field of the and usr entries should be left at 0 so that and usr do not get fsck ed again See fstab 5 which may differ from SunOS fstab 5 var usr openwin export home cdrom BackupRoot BackupRoot usr BackupRoot var BackupRoot usr openwin 4 2 exportl root exportl swap export2 root export2 swap Concatenated virtual partition home disks Figure 4 10 Default fstab file permission on its own root directory Mapping LFS File Systems to File Processors If the server contains multiple FPs it needs to decide which FP to use to mount a file system You can either specify the FP assignment in etc fstab or let the server map th
67. from an Existing Mirrored Partition 5 21 Moving a Striped or Concatenated Partition 00 eee ee 5 21 Moving a One Membered Mirrored Partition 0 000 000 08 5 22 Moving a Two Membered Mirrored Partition With No Unmount 5 22 Displaying Disk Configuration Information 000048 5 23 Chapter6 Recovering From Disk or File System Failures ADO This Chapter riere rriar oian Sameer eae ee Geeta a tae eo ae 6 1 When Does a File System Become Unavailable 0000000 6 2 Disk Problems at System Reboot 42 6 is0is ca gicgeat 6 acai ge aon ee won apes ae ke on a 6 3 Is the Problem Caused by Poor Disk Installation 6 4 Is the Problem Caused by a Malfunctioning Drive Slot 6 4 Is the Problem Caused by a Disk Without a Disk Label 6 4 Replacing a Malfunctioning Disk 0 000 eee eee ee 6 5 Replacing a Drive in Single User Mode 0 000000 000 6 5 Entering Multiuser Mode and Replacing the Drive 6 6 Using the System Without the Failing Drive 6 7 Recoverable Disk Problems in Multiuser Mode 000 008 6 8 Unrecoverable Disk or File System Problems in Multiuser Mode 6 9 Error Messages Indicating Disk or File System Problems 6 9 Pile Sy stent IS latiofi se Scans merir tos Mele Se eee Phd AE ALEEA 6 10 Types of Erro
68. from topic to topic within the manual Hypertext links are f ie text in the online documents This section covers the following topics a Installing online documentation a Starting the online documentation a Overview of online documentation A Printing a document Installing Online Documentation All of the online documentation is on the Auspex Premier Software Series CD ROM in a package named AXdocs The package includes the system manuals hardware manuals and Optional Product manuals Use the pkgadd 1M command to install the online documents Note The pkgadd process sets userid to install or root if there is no install D userid before adding a package If your system has an install userid that does not have super user permissions then the pkgadd may fail Either give the install userid super user permissions or remove the install userid Before adding the package you must find enough space for the package in a directory The pkgadd command calculates the size directory needed and displays the size In the following procedure the size given is 27 000 kilobytes but the actual size for your installation may be larger Find a directory of at least 27 000 kilobytes then use the pkgadd command to get the actual size needed You may need to do a second pass with pkgadd if you need to find a larger target directory 1 Mount the Premier Software Series CD ROM for example mount rt hsfs dev acdl cdrom This command
69. hosts bootparams b Enter the following commands on the NIS master cd var yp make Log in as root on the NetServer to be configured The system must be in multiuser mode for this procedure Execute SetupClient SetupClient Figure 2 11 illustrates the SetupClient form Table 2 9 describes the fields in the SetupClient form Auspex NetServer SetupClient Form Mon Sep 18 15 04 Setup Client Option CAdd Remove Client hostname CLIENT noname Architecture type ARCH sun4c___ SunOS version VERS 4 1 4___ lient Information Internet address 000 000 000 000 Ethernet address 00 00 00 00 00 00 Type of NIS service gt none Cclient none Swap size 32M e g 24M 32M or 64M FileServer partition EXPORT fexportt Path to Client s root EXPORT root CLIENT Path to Client s swapfile EXPORT swap CLIENT Path to executables SEXPORT exec ARCH sunos VERS__ Path to kernel executables SEXPORT exec kym ARCH sunos VERS___ Path to fusr share files SEXPORT share sunos VERS__ Path to zh home Ube TAB CR or arrow keys to move among fields when done hit CTRL F to execute P HELP F EXECUTE C ABORT L REDRAW SCRN 4R INITIAL VAL lt sp gt SELECT VAL Figure 2 11 The SetupClient form NetServer Configuration Commands a 2 29 Table 2 9 The SetupClient form fields Field Description Setup Client Option Client hostname Architecture type SunOS version Internet
70. independent of the server name Although you can use aliases in etc hosts for a host name specify the primary name not an alias in this field For example if etc hosts contains an entry as follows 144 48 252 2 MKT marketing specifying marketing for 144 48 252 2 in the form causes NSconfig to comment out the above entry and to make marketing the primary name for the interface The entry for this interface in etc hosts file will be changed to the following 144 48 252 2 MKT marketing 144 48 252 2 marketing Unless the server hostname already has an IP address associated with it in the etc hosts file one of the primary hostnames must be the same as the server hostname Otherwise an error message Hostname unknown appears when you use the hostname command Internet address for each Network interface A mode for Ethernet and FDDI interfaces Yes turns on the REFLECT mode for that interface which then reflects back all NFS responses from the interface to the MAC address from which the request arrived REFLECT bypasses normal IP routing See the ifcontig 8C man page for further information 2 14 a Version 1 10 System Manager s Guide UP or Dwn The default name is the server name followed by en 10Base T Ethernet interfaces ahmen 100Base T Ethernet on an HME adaptor or fn FDDI interfaces where nis the interface number NOTE ATM and interfaces are available
71. is a daemon that manages the state of virtual partitions ax_vpstat 8 displays the status of virtual partitions ax_mattach 8 attaches a new member to a mirrored partition ax_mdetach 8 detaches a member from a mirrored partition ax_mconvert 8 converts a striped or concatenated virtual partition to a one membered mirrored partition ax_mrestore 8 restores mirrored partitions to an active state Virtual Partition Commands on page 5 10 Virtual Partition File on page 5 9 Expanding a One Membered Mirrored Partition on page 5 18 vp man page Commands for Auspex devices The device ad for Auspex disk and CD ROM was added to the device options that may be used with boot 8S The command MAKEDEV auspex defines Auspex specific devices The dkinfo 8 and format 8 commands are modified to support Auspex disks eject 8 supports acd for Auspex CD ROMs Also ax_label 8 partitions and labels a drive and ax_Islabel 8 shows a disk s label and partitioning The ax_hot_plug 8 command provides the ability to hot plug drives that is the ability to add replace or remove disk tape or CD ROM drives while the operating system is running Removing and adding drives can be done from a single command line Two additional device commands ax_add_device 8 and ax_remove_device 8 also provide hot plug capabilities Device Names on page 4 2 Using the Format Command on pag
72. is set to the proper mode full or half duplex The 100Base T Ethernet interface uses either an RJ45 or an MII connector The connector is labeled 100Base T If the connector is cabled to a 100Base T Ethernet hub on the other end the interface functions at 100Base T speed If the connector is cabled to a 10Base T Ethernet hub on the other end the interface functions at 10Base T speed In the same way to run in full duplex mode the connector must be cabled to another full duplex interface or to an Ethernet switch that supports full duplex operation on the other end Sun clients set up for Ethernet communications use auto negotiation to determine the duplex mode of the other end of the connection When connected to a NetServer through a switch a hub or directly back to back the auto negotiation feature causes unpredictable results A Sun client using any such connection must be manually set to the proper duplex mode See the Sun client documentation for details on setting the duplex mode 2 32 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Booting Up and Shutting Down the NetServer About This Chapter This chapter provides procedures for accessing the PROM monitor and booting a server It also describes the commands that shut down a server This chapter assumes that you have finished the initial server configuration described in Chapter 2 Before you boot the server familiarize yourself with the major changes tha
73. level 0 is a full system backup and each subsequent dump level 1 to 9 backs up only those files that have changed since the last dump of a lower level The counterpart of the dump command is restore which recovers file systems that were backed up by dump Dump Command The dump command has the following syntax usr etc dump options tape_device_name file_system_to_dump Table 8 1 describes the command arguments Refer to the dump 8 man page for more information Table 8 1 Arguments for the dump command Argument Meaning options The most commonly used options are listed as follows 0 9 Dump level you have chosen b Blocking factor number of blocks written at a time d Tape density f Name of the device to which dump backs up the file system s Size of the tape in feet u Writes the date of the dump and partition name to etc dumpdates T Date string you want dump to insert in etc dumpdates By default the current time at which dump starts is supplied Use this option only when you use a script to dump a file system that has been cloned by ax_clonefs In the script create a timestamp such as d1 using a date string to specify the time when ax_clonefs finishes copying the file system This ensures that the next incremental dump backs up all the changes since the last time the file system was cloned The timestamp should use the following date string date a Gh sd ST 19 y The following is an example of a dum
74. line to var spool cron crontabs root so your server automatically sends the email message 00 5 usr auspex reporter sh gt dev null 2 gt amp 1 The script file that sends the email message is usr auspex reporter sh The message uses a subject line that includes the string you specify in the Customer Site field in the NSconfig form Refer to the man page for reporter sh 8 for more information reporter sh calls nfsstat 8C with the z option for network file statistics The z option zeros out and reinitializes the nfsstat statistics fields If you do not want the fields to automatically reinitialize edit the script usr auspex reporter sh Comment out the nfsstat command that uses the z option and uncomment the nfsstat command that does not use the z option If you set up your system to mail weekly reports you can also modify the destination address to add recipients other than Auspex to the mailing list You add and modify recipients by editing the following line in usr auspex mailmessages sh and usr auspex reporter sh REPORT_STAFF dl1l customermessages auspex com You can also direct the report to a local user by adding his or her email address to the line For example REPORT_STAFF dl1 customermessages auspex com jsmith abc com New Installations a 2 7 AUSPEX W Additionally you can specify recipients for system core dumps By default usr auspex mailmessages sh identifies the recipient in the fol
75. member listed in etc vpartab to be the Dirty partition ax_mrestore copies the data from vp1 to vp2 Read operations continue on vp1 only write operations go to both vp1 and vp2 ax_mrestore completes and normal mirror operations resume vpi vp2 IHi ai System crashes ACTIVE Hi ESTORING i Le lt lt Figure 5 7 Automatic recovery from mirrored partition error Caution If at any point you want to stop ax_mrestore 8 and return the state A of the partition being restored to DIRTY use the kill 1 command with no signal arguments That is use only the process ID of ax_mrestore in the kill command Using kill with a signal argument terminates ax_mrestore immediately leaving the partition stuck in the RESTORING state until the NetServer is rebooted Automatic Error Recovery on Mirrored Partitions a 5 13 Defining and Reconfiguring Virtual Partitions This section describes how to use the virtual partition commands to create and manipulate virtual partitions The procedures described include the following most of them require you to log in as root a Defining a virtual partition a Expanding a concatenated or striped partition a Changing a striped or concatenated partition to a mirrored partition a Adding anew member to an existing mirrored partition a Removing a member from an existing mirrored partition Recommendation Print out a copy of
76. mounts the CD in drive slot 1 on cdrom 2 Mount usr with read write privilege mount o remount rw usr 3 Add the document package to the server by entering the following command t pkgadd d cdrom The system returns The following packages are available 1 AXEC1 etherchannel HP VII HP VIII 1 10 2 AXNTBios Auspex NetBIOS HP VII HP VIII 1 10 3 AXNeTsrv Advanced Server for UNIX Systems B 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX HP VII HP VIII 1 10 4 AXatm2 ATM 2 HP VII HP VIII 1 10 5 AxXbackup FastBackup HP VII HP VIII 1 10 6 AXdgrd DataGuard HP VII HP VIII 1 10 7 AXdocs Auspex System Documentation HP VII HP VIII 1 10 8 AXdrvgrd DriveGuard HP VII HP VIII 1 10 9 AXftp NP Resident FTP HP VII HP VIII 1 10 10 AXsrvgrd ServerGuard HP VII HP VIII 1 10 Select package s you wish to process or all to process all packages default all q Enter the number 5 for the 1 10 document package Processing package instance lt AXdocs gt from lt cdrom gt Auspex System Documentation HP V HP VI HP VII HP VIII 1 10 The following question needs a path to the destination directory for the document package not a path to the source on the CD Enter a question mark Enter path to package base directory q Installation of this package requires that a UNIX directory be available for installation of appropriate software This dir
77. move among fields when done hit CTRL F to execute Q HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VAL Figure 6 1 The NSinstall form Table 6 1 The NSinstall form fields Field Description Possible values CD ROM slot The number of the slot where the Auspex Default is 1 It can be any number NetServer Software CD ROM is inserted appropriate to your server Slot numbering starts from 0 Target disk slot The number of the slot where the target The target drive can be in any slot so drive is inserted For example to repair a it can be any number appropriate to damaged root drive specify 0 To install file your server Slot 0 is allowed only if systems to a spare root drive enter the your system was booted from the number of the slot where the spare drive is CD ROM Default is 3 inserted Target Host Type of HP with which the newly installed Current the HP currently installed Processor Type software will be used on the current server and HP VIII Default is Current Using NSinstall to Install Software From CD ROM a 6 31 Table 6 1 The NSinstall form fields Continued Field Description Possible values Reboot after Whether the system automatically reboots Default is No Consider Yes only if install in single user mode after the installation you install the software to drive 0 Install root Partition newfs first Install usr Partition newfs first Ins
78. network interface Aa nfsstat 8C which is part of SunOS is modified by Auspex to display NFS statistics per interface The following three sections describe how to use these commands For further information on the syntax refer to the corresponding man pages Displaying and Flushing ARP Tables Used by an NP Board Because a NetServer can have multiple network processors and each of them maintains a different ARP table Auspex provides a command to manipulate ARP tables per board The command for flushing and displaying the ARP table is auspex ax_arp which has the following syntax ax_arp f d I interfac where interface is a network interface number such as ae0 for an Ethernet interface or afddi0 for an FDDI interface Note The output of the command pertains to the entire board that contains the specified interface To display the current entries in the ARP table use the a option The following is an example of the ax_arp a command and its output host0 gt ax_arp a I ae0 hostl el 144 48 252 4 at 0 0 3c 0 39 1d This example shows the ARP table used by host0 ae0 contains one entry which resolves the IP address 144 48 252 4 of host1 e1 to the physical address 0 0 3c 0 39 1d The f option displays and flushes the ARP table used by an interface Flushing a table means deleting all table entries The following shows the ax_arp f command and output host0 gt ax_arp f I ael hostl el 144 48 252 4 at 0
79. nfsd ax_nfsd rarpd bootparamd statd lockd link editor directory cache ax_isolated checking for crash dump preserving editor files clearing tmp standard daemons update cron uucp starting network daemons inetd printer xdm Fri Nov 8 14 31 53 PDT 199 hostname login a7 Standard Boot Messages a _ E 5 Table E 1 NetServer boot event explanation Ref number Explanation 1 Boot process initiated by Host Processor EPROM 2 Power on self tests running in RAM 3 Host Processor PROM banner 4 Memory initialization and testing 5 EPROM boot messages 6 Software copyright and version information 7 Host Processor type 8 Host Processor memory size 9 Probe of HP SCSI bus 10 Serial port autoconfigured 11 IOP boards NP boards and SP boards detected and autoconfigured 12 Labeled disk drives detected and autoconfigured an unlabeled disk drive does not appear in the list 13 SBus cards on the IOP boards detected 14 Initiate start up sequence for Auspex processor boards 15 Write Accelerator initialization 16 Configuring network interfaces 17 Host login prompt E 6 A Version 1 10 System Manager s Guide AUSPEX W W AUSPEX System Startup Error Messages System startup error messages may be received from the PROM monitor or from boot programs SP SP SPs SP No VME STATUS REG ISTER ACCESS FAILED There is a problem accessing the SP Check to be sure the
80. no unwritten data is found in the cache at initialization the write cache state is set to OFF If unwritten data is found checksums validate the integrity of the cache control data If the control data is valid the contents of the cache are written to the appropriate disk After all the data in the cache is written successfully the state switches to OFF If unwritten data is found in the write cache at initialization but the cache control data is invalid the state switches to BADCHECKSUM If the data in the cache cannot be written to disk because of disk errors or disk changes the state is set to DIRTY The ON state indicates the write cache was successfully enabled The voltage of the battery is low and the write cache is disabled This state occurs when no NVRAM is installed on the SP If the write cache state is BADCHECKSUM or DIRTY after initialization the server enters single user mode and you can use the ax_write_cache command to change the state of the write cache as described in Managing the Write Accelerator on page 7 9 7 6 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Uniatzes Da Replace battery on y write accelerator and reboot server etc rc boot tries to initialize y Yes Battery low py Battery low No 1 No Unwritten data in write cache Yes Data integrity OK y Yes SP writes data to disk Badchecksum N
81. number of partitions and the individual partition sizes vary according to the partition table the disk uses Note When describing the size of a partition the term MB represents 2 048 D disk sectors which is equivalent to 1 048 576 bytes When used to describe the size of a disk MB is 1 000 000 bytes As a result the size of the c partition may look smaller than the size of the entire disk For example the c partition of a 1 GB 1 000 MB disk is described as 954 MB in the partition table In most cases Auspex labels the disks at the factory using the default partition table Use this default partition table for clients root swap and home directories and the stripe drives for striped virtual partitions see Different Types of Virtual Partitions on page 5 4 The root partition table is used for the root disk ad0 If the current partitioning scheme on a drive does not meet your needs use the format or ax_label command to repartition your disk For example if the 4 GB drive you received from Auspex is partitioned as a data disk and you want to use it as a root disk use format or ax_label to change the partition table to root Refer to Using the Format Command on page 4 36 or Using ax_label to Partition and Label a Disk on page 4 41 for information on how to partition a disk Table 4 5 through Table 4 7 list the partition sizes for the default stripe and root partition tables as specified in etc format d
82. operations varies greatly from one file system to another balance the load among FPs manually by adding the fp n option to the appropriate entries in etc fstab File System Formats Supported by UFS The format of a UFS file system on the HP or an FP can be either of two types a Standard UNIX 4 2 BSD This format is sometimes called the Fast File System FFS or Static Table Format a Fat Fast File System FFFS This format is sometimes called the Dynamic Table Format or the Tahoe File System It is the default format for LFS file systems The primary advantage of the FFFS format is that it supports more inodes per cylinder group and more cylinders per cylinder group than the FFS format FFS is limited to a maximum of 2048 inodes and 16 cylinders per cylinder group The fsck 8 command has an option that allows you to convert from FFS to FFFS file systems If you execute fsck 8 with the c option the specified file system is converted from FFS to FFFS format The output of the dumpfs 8 command indicates which format dynamic or static is used by the file system being examined and tunefs 8 handles both formats For more information refer to the fsck 8 dumpfs 8 and tunefs 8 man pages 4 22 a Version 1 10 System Manager s Guide W AUSPEX NFS Version 3 and UFS Limitations Client operating systems query the rpc daemon about NFS Version 3 support With the addition of NFS Version 3 in release 1 10 the client receives
83. partition for example dev vp1 You must number the virtual partition according to the SP that controls the partition Table 5 1 lists the valid partition numbers Virtual Partition Driver File and Commands a 5 9 Table 5 1 Numbering virtual partitions SP that manages the virtual partition Virtual partition numbers for SP boards SPO SP1 SP2 SP3 SP4 0 255 256 511 512 767 768 1023 1024 1279 2 The second field indicates what kind of virtual partition is being defined If the partition is a striped partition it also indicates the stripe size For example if you specify striped size 128K the first 128 KB of the virtual partition is from the first physical partition the next 128 KB is from the second physical partition and so on 3 The third field lists the physical disk partitions making up the virtual partition for example adih ad2h ad4h Virtual Partition Commands The following list describes the commands and a daemon for creating and managing virtual partitions a ax_diskconf 8 prints a report showing the current disk configuration of the NetServer The report shows the physical and virtual partition to which each file system belongs For more information on ax_diskconf 8 refer to Displaying Disk Configuration Information on page 5 23 a ax_expand 8 enlarges a striped or concatenated virtual partition without first unmounting the file system For more information o
84. partition resides on a separate disk drive Both member virtual partitions must be on the same SP as well as on the same SP as the mirror itself 5 8 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Virtual Partition Driver File and Commands This section describes the driver file and commands that manage virtual partitions Virtual Partition Driver vp 4 is the device driver that acts as the interface to virtual partitions constructed from physical disk partitions Virtual Partition File vpartab 5 is a file containing the table that describes virtual partitions Edit this file to add or modify virtual partition definitions Figure 5 5 show an example of the file etc vpartab Virtual partitions 1 and 2 Concatenated dev vpl concat adlh ad2h ad4h dev vp2 concat ad4d ad5c Virtual partition 3 Striped dev vp3 striped size 128k ad7h ad8h Virtual partition 4 Concatenated This is necessary because physical partitions cannot be mirrored and virtual partition 5 is constructed by mirroring physical partition ad8c and virtual partition 3 defined above dev vp4 concat Virtual partition 5 Mirrored dev vp5 mirrored Figure 5 5 Sample vpartab entries Comment lines begin with a pound symbol and are ignored Each virtual partition definition in the table consists of the following tab separated fields 1 The first field contains the device name of the virtual
85. section describes how to replace a drive while the system is in single user mode Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 1 Comment out the entries in etc fstab and etc vpartab that reference the bad drive To comment out an entry type a sign in the first column of the entry 2 Run ax_loadvpar to load the revised etc vpartab 3 Use ax_hot_plug to replace the drive For example if the drive is installed in slot 4 enter the following ax_hot_plug add 4 If the drive is not recognized by the SP simply remove the drive and insert a new one in the slot at the prompt Caution Do not remove a disk from a slot while the disk is spinning down Removing a disk drive while it is still spinning can cause permanent damage to the drive 4 Remove the signs in etc fstab and etc vpartab that you added in step 1 5 Run ax_loadvpar to load the virtual partition table 6 Use newfs to create the file systems that used to exist on the replaced drive For example if ad4c is a member of vp2 enter the following command newfs dev rvp2 Disk Problems at System Reboot
86. server automatically maps file systems to FPs when you execute the mount a command However if etc fstab contains an fp n option the server assigns the FPs using the default method as described in the Using etc fstab to assign file systems to FPs section For example if etc fstab contains the following entries the server assigns backupdata and disks to FPO and cad to FP1 dev ad3g backupdata lfs fs 4 2 1 2 dev ad4g disks lfs fs 4 2 1 2 dev ad3g cad lfs fs 4 2 fp 1 1 2 Also if you specify an invalid FP number in the fp n option the server automatically assigns the file system to FPO and displays two warning messages For example if your server contains fewer than four FPs and you specify fp 3 for the export file system in etc fstab the server assigns the file system to FPO and generates these messages dev ad3 invalid FP3 specified mount_lfs on export Invalid file processor specified mount_lfs mounting export on FPO Note If no fp n option is specified in etc fstab the mount a command distributes file systems evenly on the FPs This distribution however does not guarantee that the load is evenly spread out over all FPs because the amount of activity on each file system varies You might need to assign file systems manually using the fp n option if some file systems are more heavily used than others If you want to see the mapping between FPs and file systems while you are mounting t
87. server manually Only Auspex qualified field personnel need the information about executing diagnostics commands Entering Monitor Mode When the System is Powered Off Caution Do not use this procedure after the automatic boot process has A started or while the operating system is running File system damage and data loss can result 1 Locate the power switch on the power distribution unit PDU Refer to the hardware manual for your NetServer for more information on the PDU location 2 Set the switch to the ON position 3 Wait while the system performs the power on self tests If the self tests execute successfully this message appears Self test passed 4 Once this message appears escape the automatic boot procedures by pressing the Caps Lock and Pause Break keys simultaneously on a DEC VT510 terminal the Break key on an X or a Link terminal or the Shift and Break keys simultaneously on a Wyse terminal Booting the NetServer a 3 3 5 The system enters the PROM monitor The monitor prompt appears as follows HP gt The monitor prompt indicates that the system is ready to accept PROM monitor commands Entering Monitor Mode When the System is Running To access the PROM monitor when the operating system is running halt the operating system Use one of the following procedures depending on whether the server is running in single user or multiuser mode a In single user mode Halt the operating system using the hal
88. server that exports the file system netserver1 e5 in this example You can obtain the name of an interface from the NIS database or the etc hosts file on the server on which you modify fstab if NIS is not used at your site Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 4 24 a Version 1 10 System Manager s Guide W AUSPEX Adding or Replacing a Drive The changing needs of your site may require that you add more disk tape or CD ROM drives This section describes the procedure used to configure new drives in the system The procedure includes such functions as a installing a new drive or replacing an existing drive a checking the partition table on a new disk drive a initializing a new file system on the new disk drive 4 mounting and exporting the file system to client workstations Note See the DriveGuard Manager s Guide for adding or replacing drives in a i DriveGuard array Drives are added into vacant slots in an installed HDDA drawer or drive rack The slots in an HDDA drawer accept only disk drives The slots in a drive rack accept the following types of dri
89. sl sl sl sl s sl sec SEC sec sec 128 gt 128 gt 128 gt 128 gt ASPX net acc root on ad0a swap on ad0b dump on adOb unit 1 fstype fstype fstype 4 2 checking root and usr filesystems dev rad0a dev rad0a dev rad0g rc auspex running ax_c ax_write_cac ax_write_cac ax_write_cac ax_write_cac ax_write_cac ax_write_cac rc auspex dev rad2c dev rad0f dev rad0od dev rvp3 i dev rvp6 i dev rvp9 i Nov 8 14 rc mounting Fri c auspex tarting NIS c a s s ypset y s s hecking quotas 9009 files 249 frags is stable Running ax_s hkdrive he SPO ini 20571 used 4784 blocks tartup at iopl SBus slot 4 spec size 409600K spec size 409588K 38521 free 0 4 fragmentation tializing write cache he SP1 ini tializing write cache he SPO he SPO he SP1 he SP1 ena wri ena wri Running from rce boot is clean is clean is clean s clean s clean s clean 231721 PDT 4 2 and done Running from rc services this is insecure pbind ypupdated tarting RPC key server bling write te cache is bling write te cache is st 1996 tarting rpc port mapper ypserv ypxfrdypbind ypsetme tarting Auspex key daemon network interface configuration ae0 flags 43 lt UP BROADCAST RUNNING gt inet 123 456781 ether 0 0 3c 0 a7 90
90. specific information Continued File name Purpose of the file etc hosts equiv etc inetd conf etc netgroup etc networks etc passwd etc printcap etc netmasks etc raidtab etc rc local etc rc local lt hostname gt etc rc boot etc remote etc rpc etc protocols and etc services etc sendmail ct etc syslog cont etc ttytab etc vpartab Defines trusted hosts See hosts equiv 5 Lists servers to be started by inetd 8C See inetd conf 5 Defines network groups recognized by the system See netgroup 5 Defines network name information See networks 5 Defines users password and account information See passwd 5 Defines printers See printcap 5 Defines network masks used to implement IP subnetting See netmasks 5 Holds the network masks that etc rc boot uses for running the ifconfig command etc netmasks is configured automatically by NSconfig See rc 8 Holds DriveGuard RAID configuration information Starts site configured and general programs at system boot time including NIS services keyserv other Sun utilities and local utilities etc rc local is configured automatically by NSconfig See rc 8 Starts daemons and does setup for local applications and machine specific services Sets ifconfigs for broadcast addresses etc rc boot is configured automatically by NSconfig See rc 8 Customizes serial port information used by tip 1 Contains informati
91. summary screen Disregard these messages because ax_perfmon can still successfully save data pertaining to the applicable parameters to the output file i Note If the filter file in an ax_perfmon command contains parameters not Figure 9 8 shows a sample output file created by ax_perfmon with a filter file Parameter names Header ths Data in one ax_perfmon Delimiter to separate the sample parameter names and the parameter values agrams Each number in a sample corresponds to one np UdpOutDatagrams parameter np IpForwDdtagrams np if np if np if e_i Interface number for example 0 means the first a interface and 1 means the np if second interface np if np if EGINNING_OF_DAT Aug 17 14 1995 80 298 179511 0 0 22 3054 4 Aug 17 14 23 1995 9 21463964 198 93634 0 0 8 1239 0 0 Aug 17 14 23 1995 10 18 58 58 167 85437 0 0 33 4315 0 0 Jun 17 14 23 1993 8 30 98 99 142 71924 0 0 149 19463 0 0 Jun 17 14 1993 23 18 11 112 402 1244 0 0 178 71613 0 Q np if Figure 9 8 Sample output file created by ax_perfmon with a filter file The first line of the output file is the header equivalent to the header on an ax_perfmon screen Following the header is a list of parameter names Refer to Types of Statistics Displayed by ax_perfhist on page 9 40 for more information on parameters BEGINNING_OF_DATA serves as a delimiter separating the list of parameter names and the actual para
92. system cy tows ees ea Res 4 15 The Auspex usr openwin file system 0 0 00 0 000 cece ec eee eee 4 16 The export file system for servers with SunOS executables 4 17 The jexportN filesystem for clients vs sauces aii Phone wx se eres 4 17 Default fstab file mezi snis iiin E Mau are Ae AN a 4 20 Virtual Partitions Example of a concatenated partition of three physical partitions 5 4 Example of a striped partition of four members 5 5 Example of a mirrored partition sc vince co eis ee psy dalle WR RS 5 6 Disk space truncated from members of a striped partition 5 8 Sample vpartab entries 0 cee eee es 5 9 Example from using the ax_vpstat command 5 12 Automatic recovery from mirrored partition error 5 5 13 Expanding a striped virtual partition using ax_expand s 5 17 Using ax_mconvert to change a partition from striped to mirrored 5 20 Figures a xiii Figure 5 10 Figure 5 11 Chapter 6 Figure 6 1 Chapter 7 Figure 7 1 Chapter 8 Figure 8 1 Figure 8 2 Figure 8 3 Figure 8 4 Chapter 9 Figure 9 1 Figure 9 2 Figure 9 3 Figure 9 4 Figure 9 5 Figure 9 6 Figure 9 7 Figure 9 8 Figure 9 9 Figure 9 10 Figure 9 11 Figure 9 12 Figure 9 13 Figure 9 14 Figure 9 15 Figure 9 16 Figure 9 17 Figure 9 18 Figure 9 19 Figure 9 20 Figure 9 21 Figure 9 22 Figure 9 23 Chapter 10 Figure 10 1 Ap
93. the s flag which causes the reboot not to perform a sync 2 operation on each file system The s flag ensures that the reboot does not hang if the NetServer has mounted a file system from another server that can not be reached See Commands for Shutting Down the NetServer on page 3 9 on page 3 9 Note When booting the NetServer verify that the diagnostic switch on the HP is set to the NORM position not the DIAG position Refer to your hardware documentation for more information a The shutdown 8 command includes the 1 option to send warning messages only to logged in users This option speeds up the shutdown process if the list of machines mounting files is not up to date or if machines mounting files do not have rwalld 8C support Single User Mode vs Multiuser Mode The UNIX operating system supports two modes of operation single user and multiuser In multiuser mode users may log in to the UNIX host whereas in single user mode only the system console is available Commands typed in single user mode execute with super user root privileges Furthermore many network services and daemons start only when the system enters multiuser mode Thus when the system runs in single user mode it is inaccessible from the network and it cannot serve up files for clients The NetServer boots automatically to multiuser mode at power on Under normal operating conditions you do not need to boot the system manually However some kinds
94. the fields in the NSconfig form In some cases if you select a certain value for a field an additional field appears on the form For example if you select Master Slave or Client for the type of NIS service you are prompted for a domain name These additional fields are indented in the table The table also lists the possible values for each field when appropriate fy Note The NSconfig log file NSconfig log is located in the directory var log 2 12 a Version 1 10 System Manager s Guide W AUSPEX f NetServer NSconfig Form Mon Mar 2 10 01 k Server Hostname SrvA Fast Ethernet Interface ahme0 Enable Disable Interface ahme0 gt UP Hostname SrvA Internet Address 192 222 222 2_ Reflect Mode gt NO Subnet Mask LfE ELLEOO Broadcast Type gt 0 Mode gt Manual Duplex Mode gt full Speed gt 100 Enable SMB on ahme0 gt N Fast Ethernet Interface ahmel Enable Disable Interface ahmel gt UP Hostname SrvA Internet Address 192 211 221 2 Reflect Mode gt NO Subnet Mask EFEFEF 00 Broadcast Type gt 1 Mode gt Manual Duplex Mode gt half Speed gt 100 Enable SMB on ahmel gt Y NetBIOS Node Type gt B X Console gt NO NIS service type gt slave Domain Name yourcompany com Create Default Route gt Yes Hostname or IP address of default gateway 192 11 11 1 Run RouteD gt NO Run SnmpD gt NO Change Date and Time gt NO Change Time Zone gt NO Notify Au
95. the files in etc to include site specific information Refer to Table 2 1 on page 2 5 for a list of site specific files c If you cannot determine what is wrong with the files restore the root partition from the most recent level 0 backup For example enter the following if the tape in slot 9 is the backup tape mount dev ad0a a cd a restore if dev nrast9 After restoring the files go to step 10 If the root partition is severely damaged do either of the following a Run newfs on the root partition of the damaged disk and restore the root partition from the most recent level 0 backup For example mount dev ad0a a ed a restore if dev nrast9 a Ifa backup is not available go to Using NSinstall to Install Software From CD ROM on page 6 29 to restore the root partition from the CD ROM After restoring the partition modify the files in etc to include site specific information Refer to Table 2 1 on page 2 5 for a list of site specific files If you restore the root partition from a backup you must run installboot on the root partition to make the disk bootable For example use the following commands to install the bootblocks in ad0a mount dev ad0a a ed usr mdec installboot a boot bootad dev rad0a To restore other damaged partitions that contain site specific information that can be easily edited or contain no site specific information follow the instructions in Using NSinstall to
96. the next section Types of Errors that Cause File System Isolation Two major types of error conditions cause the FP to isolate a file system a Errors detected by LFS a Disk problems Errors detected by LFS Errors detected by LFS are those file system corruption problems that can be fixed by fsck The following lists all file system corruption conditions that cause a file system to be isolated 1018 realloccg bad optim 1019 ialloc dup alloc 1038 alloccgblk cyl groups corrupted 1039 alloccgblk can t find blk in cyl 1040 ialloccg map corrupted 1041 ialloccg block not in map 1043 free freeing free block 1044 free freeing free frag 1046 ifree freeing free inode 1047 alloccg map corrupted 1048 alloccg block not in map 1070 isblock 6 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX 1071 clrblock 1072 setblock 1073 getmp bad magic 1074 ufs_statfs 1077 rwip type 1078 ufs_readdir dir reclen 0 1079 ufs_readdir dir reclen gt 512 1094 ufs_readdir_nfs dir reclen 0 1095 ufs_readdir_nfs dir reclen gt 512 The following examples of error messages in var adm messages are caused by a file system corruption problem Mar 15 14 04 19 host1 FPO 1095 ufs_readdir_nfs dir reclen gt 512 Mar 15 14 04 19 hostl FPO isolating filesystem mounted on disks home Mar 15 14 04 43 host1 mountd 270 rpc mountd encountered isolat
97. the problem and how to correct it Recovering from Permanent Disk Errors Without File System Isolation and Recovering from Intermittent Disk Errors Without File System Isolation give instructions on how to recover from a disk error that does not trigger file system isolation Repairing Disk Sectors After an Unrecovered Media Error describes how to fix errors on different types of disk sectors The procedures are applicable whether or not the disk sector causes file system isolation Error Messages Indicating Disk or File System Problems If the error is not recoverable error messages appear on the console as well as in var adm messages The error messages are generated by these sources A A A The SP Note that an SP error message gives the exact location of the problem disk sector The following is an example of an SP error message Nov 17 17 47 36 server_name SP1 1196 Disk Media Error slot 36 drive 1 sense key 3 sense code 12 sector 3884624 Note If the problem is caused by the file system and not by the disk hardware D the SP does not display any error messages The application program The FP or vmunix Because the FP resides on an I O Processor IOP board the message refers to the FP as IOPn FPn CPU B For example Nov 17 17 47 36 server_name IOPO FPO 0 CPU B vp2 fatal error e Permanent drive error on read block 9382416 The following example shows a vmunix error message
98. the release button until the tape is released Do not reuse the tape Common Tape Drive Errors a 8 21 Tape Drive Errors This section discusses the errors caused by confused tape drive logic or malfunctioning drive read write heads In some cases you need to replace the drive and send it to Auspex for repair Write Error Command Error messages Analysis Solution Online Error Scenario Symptoms Analysis Solution tar cvf dev rast2 vmunix ast0O fatal error d SP data transfer error SPO 1198 SP Data Transfer Error slot 2 drive 1 Sector 0 The tape drive cannot write to the tape Use the mt f dev nrast0 status command to display status information about the drive If the status is sense key 0x10 fatal the drive logic is confused This happens usually when the drive is not properly seated in the drive rack or the drive hardware is damaged Remove the tape drive and replace it in the drive slot to reset the tape drive logic After the drive executes the self test repeat the write operation If the error happens more than three times the drive hardware probably is damaged and you need to replace the unit The tape drive fails to return to the online status The tape drive LED does not indicate that the drive is online When online the 8200 drive has only the right LED on and the 8500 drive has only the top LED on The mt f dev nrastn status command returns sense key 0x0 no sense
99. to copy data from the old drive to the new drive For example if you are copying data from slot 4 to slot 9 enter the following ax_sputil copy 4 9 The copy operation copies the disk label as well as the contents of the disk For more information refer to the ax_sputil man page Alternatively you can use ax_clonefs to copy each partition from the old drive to the new drive 7 Use ax_hot_plug to remove the new drive from slot 9 and replace the old drive with the new drive in slot 4 ax_hot_plug remove 9 add 4 For more information refer to the ax_hot_plug man page 8 Mount and export the file systems on the new drive Repairing Disk Sectors After an Unrecovered Media Error A media error occurs when a disk sector rather than the entire disk is damaged The following two examples of SP error messages indicate that media errors occurred In an actual message nnn is the number of the bad sector Nov 17 17 47 36 host_name SP1 1211 Recovered correctable ECC error slot 21 drive 1 sector nnn Nov 17 17 47 36 host_name SP1 1196 Disk Media Error slot 22 drive 1 sense key 3 sense code 11 sector nnn How to recover from a media error depends on which disk sector is bad Because only a particular sector is damaged the recovery procedure involves remapping the bad sector and restoring the file affected by the bad sector instead of replacing the entire drive The following sections describe the procedures for recoverin
100. to copy the root partition from the backup root disk onto the repaired root disk b Run installboot on the restored root partition to make the disk bootable For example if the disk you are repairing is in slot 1 enter the following commands to install the bootblocks in ad1a mkdir a mount dev adla a cd usr mdec installboot a boot bootad dev radla 6 26 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX c Use halt to halt the processor When the system is in monitor mode replace the backup root drive with the restored root drive and boot the system to multiuser mode The root drive recovery procedure for a damaged root partition is now complete If the entire root disk was damaged and you are replacing it with a new disk use ax_sputil to copy the entire contents of the backup root disk onto the new root disk Use halt to halt the processor When the system is in monitor mode replace the backup root drive with the restored root drive and boot the system to multiuser mode The root drive recovery procedure is now complete Booting From the CD ROM in Single User Mode Use this procedure if you have no backup root disk or if your backup disk fails to boot 1 2 3 Use the halt command to bring the system down to monitor mode Install the distribution CD ROM in slot 1 Power up the NetServer Then boot the server in single user mode by entering the following command HP gt b a
101. to define the contents of all histograms first and load the statistics all at once later click on the 2 Step Load box in the File Load Window If this box is checked follow these steps to load statistics to the histograms 1 For each histogram follow all the instructions in Loading Statistics From a File on page 9 21 to select an input file a disk or virtual partition number if applicable and a parameter Notice that no statistics appear in the histogram after you select a parameter 2 Inthe File Load Window use the right mouse button to click on the Load Hist button In the pull down menu select Load Hist to load statistics to the current histogram or Load All Hists to load statistics to all histograms The 2 step load procedure is particularly useful if you have a large statistics file and you want to display multiple histograms It eliminates the need for waiting for each histogram to display before you can specify the parameter and properties for the next histogram Eliminating a Histogram from the Data Display Window To display fewer histograms remove the current histograms one by one For example if the Data Display Window contains four histograms follow these steps to change the window to display only histogram number 3 1 In the File Load Window click on the number of the histogram that you want to remove In this example click on 1 2 Inthe File Load Window click on the Remove Hist button to remove the select
102. 0 disk I O operations per second The bars in the bottom section of the System Summary screen measure up to 80 operations per second so a bar that is consistently full may indicate a disk bottleneck Disk traffic typically occurs in bursts When a disk is active it often shows 40 to 50 operations per second This is a problem only if a particular disk seems to be much more heavily used than the others If two heavily used file systems are on that one disk consider moving one of the file systems to another disk In cases where the files are not convenient to separate use virtual partitions to stripe the file system across multiple disks Analyzing Performance Monitor Data a 9 47 AUSPEX W A DolIneed write acceleration or more of it The Write Accelerator is optional on the NetServer The Write Accelerator improves the client NFS response time and the single user write performance for large files The SP statistics help you evaluate the need for write acceleration on an SP If the write I O activity for an SP appears high adding a Write Accelerator can improve the performance With a Write Accelerator installed on an SP you can enable the Write Acceleration function for the busy file systems with an option added to the entry in etc fstab The file systems experiencing a high number of writes receive help from the write cache reducing the overall disk write activity To determine which file systems might require write acceleration review
103. 000 A 112 meter tape in this drive blocking factor 126 has the same capacities as with the 8505 drive above 7 0 GB size 8141 density 141000 blocking factor 126 Quantum CompactTape IV 40 0 GB size 115000 1800 feet density 81633 DLT4000 blocking factor 126 20 0 GB size 115000 native density 81633 blocking factor 126 20 0 GB size 76800 compressed density 62500 blocking factor 126 10 0 GB size 76800 density 62500 blocking factor 126 1 4 inch 150 0 MB size 14400 600 feet density 1000 blocking factor 126 1 2 inch 150 0 MB size 2400 2400 feet density 6250 blocking factor 126 40 0 MB size 2400 density 1600 blocking factor 126 Note For the 4 mm and 8 mm tapes listed in Table 8 3 size arguments D compensate for the inter record gap used in the dump program Using size to compensate for this gap forces dump to use the full capacity of the tape For 1 4 and 1 2 inch tapes dump uses an inter record gap of 0 12 inch however there is no inter record gap for 4 mm and 8 mm tapes Without compensating for this gap dump uses only part of the tape Also the suggested value for size has made allowance for rewriting bad blocks 8 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX For tape lengths not listed in Table 8 2 you can estimate tape capacity based on the length of the tape For example assuming you use the same density and blocking factor the capacity of a 54 meter tape in an Exabyte 8200 drive is abou
104. 10 The last sample in the histogram By default the histogram displays all samples However you can for example specify that only the first 10 samples are displayed by setting Last X Label to 10 The increment on the X axis horizontal axis It controls the number of labels displayed on the X axis For example if you specify 10 the labels on the X axis are 10 seconds apart If the first label is 11 32 10 then the second label is 11 32 20 A higher increment causes fewer timestamps to be displayed The shape of the histogram is not affected by the X inc The default value of X inc is 0 which means that the X axis is automatically scaled The first value marked on the Y axis vertical axis By default the first value is 0 The last value marked on the Y axis By default it is the highest value attained by the statistics in the current histogram Displaying Performance Data in Histograms a 9 25 Table 9 3 Fields in the Histogram Properties Window Continued Field Function Y inc The increment on the Y axis It controls the number of labels displayed on the Y axis For example if you specify 2 the labels are two units apart A higher increment causes fewer labels to be displayed The shape of the histogram is not affected by Y inc The default value of Y inc is 0 which means that the Y axis is automatically scaled X Pos The spacing between the frame containing the histogram and the left edge of the window A greater X Po
105. 10 Tahoe File System 4 22 Tape addingadrive 4 25 blocking factor 8 7 capacity 8 5 to 8 7 cleaning drive heads 8 20 density 4 2 distribution medium 2 25 drive capacities 8 5 drive devicenames 4 2 drive errors 8 19 8 22 dump and restore arguments 8 7 I O error 8 19 jam error 8 21 LED indicator media error 8 20 memory error 8 20 no rewind option 4 2 8 3 on line error 8 22 read error 8 21 rewind option 4 2 write error 8 6 8 19 8 21 8 22 TCP IP 1 4 Temperature requirement for battery 7 5 Terminal console X 2 11 local 2 19 type 2 11 6 30 Terminology xvii Throughput NetServer 7 3 Time required for formatting a drive 4 39 Time zone 8 22 Index 14 A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX list C 1 setting 2 10 2 16 Toggle field 2 9 Tracking hardware upgrades 10 5 ttya 2 18 ttyb 2 18 10 12 tunefs command 4 22 Types of statistics displayed by ax_perfhist 9 40 Types of tapes for backup 8 3 DP checksumming 2 5 FS file system format 4 5 4 22 7 8 mount command 4 22 nattended backup scheme for 8 10 nintialized write cache status 7 6 NIX 4 2 BSD file system format 4 22 device driver 6 25 nmounting a CD ROM 4 5 nreadable disk blocks 6 25 nreadable disk label 6 24 nreadable superblock 6 24 nrecovered disk errors 6 9 6 10 nrecovered file system problems 6 9 nwritten data in the write cache 7 6 parrow key 2 10 Gee eed Gee 4 cae GC V Version SunOS 2 30 Virtual part
106. 10 8 A Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Dealing with Server Problems This section provides special maintenance information about error checking system crash procedures and message logging which can be useful if you encounter system problems This section also gives useful tips for resolving problems if they occur This section also describes how to obtain assistance from Auspex Rebuilding the System Kernel The procedure for rebuilding the Auspex NetServer kernel is identical to the equivalent Sun procedure The directory usr sys aushp conf holds binaries and object files needed to rebuild the kernel The default configuration file in the conf directory is named AUSPEX1 The following steps outline the procedure for rebuilding the kernel 1 As root change directories to usr sys aushp conf and copy the default configuration file cd usr sys aushp conf cp AUSPEX1 NEWCONFIG where NEWCONKHIG is the new configuration file 2 If necessary edit the new configuration file vi NEWCONFIG edit information wq Note When modifying the configuration file edit only sections of the file G that describe non Auspex devices Do not change any lines in the Auspex Devices section of the file 3 Complete the kernel rebuild config NEWCONFIG cd NEWCONFIG make mv vmunix vmunix old mv vmunix reboot This completes the procedure for rebuilding the kernel For more information re
107. 2 Oc Age Distributions a or A 1 minute old z or Z 26 minutes old Lower case letters AGE buffers Upper case LRU buffers data cache AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA meta cachet AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA inode cache 222222222222 2222222222222 22222222222 e Pee PPE eee EPP eee PP ee FF 2222 name cache 2222222222222Z2222222272222 22222 e eee eee eee eee eee eee eee eee eee Figure 9 5 Sample FP Statistics screen For information on subsequent FPs type f The screen contains the following information A F The first line of this area displays the CPU utilization of the current FP The second line is an operations per second bar in which letters represent kinds of operations The third line shows the distribution of the operations it indicates the percentage of total operations that each operation type represents The types of operations are listed b P activity elow the operations distribution bar Cache activity F All cache memory is on NP boards Four types of caches are displayed Data cache is the user data disk blocks in primary memory Meta cache is the nonuser data disk blocks in FP memory which store information about file systems Inode cache is the inodes cached in FP memory Name cache is the file name information cached in FP memory or each cache the following information is displayed Eject age which is
108. 3 2 single user option 3 6 Break key 10 12 Broadcast address 2 15 Buffered I O daemons BIODs 7 3 C c continue command 10 12 Cache age distribution of contents disk 1 6 information about four cache types 9 12 write 7 2 Capacity of tape drives 8 5 Capturing a screen in ax_perfhist 9 39 CD ROM addingadrive 4 25 booting from 3 6 4 5 6 27 device name 1 13 4 2 distribution medium 2 25 ejecting 4 6 how touse 4 5 in various commands 4 2 loading 4 5 location 4 26 mount point 2 25 mounting and unmounting file systems 4 5 numbering 4 2 CD ROM drives acdn 1 13 Changing from single user mode to multi user mode 6 4 the drive configuration 7 11 the partitioning onadisk 4 39 the root password 2 3 Checking an isolated file system 6 14 Checking the write cache state 7 9 Checksum for validating write cache control data 7 6 checksumming and NetServers 2 5 Cleaning tape drive heads 8 20 Clearing a histogram 9 22 Client diskless 2 20 effects of file system isolation on 6 13 information worksheet D 1 D 4 server 2 15 when to reboot 2 30 9 13 A Index 3 workstation 2 28 client UDP checksumming 2 5 Client files distributing among multiple disks 4 11 Cloning a virtual partition important guidelines 8 13 meaning of 8 12 procedure for 8 14 Commands arp 10 15 ax_add_device 1 13 4 27 5 11 6 4 6 5 ax_admmsg_ 1 15 ax_arp 1 12 10 14 ax_clonefs 1 14 4 27 4 32 4 33 8 2 8 14 8 15 ax_config 1 15 10 2 ax_diskconf 1
109. 3 9 3 10 3 11 etc remote 2 6 2 18 etc rpc 2 6 etc sendmail cf 2 6 etc services 2 6 etc shutdown 3 4 etc syslog conf 2 6 etc ttytab 2 6 2 17 2 19 etc vpartab 2 6 4 26 4 31 5 9 export 2 21 4 17 6 33 export exec 4 11 export exec sun4 sunos 4 1 2 21 home 2 30 6 33 tftpboot 2 28 usr 2 30 4 10 4 15 6 32 usr adm messages 10 10 usr etc 2 9 usr kvm 2 30 usr openwin 4 10 4 16 6 32 usr share 2 30 usr share lib zoneinfo 2 17 var 4 10 4 14 6 32 10 10 var adm config report 10 1 10 2 var adm messages message logging 10 11 messages related to file system isolation 6 15 var log auspex messages 10 11 var spool cron crontabs root 2 7 var yp 2 29 Numerics 1 35 GB disk 4 7 1 5 1 tapes encoding 8 18 1 7 tape encoding 8 18 1 76 GB disk 4 7 1 GB disk 4 7 2step load 9 24 A Index 1 2 GB disk 4 7 3 GB disk 4 7 4 2 file system type 1 14 4 13 4 GB disk 4 7 4 mm tape drives data format 8 5 naming conventions 4 3 8 mm tape drives naming conventions 4 3 9 GB disk 4 7 A acd device 1 13 4 5 Active file system 8 4 Active status virtual partition 5 12 Adding a file system back on line 6 14 Adding drives 4 25 7 4 Adding entries to etc vpartab 5 14 Address broadcast 2 15 Ethernet 2 30 Internet 2 11 2 14 2 30 X terminal 2 11 AGE buffer statistics 9 13 Aliases for host names 2 14 Alternate boot device 3 6 Alternate superblock 6 24 Analyzing performance monitor data 9 47
110. 4GB cyl 4094 alt hd 16 sec 128 gt ad4 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad5 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad6 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad7 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad8 lt Seagate 4GB cyl 4094 alt hd 16 sec 128 gt ad9 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad1l0 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt adll lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad12 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad13 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt adl4 lt Seagate 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad15 lt MICROP 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad1l6 lt MICROP 4GB cyl 4094 alt 1 hd 16 sec 128 gt ad17 lt MICROP 4GB cyl 4094 alt 1 hd 16 sec 128 gt O 71 12 0 device ad 0 0 0 mvic vme asp 6d 1180 ad 0 0 0 File and args 2 2 Standard Boot Messages 2 ad18 lt MICROP 4GB ad19 lt MICROP 4GB ad20 lt MICROP 4GB ad21 lt MICROP 4GB ASPX buf e unit 0 ASPX buf e unit 1 ASPX buf e unit 2 ASPX net acc unit ASPX bu ASPX bu ASPX bu f e unit 3 f e unit 4 f e unit 5 cyl cyl cyl cyl at iop0 SBu at iop0 SBu at iop0 SBu 4094 al 4094 al 4094 al 4094 al Pe PP nn nat tot ct 0O at iopO SBus at iopl SBu at iopl SBu at iopl SBu s sl s
111. 5 Virtual Partition Restrictions yt sii gare set he ate pao gu he de esate bee 5 7 Restrictions Common to All Kinds of Virtual Partitions 5 7 Restrictions Applicable to Concatenated and Striped Partitions 5 7 Restrictions Applicable to Mirrored Partitions 00000 5 8 Virtual Partition Driver File and Commands 0000 cece eens 5 9 Virtual Partition Driver sich a a ieee ea ee ee 5 9 Virttial Partition Pile rei ayn a e ee oie ee Ne eda 5 9 Virtual Partition Commands 0 6 ce eens 5 10 Automatic Error Recovery on Mirrored Partitions 0005 5 13 Defining and Reconfiguring Virtual Partitions 0 00 e eee 5 14 Defining a Virtual Partition usuusu eke ae ee een ee GS 5 14 Expanding a Concatenated or Striped Partition 005 5 15 Expanding a Concatenated Partition 0 0 0 0 0c eee eee eee 5 16 Expanding a Striped Parthonas wits isla Weekes Son rreren 5 16 Expanding a One Membered Mirrored Partition 4 5 18 Expanding a One Membered Concatenated Partition 5 19 Contents A vii Expanding a One Membered Striped Partition 5 19 Expanding a Two Membered Mirrored Partition 004 5 19 Changing a Striped or Concatenated Partition to a Mirrored Partition 5 19 Adding a New Member to an Existing Mirrored Partition 5 21 Removing a Member
112. 5 1 dainval Auspex Processor Error Messages E 9 Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1036 1 closedq 1037 3 free inode s d had d blocks 1038 1 alloccgblk cyl groups corrupted 1039 1 alloccgblk can t find blk in cyl 1040 1 ialloccg map corrupted 1041 1 ialloccg block not in map 1042 1 free bad size 1043 1 free freeing free block 1044 1 free freeing free frag 1045 1 ifree range 1046 1 ifree freeing free inode 1047 1 alloccg map corrupted 1048 1 alloccg block not in map 1049 1 direnter 1050 1 direnter target directory link count 1051 2 s bad dir ino d at offset d s 1052 1 dirprepareentry new block 1053 1 DIRBLKSIZ gt fsize 1054 1 dirprepareentry invalid slot status 1055 1 dirmakeinode no attributes 1056 1 direnter dquot 1057 1 DIRBLKSIZ gt fsize 1058 1 dirremove 1059 1 iget bad dev 1060 1 iget bad fs 1061 1 free inode isn t 1062 1 iput 1063 1 irele 1064 1 idrop 1065 1 iinactive 1066 1 itrunc newspace 1067 1 itrunc1 1068 1 itrunc2 1069 1 update ro fs mod 1070 1 isblock 1071 1 clrolock E 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1072 1 setblock 1073 1 getmp bad magic 1074 1 ufs_statfs 1075 1 ufs_badvfsop Shouldn t be called 1076 1 rwip 1077 1 rwip type 1078 1 ufs_readdir dir rec
113. 5 Servertacdelc Hist4 Fri Feb 3 14 48 52 1995 Serversacdele Ti 6 5 5 4 23 2 1 9 Fri 14 48 55 Fri 14 49 03 Fri 1434931 Fri 14 48 55 Fri 14 45 03 Fri 14 49 1 HP cpu usr_time Ave 7 0 SPO cpu busy Ave 0 9 SEPP NNAW Be coi onmhk ab oah Hist2 Fri Feb 3 14 48 52 1995 Servertacdelc HistS Fri Feb 3 14 48 52 1995 Serversacdele 18 SCORPNWEUO ON Dwoma NE Fri 14 49 03 Fri 14 49 41 Fri 14 48 55 Fri 14 49 03 Fri 1424934 re Ave 8 4 SP1 cpu busy Ave 1 2 Hist3 Fri Feb 3 14 48 52 1995 Servertacdelc Hist6 Fri Feb 3 14 48 52 1995 Serversacdelc 12 0 76 1 65 2 54 3 43 5 32 6 21 7 nah f 0 0 Fri 14 48 55 Fri 14 49 03 Fri 1434931 Fri 14 49 03 Fri 1434931 HP cpu idle_time Ave 84 6 e ope eph itu Aver 5 4 Figure 9 19 Displaying six histograms as specified in a configuration file Displaying Performance Data in Histograms A 9 31 AUSPEX W To remove a configuration file open the Configuration Window and hold down the right mouse button Click on the Delete Configuration File option When the system asks you to confirm click on the Delete button The configuration file is removed from your directory Examining Samples in Histograms You can use the zooming feature provided in ax_perfhist to take a close look at statistics that interest you Zooming means that only samples collected within a specified period are displayed
114. 6 22 On line backup 8 12 On line documentation printing B 7 setting the display environment B 5 Online documentation FrameViewer license B 2 overview B 2 Online documentation printing xxi Online help xx OpenWindows 3 0 4 16 Operating system installing from the CD ROM 6 29 SunOS 3 2 Operator errors in handling tape drives 8 19 Optional product documentation B 1 Optional products 100Base T Ethernet 1 8 ATM 1 8 FTP on the Network Processor 1 8 Options for tape drives no rewind 4 2 8 3 rewind 4 2 Overall Properties Window 9 28 P Page Down 2 10 Page Up 2 10 Panic messages 10 11 Panics 10 10 E 8 Parameter button 9 23 Parameters menu in ax_perfhist 9 41 Partition changing 4 39 default disk partitions 4 7 managing 5 14 raw disk 4 3 scheme 4 7 4 28 size 4 7 virtual 5 2 Partition menu label option 4 40 select option 4 39 Partition option in the format menu 4 39 Partition tables root drive 4 8 Partitioning information in sector 0 6 24 passwd command 2 3 Performance data capturing and displaying in real time 9 2 Performance Monitor analyzing data 9 47 commands entered interactively 9 5 FP statistics 9 3 how to run 9 3 A Index 11 NP statistics 9 3 purpose of 1 12 SP statistics 9 3 virtual partition statistics 9 3 write cache statistics 9 3 pkgadd 1M 1 7 Port baud rate 2 19 console 2 18 modem 2 18 serial 2 18 10 12 Power on diagnostics 3 3 messages 6 3 Primary host name for a network inte
115. 8 Print C mimand s tik enn a a E EA a ag ier A ATEA is 9 38 Print Command Output aessa nia eee ahi es 9 40 Types of Statistics Displayed by ax_perfhist 00 0 0 00s 9 40 PU enticed E at cba ER Te ane Ra es MCU SNC a 9 41 INE enue acre te arrestee ee ee We ease 9 41 BE ik oe Saaana a Oe SAD Cede a eee UE ae AOE nies 9 43 SPE Aisa aa 8 ery aoe PR Re ea en cath nem a aloe td aes 9 45 Starting ax_perfhist with Options and Arguments 005 9 46 Analyzing Performance Monitor Data ic tii ea ets eS oe ROG Ce Ee GS 9 47 Chapter 10 Special Maintenance Abou t This Chapter sid ses a ve dae ate apr rar in pet ea cede aa beds outdca ie 10 1 Configuration Management sco icin chest dm aniia ple rheae sh pa aaienk s 10 2 Starting aX CON FS ssaa re Con Maced Beet faa E ets OAE Sa AEE Aa aeai 10 4 Interactive Mode n prsaiencat Gem eaemlacacare de ated abe doh atealpauener dana 10 4 Displaying Configuration Data 0 0 6 eee eee eee eee 10 5 Non interactive Mode sic cise a tte cealyids Yoh ay ted ah gach apie nena era a rst hans 10 5 Displaying Configuration Information 6 6000 e eee eee 10 5 Copying Configuration Information to a File 0005 10 6 Flash PROM Download Utility suasusu nnana rarene rreren eree 10 7 ax Toad fl shzine Aan ree te a e a ey a a Ea 10 7 Dealing with Server Problems 4525 h0es4ene seein ime yea ee aise 10 9 Rebuilding the system Kermel 9325 144 iain o
116. A 10 aA Version 1 10 System Manager s Guide AUSPEX W W AUSPEX newfs 8 nfsd 8 nfsstat 8c panic 8s rc 8 rc auspex 8 rc boot 8 rc local 8 rc shutdown 8 reboot 8 reporter 8 reporter sh 8 restore 8 rrestore route 8c savecore 8 shutdown 8 stopnfsd 8 sundiag 8 swapon 8 ypserv 8 ypbind ypxfrd Create a new file system NFS daemon Network file system statistics Explains what happens when the system crashes Scripts for autoreboot shutdown and daemons Restart the kernel Prepare and mail site reports to Auspex Customer Service Incremental file system restore Manipulate the routing tables Save core dumps of the operating system Close down the system at a given time Auspex utility to shut down the NFS System diagnostics Specify additional device for paging and swapping NIS server and binder processes Maintenance Commands A A 11 W AUSPEX Online Documentation About This Appendix This appendix provides information on installing online documentation and setting up your workstation s X environment to view the online documents Online Documentation a B 1 Online Documentation Auspex offers online documentation to users who have an X display that runs the X Window System protocol MIT X 11 Release 4 or later or that runs Sun OpenWindows 2 0 or later The online documentation provides hypertext links embedded in the text which allow you to move easily
117. Auspex Systems Data Display Tool File 7 Properties v Print 7 Qty Hists 1 Dhe Configure the properties of histograms for Remove all histograms from example their size position within the the Data Display Window window and so on Figure 9 10 Buttons in the Data Display Window 9 20 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Loading Statistics From a File The graph displayed by ax_perfhist contains no statistics see Figure 9 9 until you load the data from a file saved by ax_perfmon This section describes the histograms that are displayed using the default window properties alternatively you can configure the window properties before loading the statistics Follow these steps to load the statistics 1 Click on the File button to display a pull down menu Select Data Files from the menu The File Load Window appears as shown in Figure 9 11 File Load Window co acdelco dat in dat proted dat splinter dat Data File O Sum Params Directory Figure 9 11 File Load Window 2 In the Directory field type the name of the directory containing the statistics files The If the File Extn field is empty clicking on the Scan Dir button displays all the filenames in the directory If you name your statistics files consistently ending the filenames with a period followed by a constant string such as dat ax_perfhist displays only the files with the speci
118. Data Display Window and File Load Window allows you to select a different type of statistics quickly Once you select an input file you can send the File Load Window to the back allowing only the pull down menu buttons to show as illustrated in Figure 9 12 When you activate any of these buttons to select a new parameter the File Load Window does not obscure the Data Display Window because it remains in the back and you can view the new statistics immediately Auspex Systems Data Display Tool ay nists O 2 Step Load O Sum Params Figure 9 12 Displaying the Data Display and File Load windows simultaneously Clearing a Histogram To clear the histogram click on the Remove Hist button in the lower right corner of the File Load Window This button clears the statistics from the Data Display Window only It does not delete the statistics file from your directory 9 22 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Displaying Multiple Histograms To display up to eight histograms simultaneously first specify the histogram number then specify the type of statistics in each histogram pecily yp 8 The following example shows how to display a second histogram in the Data Display Window 1 Use one of the following ways to display multiple histograms A Inthe File Load Window click on 2 in the Hist list a Inthe Data Display Window click on the
119. Displaying Performance Data in Histograms a 9 27 v Auspex Systems Data Display Tool File v Properties 7 Print 7 Qty Hists 1 Clear All Hist1 Fri Feb 3 14348352 1995 Servertacdelco File acdelco dat 18 54 18 04 17 54 17 04 16 54 16 04 15 54 255 Fri 14 48 58 Fri 14 49 00 Fri 14 49 03 Fri 14 49 05 Fri 14 49 08 Fri 1424924 HP cpu sys_time Ave 8 4 Figure 9 16 Server statistics plotted on a continuous graph Configuring the layout of the Data Display Window You can design the overall appearance of the histograms in the Data Display Window For example you can determine whether histograms are arranged in columns or rows The following procedure describes how to configure the overall properties of histograms 1 In the Data Display Window click on the Properties button to display a pull down menu Select All Histograms in the menu The Overall Properties Window opens see Figure 9 17 gy Overall Properties Window Display Average Display Value Hide Value Automatic Y Scale Preset Value Figure 9 17 Overall Properties Window 9 28 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX 2 The Overall Properties Window lists the possible values for each parameter Click on the appropriate value to select it If you want to undo a modification and restore the original parameter value click on the Reset button Refer to Table 9 3 for the descriptions of the parame
120. Domain name Export path Root path Swap path Executables path Kernel executables path Home path usr share files path Ethernet addr Notes Internet addr Domain name Export path Root path Swap path Executables path Kernel executables path Home path usr share files path Client Information Worksheet a D 5 Disk Drive Information Worksheet Drive Type Information Disk Drive no slot number Drive size 1 GB 1 35 GB 1 76 GB 2 GB 3 GB 4 GB 9 GB Drive Partitioning Drive partitioning scheme root default stripe other Partition Size Mount Point Major Directories a Notes D 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Virtual Partition Information Worksheet Virtual Partition Information VP No Members Size Type Mount Point Major Directories Notes Virtual Partition Information Worksheet a D 7 W AUSPEX Messages About This Appendix This appendix describes messages unique to the NetServer Auspex NetServer messages fall into the following categories A Standard boot messages A System startup error messages a Auspex processor error messages This appendix makes no attempt to document standard Solaris messages For information about Solaris messages refer to Ap
121. EX D Mounting File Systems on the FP and HP This section describes mounting file systems on the FP and HP It also describes LFS which is specific to Auspex servers the modifications that Auspex made to the etc fstab file format and the commands related to LFS File Systems Mounted on the File Processor The FPs which support the LFS file system handle file system requests from the NPs For UNIX to mount file systems managed by the FPs it must use a file system type specific to the Auspex NetServers called Local File System LFS This is the file system type specified in the etc fstab file for the mounted file system LFS is much like NFS but the requests go across the VME backplane instead of over the network Individual files in LFS file systems can exceed 2 GB in size when available over NFS Version 3 Such files can be as big as the file system itself Clients accessing files over 2 GB in size must also have large file capabilities File systems normally are mounted on the FP Also only file systems managed by the FP can take advantage of a NetServer feature called file system isolation as described in Chapter 6 Recovering From Disk or File System Failures This feature allows an LFS mounted file system to go offline if it is corrupted or if there is a disk problem without affecting the operation of the entire server Mount a file system on the HP only if the conditions described in the next section apply File Sys
122. FP Do When an Attempted I O Fails This section describes what happens when the FP encounters a file system problem When you try to delete or create files in an inconsistent file system or a file system on an unreachable disk the FP cannot complete the required I O operation It can however take the file system offline so you can do the following a Investigate the cause of the problem by examining the SP and FP error messages logged in var adm messages a Follow the appropriate procedure to repair the file system or disk and then bring the file system back online An isolated file system remains mounted but any process that tries to access it hangs Note After a file system is isolated a user s attempt to log in to a server hangs D if the login or any shell startup file contains references to the isolated file system For example if usr local is LFS mounted and usr local bin is included in the PATH of login you cannot log in as root once usr local is isolated To ensure that you can log in to a server as root after a file system is isolated do not include commands that try to access an LFS file system in the root s login file or any shell startup file such as cshrc and profile The ax_isolated daemon is started by etc rc local to monitor file system isolation The syntax of the ax_isolated command is as follows ax_isolated m user s n By default ax_isolated sends email to root if a mounted file system is
123. Fast File System 4 22 FFS Fast File System 4 22 Field input for configuration forms NSconfig form 2 14 NSinstall form 6 31 SetupClient form 2 30 SetupExec form 2 25 SetupTty form 2 19 toggle 2 9 File 6 10 damaged inode 8 17 device 4 4 restoring from 1 5 1 produced dump tapes 8 18 virtual partition 5 9 File Load Window 9 21 9 22 input file list menu options File Processor contraints 4 22 error messages 6 9 6 25 mounting file systems on 4 18 File system export 2 21 4 17 usr 4 15 usr openwin 4 16 var 4 14 active 8 4 adding back on line 6 14 backing up on line 5 2 disk formats 4 22 distributing across disk drives 5 2 5 4 9 47 inconsistency 6 3 6 13 isolation 2 31 6 10 isolation after reparable media errors 6 17 isolation caused by corruption 6 18 isolation caused by irreparable media errors 6 15 killing sleeping processes on 6 15 larger than a physical disk 5 2 listing using ax_diskconf 5 23 mapping toFPs 4 20 on an inaccessible drive 6 19 protecting from disk or media failure 5 2 quiescent 8 4 releasing 6 14 root 4 13 specifying the type 4 19 Tahoe 4 22 2 10 9 37 A Index 7 type4 2 1 14 4 13 type LFS 1 14 4 18 4 30 5 11 Filter file creating 9 16 used in ax_perfmon 9 16 to 9 17 First time NetServer configuration 2 3 Flushing an ARP table 10 14 Flushing unwritten data from the Write Accelerator 7 10 FMK Functional Multi processing Kernel 1 10 FMP Functional Multi processing advantages o
124. Fri Sat 1 16 0 5 9 5 9 16 32 5 0 5 9 5 33 48 9 5 0 5 9 49 64 5 9 5 0 5 65 80 9 5 9 5 0 Week 2 File System Sun Mon Tue Wed Thu Fri Sat 1 16 1 5 9 5 9 16 32 5 1 5 9 5 33 48 9 5 1 5 9 49 64 5 9 5 1 5 65 80 9 5 9 5 1 Figure 8 1 Backup scheme for 42 drives 8 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Backing Up the Root Disk Because NetServer software prevents the root and usr partition from becoming members of virtual partitions and because the NetServer requires swap space on the root disk the root disk cannot be mirrored or hot plugged Auspex recommends that you back up the a and g partitions of the root disk onto tape Additionally you can replicate the root partitions onto another disk drive for use as a backup boot source in case the root disk fails You can then quickly replace a damaged root disk by moving the backup drive into slot 0 Auspex recommends that you use the online backup procedure described in Online Backup on page 8 12 to copy a root disk Recommendation If you changed the root drive configuration use online backup to copy the drive only after you have used it to boot the system successfully Then use dump to back up daily incremental changes to files on the backup drive If you perform online backups for the entire drive every day without first verifying that the modified drive is bootable yo
125. It is recorded in etc dumpdates The name of the clone partition dev rvp11 is not recorded For more information on the dump command syntax refer to Using the Dump and Restore Commands on page 8 2 8 16 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Restoring a File With a Damaged Inode This section describes how to restore a file when you suspect data is corrupted and fsck shows errors in an inode The following is a sample fsck output fsck dev ad9a dev rad9a Last Mounted on export root Phase 1 Check Blocks and Sizes BAD DUP FILE I 26 OWNER root MODE 644 SIZE 404 MTIME Sep 18 16 13 1995 CLEAR yn n Before you replace the corrupted file with the backup copy determine the filename associated with the incorrect inode from the fsck output In the previous example the inode number is 26 The following procedure restores the file associated with inode 26 1 Use ncheck to determine which file is associated with the inode ncheck i 26 dev rad9a dev rad9a 26 cshre In this example the file cshrc is associated with inode 26 2 Locate the backup tape that contains cshrc and install it on the server Then invoke the restore command in interactive mode and verbose mode For example if the backup tape is in slot 2 enter the following restore imvf dev nrast2 The restore command displays files by name and inode 3 Select the file to be restor
126. O e0 packets sec NPO el packets sec Ave 123 1 Fri Feb 3 15 25 57 1995 Serversacdelco Filet acdelco dat Sum of statistics from the first two histograms That is the statistics show packets per second on NPO both interface 0 and interface 1 To create this combined histogram select this histogram load the parameter used in the first histogram click the Sum button and load the parameter used in the second histogram Figure 9 22 Summing two histograms 9 36 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Deleting an Input File To remove a Statistics file that you have saved with ax_perfmon follow these steps 1 Open the Load File window 2 Inthe input file list hold down the right mouse button to display a pull down menu Select Delete File in the menu A box appears asking whether you want to delete the file Click on Delete and the file is removed from your directory The pull down menu contains parameters other than Delete File These parameters are described in the next section Pull Down Menu Displayed in the Input File List In the File Load Window a pull down menu appears when you point to the input file list and hold down the right mouse button The menu offers a convenient way to sum statistics from different files load statistics from an input file delete a file from your directory and so on Table 9 6 explains the function of each op
127. QD W AUSPEX Note Because ax_mconvert must create a new virtual partition to be the member of the mirrored partition if all virtual partition numbers have already been used ax_mconvert generates this error message ax_mconvert all VP devices in use Adding a New Member to an Existing Mirrored Partition If a mirrored partition with a single member already exists you can add another member to the partition by entering the following command ax_mattach mirrored_partition new_member For example if you want to add vp4 as the second member to vp5 enter the following ax_mattach vp5 vp4 The new member is automatically recorded in etc vpartab If the mirrored partition in the example vp5 is mounted ax_mrestore starts to copy data from the existing member to the new member of the mirrored partition Now the mirrored partition consists of two members with the same data Removing a Member from an Existing Mirrored Partition If a two membered mirrored partition already exists and you want to remove a member from it enter the following command ax_mdetach mirrored_partition old_member For example if you want to remove vp4 from vp5 enter the following ax_mdetach vp5 vp4 The change is automatically recorded in etc vpartab partition edit etc vpartab to comment out references to the partition vp4 in the previous example i Note If you are planning to remove the drive containing the detached virtual Now the mirrore
128. Qty Hists button and select 2 2 To define the contents of histogram number 2 hold down the right mouse button while pointing to the Parameter button in the File Load Window Select a parameter from the pull down menu The statistics appear in the second histogram in the Data Display Window as shown in Figure 9 13 In this example the first histogram shows the CPU load of the FPO the second histogram shows the I O byte counts on NPO The ax_perfhist program automatically arranges and resizes the histograms to optimize the use of the Data Display Window For example if three histograms are displayed the window is divided into three equal portions If you delete two of the histograms the remaining histogram is resized to occupy the entire Data Display Window Auspex Systems Data Display Tool File 7 Properties 7 Print 7 ty Hists 2 Clear All Done Qty Histl Fri Feb 3 14 48 52 1995 Servertacdelco Filet acdelco dat Fri 14 48 58 Fri 14 49 00 Fri 14 49 03 Fri 14 49 05 Fri 14 49 08 Fri 1424924 FPO cpu busy Aver 2 9 Hist2 Fri Feb 3 14 48 52 1995 Servertacdelco Filet acdelco dat 0 0 a 1 Fri 14 48 55 Fri 14 48 58 Fri 14 49 00 Fri 14 49 03 Fri 14 49 05 Fri 14 49 08 Fri 1434931 NPO Ffddi0 Incoming Outgoing Bytes sec Ave 125632 9 Figure 9 13 Data Display Window showing two histograms simultaneously Displaying Performance Data in Histograms a 9 23 Loading Statistics Using 2 Step Load If you prefer
129. SP to an NP bypassing the HP FTP NP handles all the Internet Transmission Control Protocol processing FastBackup FastBackup increases the speed of Auspex systems doing full system backups and restores With FastBackup backup and restore operations take place entirely between SP boards and the backup devices Up to two backup devices can be added for each SP board The backup device can be a tape drive on the NetServer a jukebox or tape stacker allowing the system administrator to start the process and then let the process run without further intervention 1 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Hardware Implementation of the FMP Architecture Figure 1 1 illustrates the NetServer hardware architecture Remember that because you can mix and match some of the processor boards in one system the configuration of your server may be different from the one shown Also there are restrictions on the maximum number of boards that can be installed in a system depending on the board combination SunOS Host Processor 128 384 MB UNIX Memory Tape or a Optical cpul scsi Disk UNIX Library FMP I O Backplane Network Network Storage Write Processor Processor Processor Accelerator 128 256 MB 128 256 MB ptiegal I O Cache Memory I O Cache Memory CPU CPU CPU CPU CPU Pr
130. SPEX W W AUSPEX System Calls intro 2 Introduction to system services and error numbers chown 2 Change the user and group ownership of files mount 2v Mount file systems nfssvc 2 NFS daemons System Calls a A 3 Devices and Networking acd 4 ad 4 ae 4 afddi 4 afe 4 anp 4 apm 4 ard 4 asp 4 ast 4 atm_accept 4 atm_bind 4 atm_client_aal4 4 atm_client_null 4 atm_close 4 atm_connect 4 atm_error 4 atm_getbatchsize 4 atm_gethostbyname 4 atm_intro 4 atm_listen 4 atm_open 4 atm_recv 4 atm_send 4 atm_server_aal4 4 atm_server_null 4 atm_setbatchsize 4 atmio 4 cdromio 4 compver 4 depend 4 if 4n CD ROM driver for the Auspex Storage Processor Disk driver for Auspex Storage Processor Auspex Ethernet interface Auspex FDDI MLIT 3 interface Auspex Fast Ethernet interface Auspex Network Processor Auspex Primary Memory Auspex RAID array driver Auspex Storage Processor Tape driver for Auspex Storage Processor Accepts a pending connection request Binds an Application Service Access Point ASAP to an ATM end point Example client ATM program using AAL 3 4 Example client ATM program using null AAL Closes a connection with an ATM endpoint Establishes a connection with an ATM endpoint Prints an ATM error message Gets batch size for null AAL interface Gets the ATM address for a hostname Provides introduction to ATM user level library functions Listens for an ATM co
131. This section provides an example for recovering a disk hardware error through the ax_write_cache command on SP0 1 Display the write cache error pages to verify the hardware error ax_write_cache s 0 errors Run ax_kill 8 to kill all processes sleeping on file systems that may have become isolated because of disk failure Unmount the affected file system Run ax_hot_plug to remove and reseat the drive that failed Reseating recycles power to the drive which sometimes corrects drive problems 7 10 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX 10 11 12 Check the drive LED to see if the drive is working If the LED remains on or flashes continuously go to step 6 If the LED flashes briefly and then goes out reseating the drive may have corrected the error If so enter the following command ax_write cache flush s 0 If the flush succeeds reseating the drive corrected the problem and the recovery procedure is complete If the state of the cache is still DIRTY go to step 6 Run ax_hot_plug to replace the problem drive After replacing the drive purge the write cache by entering the following command ax_write_cache s 0 purge Verify the cache state by entering the following command ax_write_cache s 0 state If the state is OFF the purge was successful If the state is not OFF contact Auspex Technical Support for assistance Enable the write cache by entering the following command
132. Version 1 10 System Manager s Guide Y AUSPEX Copyright Copyright 1998 Auspex Systems Inc All rights reserved Printed in the United States of America Part Number 850517 001 No part of this publication may be reproduced in any form or by any means without the prior written consent of Auspex Systems Inc Auspex Systems Inc reserves the right to revise this publication and make changes in content from time to time without obligation on the part of Auspex Systems to provide prior notification of such revision or change U S GOVERNMENT RIGHTS As specified in 48 C F R 12 212 of the FAR and in 48 C ER 227 7202 1 of the DFARS the use duplication or disclosure of licensed commercial software and documentation is subject to the Auspex System s license Such rights and restrictions are similar to those set forth in FAR 52 227 19 c 1 amp c 2 Trademarks Auspex Auspex logo design Functional Multiprocessor Functional Multi processor Functional Multi processing Functional Multiprocessing Kernel FMK and FMP are registered trademarks of Auspex Systems Inc NS 7000 NS 6000 NS 6002 NS 5500 NS 5502 NS 5000 NS 3000 NetServer DataGuard ServerGuard Functional Multiprocessing NeTservices and Thrive Carefully are trademarks of Auspex Systems Inc AT amp T is a registered trademark of AT amp T Corporation Microsoft MS MS DOS Windows Windows NT and Backoffice are either registered trademarks or trademarks of Mi
133. a File systems managed by the File Processor must be mounted type LFS file systems managed by the HP must be mounted type 4 2 LFS is discussed in Mounting File Systems on the FP and HP on page 4 18 Auspex File System Structure Figure 4 4 through Figure 4 9 illustrate the Auspex file system structure The following file systems are shown a root a var a usr a usr openwin a export Your export file system structure may be different from the one described in this section depending on your version of the SunOS executables a exportN The exportN file system is not a default file system shipped with the server To create it run SetupExec and SetupClient as described in Chapter 2 In the figures an arrow gt indicates a symbolic link to the directory following the arrow For a more detailed description of the functions of the files mentioned in Figure 4 4 through Figure 4 9 see Sun s System and Network Administration Note that Auspex modifies the SunOS file structure to some extent however from the client s perspective the modifications are invisible 4 12 a Version 1 10 System Manager s Guide W AUSPEX This section mentions two file system types 4 2 and LFS which appear in etc fstab They are discussed in greater detail in Mounting File Systems on the FP and HP on page 4 18 Root file system Figure 4 4 shows the server s root file system The standard SunOS configuratio
134. a 6 5 Run fsck to check the newly created file system Restore data to the new drive from the most recent backup tapes Run fsck p to check all the file systems and then press Ctrl D to go to multiuser mode Entering Multiuser Mode and Replacing the Drive If you want the server to be in multiuser mode as soon as possible follow this procedure to replace the drive in multiuser mode 10 Caution The following line appears at the top of the etc vpartab and etc fstab files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades In single user mode comment out the entries in etc fstab and etc vpartab that reference the drive to be replaced If you change etc vpartab run ax_loadvpar to load the revised virtual partition table Run fsck p to check all the file systems and then press Ctrl D to go to multiuser mode After the system boots to multiuser mode use ax_hot_plug to replace the drive If the drive is not recognized by the SP use ax_hot_plug with the add option At the prompt remove the problem drive and insert the new one Caution Do not remove a disk from a slot while the disk is spinning down Removing a disk drive while it is still spinning can ca
135. a Inthe example shown in Table 4 8 the export file system is in ad7c This file system contains the architecture dependent executables that correspond to a diskless client s architecture allowing clients to boot their architecture specific operating system When you execute SetupExec the directories are loaded into the export directory The export root and export swap file systems contain root and swap directories for several diskless clients When you execute the SetupClient command it places the client root and swap directories in the partitions you specified in the SetupClient form In Table 4 8 export root and export swap reside on virtual partitions that consist of physical partitions on different drives Using virtual partitions allows you to back up these file systems easily and balance disk access among multiple disks For more information on the Auspex Virtual Partition Manager refer to Chapter 5 However you can choose to store the clients root and swap directories on physical partitions if you prefer to manage smaller file systems because newfs and fsck on these systems take a shorter time to complete than on a large virtual partition you want to be able to add client diskless workstations without having to reconfigure a virtual partition By default the SetupClient program sets up root and swap in exportN where N is the disk number you specify For example you can store one group of root and swap dir
136. abel 8 or format 8 to relabel the data drives and initialize file systems on the appropriate partitions using newfs 8 For more information on these commands refer to the ax_label man page format man page and newfs man page make sure the partition does not overlap another partition that is already mounted or opened For example if dev ad10h is mounted trying to use newfs to initialize dev rad10c generates a message similar to the following fy Note Before using newfs to initialize a file system on a physical partition dev radlc cannot create Device busy Naming File Systems Because any server disk partitioning scheme includes multiple disks and partitions it is important to develop a logical naming convention to identify the file systems Recommendation Do not base file system names on physical device names per Using names other than physical device names makes it easier to reconfigure disk partitions when you use virtual partitions For example name the file system used by the engineering department home engineering instead of home ad9 even though ad9 is the disk on which the file system resides 4 10 a Version 1 10 System Manager s Guide W AUSPEX Example of File System Distribution The following list explains the file systems that typically exist on a server Table 4 8 illustrates how these file systems are distributed a The file systems ad0a ad0b ad0g ad0d and adOf are reserved for the server
137. access an isolated file system they receive no responses from the server To them the server seems to have hung but in fact it continues to deliver file service for other file systems NFS I O operations continue on other file systems that are not isolated Repairing an Isolated File System with fsck If you have determined that an LFS error not a faulty disk caused the isolation you can repair the file system by running fsck interactively while the file system remains mounted While the NetServer is running fsck keeps track of dirty cylinder groups in each file system If a file system contains dirty cylinder groups fsck checks only those cylinder groups at reboot and the file system name is entered in etc cgfscklog File systems listed in etc cgfscklog are checked later on by ax_isolated f which is normally invoked from a cron script If the Write Accelerator contains unwritten data for the isolated file system that is write cache state is DIRTY the fsck command may fail and an error message appears The following is an example of the error message ad4c fatal error 10 Accessing Dirty Write Cache on read drive 0 block 64 absolute block 64 Unrecoverable Disk or File System Problems in Multiuser Mode a 6 13 To flush the Write Accelerator use the ax_write_cache command For example if the Write Accelerator is installed on SP0 enter the following ax_write_cache s 0 flush After the state of the Write Accelerator
138. ad write head To confirm use the ax_sputil verify command to see if the damaged sector can be verified For example ax_sputil verify 36 3884625 If the damaged sector can be verified see File System Isolation After Media Errors That Can Be Fixed on page 6 17 If the ax_sputil verify command hangs or returns error messages the sector cannot be verified In this case you cannot map out the problem sector Instead replace the disk and restore the file system using the following steps 3 Unexport the file system so clients cannot start processes that try to access the file system For example exportfs u export root Diskless clients trying to access the file system receive a stale file handle error message after the file system has been unexported As a result the clients crash This is acceptable in this example however because clients cannot access the bad disk anyway In this example unexporting export root affects only diskless clients 4 Enter ax_kill to kill all processes that are sleeping on the isolated file system 5 If the SP controlling the isolated file system uses write acceleration flush all unwritten data in the write cache to all disks and then purge the data that can not be written For example ax_write_cache s 0 flush ax_write_cache s 0 purge The write accelerator is automatically turned off after these steps Enable it again by entering the following command so other file systems on SPO can cont
139. address Ethernet address Type of NIS service Domain name Swap size FileServer partition Path to Client s root Path to Client s swap file Path to executables Path to kernel executables Path to usr share files Path to home Indicates whether to add or remove a client Host name of the client machine Client s architecture for instance sun4c sun4 sun3x or sun3 See Table 2 7 The SunOS version running on the client for example 4 1 3 or 4 1 4 Internet address of the client Ethernet address of the client Type of NIS service to provide to the client if any It can be either client or none A diskless SunOS client cannot be an NIS master NIS Domain Name This field appears only if the client is running NIS service Number of bytes reserved for the client s swap file SetupClient creates SWAP_PATH CLIENTNAME with this size You can append one of K or k to indicate kilobytes or M or m to indicate megabytes For example 16M or 16000K Path name of the export partition in which various client directories and files reside Path name of the client root directory This is the directory the client mounts as You must enter the client s host name by itself Path name of this client s swap file You must enter the client s host name by itself Path name of the directory in which the architecture dependent executables reside This is the directory that the client mounts as usr Path name o
140. aged status virtual partition 5 12 Data cache 9 12 Data cache buffers for the FP 9 8 Data compression 4 3 8 6 Data Display Window 9 20 9 22 configuring layout 9 28 options 9 35 two histograms 9 23 DataGuard 1 3 Date and time formats 2 17 Date and time on the server 2 16 dd command 4 38 6 26 Default boot device 3 5 Default disk partitions 4 10 Defining a virtual partition 5 14 Delete key 2 10 Detaching a member from a mirrored partition 5 11 Device boot 3 6 creating new 4 4 defining with MAKEDEV auspex 1 13 names 4 2 4 5 special files 4 4 Device driver UNIX 6 25 vp 1 13 5 9 Device names acdn 1 13 adn 1 13 df command 4 22 5 3 Diagnostic switch on the Host Processor 10 12 10 13 Diagnostics extended 3 3 power on 3 3 Dial in modem 2 19 Dial out modem 2 19 Differences between mirroring and cloning a virtual partition 8 12 Dirty status virtual partition 5 12 write cache status 7 6 Disabling the Write Accelerator 7 9 Disk 1 35 GB 4 7 1 76 GB 4 7 1 GB 4 7 2 GB 4 7 3 GB 4 7 4 GB 4 7 A Index 5 9 GB 4 7 adding 4 25 blocks unreadable 6 25 cache 1 6 configuration tool 5 23 copying data 6 22 6 23 copying the label 6 23 damaged root 6 26 defect 4 38 devicename 4 2 displaying configuration information 5 23 drive information worksheet D 1 D 6 drive serial number 7 4 dynamic table format 4 22 formats 4 22 formatting 4 38 label 4 7 4 28 5 7 5 8 6 24 numbering 4 2 partition schemes 4 7 partitioning 4 12 r
141. agnostics Utility to download Flash firmware Auspex utility to load the virtual partition table Show a disk s label and partitioning Auspex utility to attach a new member to a mirrored virtual partition Convert a virtual partition to a mirrored virtual partition Auspex utility to detach a member from a mirrored virtual partition Prints the amount and type of primary memory VME data transfer test Auspex utility to copy data from one member of a mirrored virtual partition to the other member Utility to obtain network statistics from Auspex Ethernet Processors Start NFS daemons on Auspex Network Processors Auspex command to display graphically the statistics gathered by ax_perfmon Auspex performance monitoring tool A 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX ax_ping 8 ax_purgelock 8 ax_remove_device 8 ax_request_key 8 ax_resolve 8 ax_restore_mirrors 8 ax_set_timeout 8 ax_sperr ax_sputil 8 ax_startup 8 ax_statd 8 and ax_statd2 8 ax_statfp 8 ax_statm16 8 ax_systest 8 ax_systest setup 8 ax_tcbt ax_tcbt2 ax_tcp 8 ax_tfddi ax_timed 8 and ax_errd 8 ax_trace ax_util 8 ax_vold 8 ax_vpstat 8 ax_whatrel 8 Send null FMK messages to an Auspex processor Releases remote locks held by the server for a specified Internet address Utility to allow disk drives to be removed while the NetServer is running Generate email or a file requesting
142. alid data received by the FDDI interface InAbort Number of frames received by the FDDI interface that are incomplete and cannot be processed InDrop Number of frames received by the FDDI interface that are discarded because the receive buffer was full MTU Maximum transmission unit size used by the server It defaults to 4352 Offnet MTU Maximum transmission unit size used by the server for FDDI traffic sent to networks not directly attached to the server It defaults to 4352 NP Protocol Screen The NP Protocol Statistics screen displays network statistics by protocol Figure 9 4 shows a sample NP Protocol Statistics screen OverU OverT dgram dgram cksum 1 0 NP Protocol Stats Fri Feb 2 10 00 45 1996 for NPO version2 versions DP cP eins S out errs forwarded put bytes get bytes workers segs in segs out segs w err curr estab conns active opens Passive opens workers te te ee oe oe oe oe Scocco m 149891 1 Figure 9 4 NP Protocol Statistics screen Each protocol has its own section Note that the NFS section also shows statistics by version2 and version3 Table 9 2 explains the network protocol statistics 9 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table 9 2 NP Protocol Statistics Statistics Explanation NFS statistics calls OverUDP OverTCP UDP statistics dgrams in dgrams out cksum errs IP statisti
143. an Auspex optional product key Find the FMK process with the specified name Restores all mirrored partitions to an active state Specify how long the system waits before declaring a board dead Storage Processor error description program Auspex utility for various disk related functions Auspex daemon that downloads and boots Auspex processors Auspex statistics daemons Utility to report statistics for a specified Auspex File Processor Utility to report M16 statistics for a specified Auspex Processor Auspex system stress test suite Generate an Auspex system test suite environment file template Test 100Base T Fast Ethernet port on NP boards Test Sun Fast Ethernet Adapter 2 0 on NP boards Add NP TCP workers show status of existing workers drop connections Test FDDI or CDDI port on NP boards Daemon that reports and logs errors from the Auspex processors Extracts formets or configures Auspex processor traces Auspex utility used to perform board level debugging on the Auspex processors Auspex virtual partition daemon Auspex utility to display the status of virtual partitions Utility to print the release number of files that are part of Auspex releases Maintenance Commands a A 9 ax_write_cache 8 boot 8S config 8 corescreen 8 corescreen sh 8 diskusg 8 dkinfo 8 dmesg 8 dump 8 dumpfs 8 eeprom 8s exportfs 8 fastboot 8 and fasthalt 8 fastfs 8 fsck 8 getcores 8 getcores s
144. an boot from the NetServer Software Configuration a 2 1 New Installations The following procedure outlines the steps for configuring your NetServer after you install the server for the first time with NSinstall 8 or NSupdate 8 See the Version 1 10 Software Release Note for software installation instructions The following sections provide details about each configuration step Install your console terminal Boot the server to single user mode Run NSconfig and reboot the server Change the root password If appropriate configure the server as an NIS master or slave in multiuser mode Modify site specific files as necessary and reboot the server if needed Run SetupTty Run SetupExec oC Oe ND TF ON BP Run SetupClient Installing the Console Terminal With each NetServer Auspex provides a DEC VT510 ANSI compatible ASCII terminal as a system console You can also use another model ANSI compatible terminal such as a Link terminal as well as an X terminal or UNIX workstation Using an X terminal or workstation allows you to view several windows simultaneously and invoke the online NetServer documentation The exact procedure for setting up the console depends on the terminal type the following procedure provides some general guidelines for setting up a console terminal 1 Start with the NetServer powered off 2 Set up the console terminal hardware and attach the terminal to the server s console port Refer to
145. an isolated file system varies depending on the cause of the problem See the examples in Examples of Recovery Procedures After File System Isolation on page 6 15 You must first determine whether a disk problem or an LFS problem has occurred If it is a disk problem decide whether to fix or replace the disk Try to salvage the disk because replacing a disk and restoring data from a backup tape takes more time The following example shows a file system that is isolated because of an LFS error In this case the file system can be repaired by fsck before it becomes available to network users again In the mnt file system the metadata pertaining to the inode is corrupted If you try to use rm to delete a file within mnt the command hangs and the following messages appear on the console May 19 15 34 42 Hostl FPO dev 0x0 ino 8 fs mnt May 19 15 34 42 Hostl FPO isolating filesystem mounted on mnt May 19 15 34 42 Hostl FPO 1046 ifree freeing free inode In this example because of a file system inconsistency the FP tries to free an inode that was already marked free To remedy the situation the FP takes mnt offline which causes all activity on the file system to stop and all LFS requests to the file system to return an error message This error message is transparent to user programs but is handled by the HP or NPs Other file systems on the server however continue to function as usual If network users try to
146. anager s Guide AUSPEX W W AUSPEX As the system boots the service representative can put the system in echo mode so that all input and output is displayed both on the local and remote consoles During the session the service representative may use the echo mode messaging facility to send instructions or information to your local console Follow all instructions from the service representative carefully 5 When the remote diagnostic session is completed and the NetServer is ready for normal operation follow these steps to restart the system Power off the modem b Disable diagnostic mode by setting the diagnostic switch on the HP to norm c At the monitor prompt boot the server to multiuser mode by using the b command Dealing with Server Problems a _ 10 13 Managing Network Interfaces This section gives information about commands for network management and then gives a tip for network maintenance The ax_perfmon command described in Chapter 9 provides an overall picture of server performance and of statistics on each network interface Occasionally however you might need to get detailed information on a server interface such as high level protocol statistics to investigate network problems Auspex provides three commands for you to manage and monitor individual network interfaces on your NetServer a ax_arp 8C displays and flushes the ARP table on a network interface a ax_netstat 8C shows network statistics for a
147. anded to include the following information a A list of the file systems on each drive a For each file system mounted on a virtual partition the report includes the identifier of the virtual partition and of its members if any Figure 5 10 shows an edited output of the ax_diskconf command Figure 5 11 shows edited output of the second ax_diskconf table with the p option Displaying Disk Configuration Information a 5 23 File System Mirror fusr fyar fopeny fusr openwin fyvar crash fhome fhomedocs fhome atria home sapphire fhome atticd fhome dy1p1 Physical Partitions ada ad0g adof ad8c ad0d ade ad0h ad6ctad c ad30c ad40c ad12c ad29c ad4 c ad54c ad55c fhome dy1p2 ad4 c ad54c ad55c disks usr local Capacity in 1000 megabytes 0 a 2 3 la qgdeeefFFFFFFFAhhh Pee eee eee ee eee eee E E E E lececc leececececeeef FFFFFFFFFFE l cceccecececeeeeeeeceeeeeececeeceececeeceeeeceeeececeececceceececececeece leccecec reidi 32 lines 1422 characters Figure 5 10 Example ax_diskconf display The following list explains some information in the ax_diskconf display in Figure 5 10 a The file system home docs uses vp11 as a concatenated virtual partition made up of ad6c and ad7c a The file system home atria uses a striped virtual partition made up of two members ad30c and ad40c a The file system disks usr local uses a mirrored virtual partition vp4 whose members ar
148. ape drives respectively These tables use the tape drive in the fifth drive slot as an example and assume that the drive uses the rewind option Always use rast or nrast for Exabyte 8200 tape drives because these drives do not use data compression 4 2 a Version 1 10 System Manager s Guide W AUSPEX Table 4 2 Naming a 4 mm tape drive in slot 5 Tape drive name Meaning rast5 Default tape density which is set by a hardware switch on the drive If the switch is set to rast5c rast5lo rast5loc ON the drive uses compression If the switch is set to OFF the tape accepts data in either low density or compression format depending on the format of the existing data on the tape If the tape is blank data is written in low density format Forced compression WangDAT compression overriding the hardware switch on the drive Without compression overriding the hardware switch on the drive Not applicable Table 4 3 Naming an 8 mm tape drive in slot 5 Tape drive Meaning Meaning name Exabyte 8500 drives Exabyte 8505 and 8505XL drives rast5 Exabyte 8500 format without compression Exabyte 8500 format without compression rast5c Not applicable Exabyte 8500 format with compression rast5lo Exabyte 8200 format without compression Exabyte 8200 format without compression rast5loc Not applicable Exabyte 8200 format with compression Table 4 4 Naming a DLT4000 tape drive in slot 7 Tape drive name Meaning ra
149. arger than a single physical disk drive or assemble several small unused partitions into a single partition large enough to be useful For example in Figure 5 1 the physical partitions ad7c ad8c and ad9a are members of a concatenated partition vp1 The physical partitions that make up a concatenated partition do not need to be the same size Concatenated partition vp1 y VUON Ooo TE J ad7c ad8c adga Physical partitions members of vp1 Figure 5 1 Example of a concatenated partition of three physical partitions Striped Partition A striped partition consists of 2 to 16 physical disk partitions The striped partition maps its logical address space across multiple physical partitions in fixed length segments referred to as stripes For example in Figure 5 2 a striped partition vp2 consists of four physical partitions of equal size Striping distributes access requests on heavily used file systems over multiple drives Unlike concatenated partitions adjacent blocks in a striped partition are mapped to blocks on different disks thus reducing access latency and improving throughput to a given file system 5 4 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Striped partition vp2 ad12g Physical partitions members of vp2 A stripe whose length must be a multiple of 8 KB Figure 5 2 Example of a striped partition of four members Mirrored Partition A mirrored partition consi
150. arguments for the boot command The boot command can be abbreviated to b Note The NetServer does not support booting to multiuser mode from the i CD ROM 3 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The Default Boot Device The default boot device is disk unit 0 ad0 if no device is found at this location the boot attempt fails The default boot device value is recorded in the system EEPROM and can be changed by the system administrator for instructions on how to do this refer to the eeprom man page You can also specify the device from which the server boots in the b command The next section discusses the various arguments for the boot command Using the Boot Command The following is the boot command syntax HP gt b dev sp unit part path options The b command takes the following arguments dev Device type which is ad for both Auspex disk or CD ROM drives The NetServer cannot boot from a tape drive Note The device type ad applies to both disk and CD ROM drives only when D booting the NetServer For other commands the CD ROM device type is acd Refer to CD ROM Drives on page 4 2 for more information on CD ROM drive naming conventions sp SP number which can be 0 1 2 3 or 4 for the NS 7000 700 and NS 7000 800 Series NetServers the default is 0 and 0 for the NS 7000 250 and NS 7000 150 unit A disk number relative to the SP number The supported range is 0 to 9 If you ar
151. arious commands are available for you to manage the Write Accelerator 7 2 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX How the Write Accelerator Improves Throughput The following list explains how a NetServer equipped with a Write Accelerator achieves higher throughput A Faster response for an NFS write operation Without write acceleration a client NFS write operation is completed only after the data is successfully written to the disk drive on the server With write acceleration the NetServer puts the data into the cache memory on the Write Accelerator board first and immediately informs the NFS client that the write operation is complete The client sees the faster response for NFS write operations More buffered I O daemons BIODs available for concurrent NFS writes A typical NFS client starts a number of BIODs in order to start several concurrent NFS write operations For example if an NFS client uses four BIODs to complete four conventional NFS writes each BIOD must wait until its write to disk is completed before the BIODs are available for the client to use again With the faster response time provided by the write cache the BIODs do not have to wait for the write to disk to be completed For the NFS client the net result is more available BIODs which translates to improved client throughput The actual performance benefits for the client depend on what applications the client is running Reduct
152. at which is the data file used by the format command Some of the drives listed are no longer available from Auspex but are still supported with this system software release Recommendation Although etc format dat includes a root partition table for per 9 GB disks such disks should not be set up as the root disk The 9 GB disk is not supported as a root drive Disk Partitioning a 4 7 Table 4 5 Default disk drive partition tables AUSPEX D 9 GB 4 GB 3 GB 2 GB 1 76 GB 1 35 GB 1 GB Partition drive drive drive drive drive drive drive a 1 445 MB 683 MB 480 MB 334 MB 281 MB 215 MB 159 MB c 8 668 MB 4 094 MB 2 882 MB 2 000MB 1 682MB 1 291 MB 954 MB d 4 335 MB 2 048 MB 1 441 MB 1 001 MB 842 MB 646 MB 477 MB g 2 890 MB 1 365 MB 961 MB 667 MB 561 MB 430 MB 318 MB h 4 333 MB 2 046 MB 1 440 MB 999 MB 840 MB 645 MB 477 MB Auspex NetServers use three different types of 2 GB disk drives Your 2 GB partition tables may differ from those shown here Table 4 6 Stripe drive partition tables 9 GB 4 GB 3 GB 2 GB 1 76 GB 1 35 GB 1 GB Partition drive drive drive drive drive drive drive a 2 167 MB 1 023 MB 720 MB 500 MB 420 MB 322 MB 238 MB b 2 167MB 1 023 MB 720 MB 500 MB 420 MB 332 MB 238 MB c 8 668 MB 4 094MB 2 882MB 2 000MB 1 682MB 1 291 MB 954 MB e 4 334 MB 2 046 MB 1 441 MB 1 000 MB 840 MB 645 MB 477 MB f 4 334 MB 2 046 MB 1 441 MB 1 000 MB 840 MB 645 MB 477 MB g 2 167 MB 1 023 MB 720 MB 500 MB 420 MB 322 MB 238 MB h 2 167 MB
153. ata Wea eae a eho aa a ates Ai iat aay aD 1 7 ServerGuard ite ve atiswas oh wie Saag ne PUR SAA SE a eee 1 7 Data Guard if sehen a ata tier asennad Anette east etl Mae ee ae eae 1 7 DriveGuat dips iev tats aa ced ak a ee eal ae 1 7 ING TSGRVICES meieni sare etude teed ate aint Wie a arias ed aut cathe RE oes 1 7 EtherChannel iein en aa a i ae A E dale aed E ee 1 8 rN EE A EEE A chint actiasal gt ode titel tee 1 8 FTP on the Network Processor 000s eee cence tenet eens 1 8 RAS EES AG RUA a sana af cotati sg ate Seen al een deat Sata ac tala pig ts ead ec tahoe 1 8 Hardware Implementation of the FMP Architecture 0 000 1 9 NetServer Software Architecture 0 0 0 c cece cee eee eee eee 1 10 Advantages of the Auspex Environment 0 000 c cee e eee eee 1 11 SunOS Enhancements 0 e ee eee eee eee eee nent E 1 12 Chapter 2 Software Configuration About This Chaplet sn sicasG ee namie t esis ai IO bre A 2 1 New Installations sis niia E Rig be tet cheney Gata 2 2 Installing the Console Terminal ia ss ange eae Rianne eks 2 2 Booting the Neterver sss 165 sit ca seid h Ane rerne rererere 2 2 Run NSconfig to Customize Your Server for the First Time 2 3 Changing the Root Password ax ca i 4s ih Fenda ews Be GS oe wie RSS 2 3 Configuring the Server to Be an NIS Master or Slave Server 2 4 Configuring the NetServer to Be an NIS Master 2 4 Conf
154. ation on the relationship between device names and data compression dev rastn The rewind on close option rewinds the tape after a file is written to the tape dev rastnlo This option is not used when dumping more than one file dev rastnc The n in the device name is the slot number in the range of available slots dev rastnloc Append lo c or loc to the device name to indicate whether compression is used Refer to Table 4 2 and Table 4 3 on 85 0515 001 for more information on the relationship between device names and data compression file_system_to_dump Name of the file system to be backed up You can specify the file system name or the physical partition name on which the file system is mounted For more information for the b d and s options for the dump command refer to The Size Density and Blocking Factor Options for Dump on page 8 7 Caution If you use 4 mm or 8 mm tapes to back up file systems use only digital quality tapes to prevent data loss and avoid tape drive damage Restore Command If you want to restore dump files back onto disk use the restore 8 command You can restore a dump file in noninteractive mode or you can use interactive mode to select specific files on the tape When using restore to replace dump files onto disk be aware of the following 4 Before restoring the files first mount the target partition 4 Files are restored in a relative path from the point of execution a restore overwr
155. attr fhandle_to_mbuf x_handy lt FHSIZE 1180 3 ard d Automatic rebuild has started 1181 3 ard d Automatic rebuild has terminated due to error 2x 1182 3 ard d ard number has been reassigned due to conflict 1187 3 Warning Battery voltage low on SP NVRAM write cache disabled d pages unwritten 1188 3 Warning NVRAM checksum error write cache disabled 1189 3 Warning a cached write failed on slot d 1190 3 A mirrored partition has a dirty or damaged component vp d 1191 2 Recovered unknown sense key and code slot d drive d sense key 2x sense code 2x VME addr 8x sector d 1192 2 VME Transfer Error slot d Status 2x VME address 8x 1193 2 Drive Fault slot d drive d sense key 2x sense code 2x 1194 2 Drive Offline slot d drive d 1195 2 SCSI Interface Err slot d drive d SnsKey 2x SnsCode 2x Stat 0 2x Stat 1 2x Ctl 2x Bus 2x 1196 2 SP Data Transfer Error slot d drive d SnsKey 2x SnsCode 2x sector 9x 1197 2 Drive Box Power Failure slot d Box d 1198 2 SP Data Transfer Error slot d drive d SnsKey 2x SnsCode 2x sector 9x 1199 2 Access to unknown SCSI device at slot d denied 1200 3 Recovered drive error no sense info slot d drive d 1201 3 Recovered no drive index or sector signal slot d drive d 1202 3 Recovered no drive seek complete slot d drive d 1203 3 Recovered drive write fault slot d drive d 1204 3 Recovered drive not
156. auses ax_mattach to run in the foreground This new member is automatically recorded in etc opartab The new member must be the same size as the existing member of the partition Note The data in the virtual partition is not mirrored until this step is complete Changing a Striped or Concatenated Partition to a Mirrored Partition You can convert a striped or concatenated virtual partition to a one member mirrored partition without affecting the file system mounted on the virtual partition Because no unexporting or unmounting of the file system is required users have uninterrupted access to the file system while the conversion takes place Note You cannot convert a striped or concatenated partition directly to a two Li member mirrored partition You can however use ax_mattach to attach a second member to the mirrored partition after the conversion process Also you cannot convert a partition if there is a UFS file system mounted on it Defining and Reconfiguring Virtual Partitions a 5 19 To convert a virtual partition to a mirrored partition use the ax_mconvert command Its syntax is as follows ax_mconvert partition The partition in the command must be a striped or concatenated partition currently defined in etc vpartab Applying the ax_mconvert to a mirrored partition returns an error message The ax_mconvert command creates a new virtual partition and makes it the only member of the mirrored partition The mirrored par
157. aw device 4 3 reducing latency 5 2 remapping a sector 6 23 root backup 8 11 8 14 sample ax_diskconf display 5 24 5 25 sector Q 6 24 static table format 4 22 surface analysis 4 38 Disk errors definition 6 2 during normal operation 6 8 intermittent 6 21 ona mirrored partition 5 13 recovering on a running system 6 19 Disk partition tables default 4 8 striped 4 8 Disk partitioning 4 7 Disk problems 6 11 Disk type choosing using format 4 37 Diskless SunOS client adding 2 28 loading executables 2 20 removing 2 28 Displaying a disk partition table 4 9 ARP table 10 14 disk configuration information 5 23 multiple histograms 9 23 network statistics with ax_netstat 10 15 to 10 18 NFS statistics with nfsstat 10 19 system configuration data 10 5 Distributing client files 4 11 Distributing file systems 9 47 Distribution medium CD ROM 2 25 4 5 default mount point 2 25 tape 2 25 dkinfo command 5 24 checking disk label 4 28 checking root disk label 6 27 different from SunOS 1 13 displaying partition table 4 9 used with CD ROM 4 2 Documentation applicable to the NetServer xvii installing online B 1 Domain name 2 15 2 30 Down arrow key 2 10 Drive adding 4 25 7 4 hot plugging 4 26 LED indicator 8 6 removing 4 31 SCSI ID 4 34 types supported by the NetServer 4 25 Drive configuration change recovering from 7 11 Drive rack installing 4 25 Drive slot number in an SP error message 6 8 Driver virtual partition 5 9 dump command
158. ax_write_cache s 0 enable Partition and label the new disk in the same way you prepared the old disk Use newfs to create file systems on the disk and restore the data from tape backups Remount the file system Recovering From a Drive Configuration Change This section provides an example for recovering from a drive configuration change on SP0 1 Enter the ax_write_cache command to verify the cache state ax_write_cache s 0 state If the cache state is OFF go to step 4 Otherwise flush the unwritten data to disk ax_write_cache s 0 flush Purge all other data from the cache ax_write_cache s 0 purge Re enable the write cache by entering the following ax_write_cache s 0 enable Write Cache States and Commands a 7 11 W AUSPEX Backing Up the NetServer About This Chapter A good backup strategy is vital to the protection of your data A program of regularly scheduled backups protects your data from corruption because of a system crash or other cause and protects user files against accidental deletion before attempting to back up or restore file systems Refer to Tape Drives on page 4 2 for more information on how to name the drives i Note Familiarize yourself with the Auspex tape drive naming conventions This chapter in three major parts summarizes how to back up and restore file systems on the NetServer a Backing up a physical disk partition to tape which must be performed when the part
159. axis By default it is Time which means that each value on the X axis is the time when the sample was collected If set to Samples the values on the X axis correspond to the numbers of the samples 9 26 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Y Title specifies the string printed here Y Pos determines how far this box is from the upper edge of the window X Pos determines how far this box is from the left edge of the window This dotted line represents a threshold which is set to 8 0 The line makes it easier to count the number of samples exceeding the threshold A Y inc of 0 5 causes the values on the Y axis to be printed 0 5 unit apart First value on Y axis is 0 if Lower Y Label is set to 0 This bar represents the first sample if First X Label is set to 0 7 Plot Title specifies the string printed here Auspex System Data Display Tool F Properties v Histl Fri Feb 3 14 48 52 1995 Serverzacdelco File acdelco dat 35 Fri 14 48 58 Fri 14 49 00 Fri 14 49 03 Fri 14 49 05 Fri 14 49 08 Fri 1434924 HP cpu sys_time Ave 8 4 X Axis Display is set to X Title specifies the string printed Time so each value onthe here X axis represents the time when the sample was collected If both Total Samples and Last X Label are 31 this is the 31st sample Figure 9 15 Effects of Histogram Properties parameters on a histogram
160. ay for default routing Specifies how to run the route daemon Email address or name and telephone number of the person who is responsible for this server for example jsmith companyname com Location of the server for example 2nd FI Building 2 ABC Corp San Jose Specifies whether to set both the date and time on the server The setting does not affect the time zone setting If you enter Yes enter the date and time in the associated field A one second difference exists between the time when the form is executed and when the time is set Auspex recommends that you set this field last right before executing the form Specifies whether to change the local time zone The setting does not affect the date and time If you enter Yes select the time zone name in the associated field When your server is rebooted this field determines whether the server automatically sends an email message containing the message log to Auspex The subject line in the message includes the string you specified for Customer Site notifying Auspex Technical Support that your NetServer might have problems that require attention The script file that sends the email message is usr auspex mailmessages sh Refer to the mailmessages sh 8 man page for more information 2 16 a Version 1 10 System Manager s Guide Yes or NO Yes No or Quiet If set to Yes default routing information is distributed to all other machines on the network t
161. ber of bytes processed by each network interface per second is calculated See the NBUF stats row of Table 9 1 for an explanation of the interface definitions that also show up here on the summary screen An N entered with the summary screen moves the NP portion of the screen to the next NP board All interface numbering restarts from 0 on each board For the FP two types of information are shown The number of LFS operations per second In the bar different characters represent different NFS operations refer to the FP Statistics screen for definitions of each character The ages of data cache buffers associated with the FPs The letters in the bar represent the time since the buffers were used A represents buffers used within the last minute B represents buffers used within the last two minutes Z represents buffers that have not been used for more than 26 minutes Lowercase letters have no significance to most users For the SP the display shows the number of operations per second on each device controlled by the SP The devices are arranged by drive racks each of which corresponds to a bar graph Within each bar graph R represents a read operation and W represents a write operation Note The rack numbers are relative to the SP being displayed meaning that D rack numbering always starts from 1 Refer to your NetServer hardware manual for a detailed description of the drive racks suppor
162. boot the system Remote Diagnostics If you contact Auspex for help in resolving an Auspex NetServer problem the service representative may need to perform diagnostics on the system remotely This section outlines the procedure for obtaining remote diagnostic assistance 1 Enable diagnostic mode by setting the diagnostic switch on the HP to the diag position Refer to the hardware manual for your server for the location of the diagnostic switch The following procedure describes access with a modem connected to the NetServer Other access methods may be to a terminal server which can be connected to serial port b ttyb You can access a terminal server through a modem or the Internet 2 Attach a modem to ttyb Verify the modem is powered on and the telephone cable is properly attached Refer to the modem s user documentation for information on connecting and configuring the modem 3 Depending on the nature of the problem the Auspex service representative may ask you to shut down the system If so do the following a Shut down the system using shutdown 8 or halt 8 b When shutdown is complete and a PROM monitor prompt appears on the console power down the NetServer 4 Notify the Auspex service representative to establish a connection to the modem attached to the NetServer Once the modem connection is established follow the service representative s instructions to reboot the NetServer 10 12 a Version 1 10 System M
163. cause the server to crash The HP VIII is configurable from 128 to 384 MB of memory with 16 MB 32 MB 64 MB or 128 MB add on modules Memory modules must be installed in pairs of equal size NetServer Hardware Architecture a 1 5 Network Processor NP The Network Processor combines network processing and file processing functions as well as onboard cache memory An NP may have multiple mixed network connections including Ethernet and FDDI connections An FDDI NP can support either fiber or MLT 3 interfaces and can be either a Single Attach Station SAS or a Dual Attach Station DAS The DAS connection supports fiber only NP network interface adapter boards supporting ATM FDDI and Fast Ethernet 100Base T interfaces are available as optional products For a complete list of supported network interface combinations see the Network Processor Configurations section of the Version 1 10 Hardware Release Note Storage Processor SP The Storage Processor operates multiple parallel SCSI channels The SP V supports six 10 MB per second fast SCSI channels The SP is responsible for SCSI channel management SCSI drive control and optimization DMA data transfers to the primary I O cache memory and virtual partition management Optionally the SP supports write acceleration for NFS write operations through write caching write coalescing and redundant write elimination Write acceleration can improve client NFS response time
164. changes to OFF you can retry the fsck command If the flushing is not successful there may be a malfunctioning disk in the system that prevents data from being written Check the SP error messages in var adm messages to determine which sector caused the problem Use ax_sputil reassign to reassign the bad sector and try flushing again If it still fails purge the data from the cache For information on the arguments to fsck refer to the fsck man page If you want to know more about how fsck repairs corrupted file systems refer to Sun s System and Network Administration If the file system is successfully repaired by fsck go to Releasing a File System After Repairing It by fsck for information on making the file system available again If the file system cannot be repaired because of a bad disk go to Restoring a File System that Cannot Be Repaired on page 6 15 for information on killing processes sleeping on the file system Releasing a File System After Repairing It by fsck After fixing the inconsistency of an isolated file system release the file system to make it available again with the ax_fsutil command Use the partition name or the mount point as the argument in the command For example to release the file system with the mount point mnt enter the ax_fsutil command with the release option as follows ax_fsutil release mnt If dev vp3 is mounted on mnt you can also enter the command in this form ax_fsutil releas
165. client architecture Because export directories are exported the export file system is assigned the LFS file system type in etc fstab export sun4c sunos 4 1 3 km proto root sunos 4 1 3 mim enoa KoG proto root sunos 4 1 3C sun4 sunos 4 1 3 sun4 sunos 4 1 3C exec sun4c sunos 4 1 3 gt sun4 sunos 4 1 3 sun4c sunos 4 1 3C gt sun4 sunos 4 1 3C sun4m sunos 4 1 3 gt sun4 sunos 4 1 3 sun4m sunos 4 1 3C gt sun4 sunos 4 1 3C lost found lib man sunos 4 1 3 nie share sys gt kvm sys lib man sues 41380 Sre sys gt kvm sys Figure 4 8 The export file system for servers with SunOS executables exportN file system Figure 4 9 shows the exportN file system structure where N represents the disk number The exportN directories contain the client root and swap file systems Because these directories are exported this partition is assigned the LFS file system type in etc fstab The exportN file system is unnecessary if you use one virtual partition for export root and another for export swap Refer to Example of File System Distribution on page 4 11 for more information on how to configure partitions for client root and swap directories export 1 n root swap aaki crash Figure 4 9 The exportN file system for clients For information on how to configure clients on the system see SetupClient on page 2 27 Disk Partitioning a 4 17 AUSP
166. combined statistics from input files 1 and 2 Display the samples up to the first threshold marker For example if Threshold is set at 10 0 this option causes the histogram to use 10 0 as the maximum value on the Y axis If you want to display the entire histogram again invoke the Load File option Remove the input file from your directory Displaying Performance Data in Histograms A 9 37 Printing Histograms To print the histograms in the Data Display Window click on the Print button in the Data Display Window The Print button invokes the predefined print command from the Print Options window You can print the histogram to a printer or a file This section describes the Print Options Window which contains parameters determining how histograms are printed After setting the print options click on the Apply button in the Print Options Window The Reset button restores the values of the print options if they have not been applied Figure 9 23 explains the functions of the print options for detailed information about configuring the options refer to the following sections Print command automatically No argument or parameter Possible commands for capturing the changes based onthe argument values are configurable Enter screen before the screen image is and parameter values configured the print command manually sent to a printer or saved to a file in this window Print Options Window Control Screen Capture Prints the s
167. creen image Output to the printer or Printer Name saves the image to a file Print Command xwd frame xpr device ps lpr P Name of the printer Print command for capturing the screen image and sending it to a printer or file Figure 9 23 Print Options Window Print Command In the Print Options Window you can specify the command that ax_perfhist uses for printing the Data Display Window The default print command does the following a Captures the screen using xwd Aa Sends the output to a printer using lpr The following is an example of the print command xwd frame xpr device ps lpr P Note Treat the default print command as a suggested command it may need i customizing depending on the printer setup at your site For example if you want to print to a printer other than the default one append a printer name to the command 9 38 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Changing the print command automatically The Print Options Window contains several parameters that configure the print command If you want the print command to be configured automatically select Auto for the Control option When Auto is selected the change you make to any print option is reflected immediately in the Print Command field Select Manual for the Control option to enter a command manually in the Print Command field Configuring the print options The information in this section applies only
168. crosoft Corporation Sun Sun Microsystems the Sun Logo Solaris SunOS ONC ONC NFS and NFS are trademarks or registered trademarks of Sun Microsystems Inc All SPARC trademarks are used under license and are the trademarks or registered trademarks of SPARC International Inc in the United States and other countries UNIX is a registered trademark in the United States and other countries of The Open Group VMEbus is a trademark of VMEbus Manufacturers Group DEC and VT 510 are trademarks of Digital Equipment Corp ForeRunner is a trademark of FORE Systems Inc Acrobat is a trademark of Adobe Systems Inc Auspex NetServer System Software is derived from UNIX licensed from The Santa Cruz Operation Inc and SunOs 4 1 4 and ONC NFS 4 1 licensed from Sun Microsystems Inc Auspex NetServer System Software Version 1 10 incorporates AT amp T s Advanced Server for UNIX Systems Auspex Optional Products Premier Software Series for NeTservices incorporates AT amp T s Advanced Server for UNIX Systems and NETBIOS ix NETBIOS ix is a registered U S trademark of Micro Computer Systems Inc Microsoft may have patents or pending patent applications trademarks copyrights or other intellectual property rights covering subject matter in this document The furnishing of this document does not give you any license to these patents trademarks copyrights or other intellectual property rights except as expressly provided in any wr
169. cs pkts in pkts out pkts w error pkts forwarded FTP statistics puts gets put bytes get bytes workers TGP statistics segs in segs out segs w err curr estab conns active opens passive opens workers Version2 or version3 NFS calls Number of NFS calls over UDP Number of NFS calls over TCP Number of datagrams received Number of datagrams sent Number of checksum errors Number of packets received Number of packets sent Number of packets received with errors Number of packets forwarded Number of puts Number of gets Number of bytes in puts Number of bytes in gets Number of FTP workers configured Number of segments received Number of segments sent Number of segments received with errors Number of connections currently established Number of active opens Number of passive opens Number of TCP workers configured Capturing and Displaying Real Time Performance Data a 9 11 FP Statistics Screen The FP Statistics screen displays statistics for one FP Figure 9 5 shows a sample FP Statistics screen 1 0 FP Stats wizard Thu Nov 21 14 21 52 1996 for FPL CPU 92 Ops sec t Op Dist G getattr L lookup R read Wewrite D readDir a access N reName S Setattr K readlinK C create r remove d readDir c commit X other Eject Eject Hit Access Age Count s Rate Count s data cachet 0200 1002 Oc 0300 meta cachet 0 00 1002 Ol 0300 inode cachet 0300 1002 Oo name cachet 000 100
170. cs collected For example if the maximum value among all the samples is 28 1 the highest value on the Y axis is 28 1 Automatic scaling is disabled The highest value on the Y axis in the current histogram is maintained when you load different statistics to it For example if the highest value in the histogram is 28 Preset Value uses 28 as the highest value regardless of the parameter or input file you subsequently load to histogram 1 When multiple histograms are displayed they are arranged in columns When multiple histograms are displayed they are arranged in rows Number of columns or rows used to display the histograms in the Data Display Window Saving and Applying a Configuration File If you often use a specific layout of the Data Display Window you can save the layout information to a configuration file When you apply the configuration file the Data Display Window arranges the histograms according to the specifications in the file Displaying Performance Data in Histograms a 9 29 Also if you plan to use the ax_perfhist command non interactively for example for printing out histograms define a configuration file Refer to Starting ax_perfhist with Options and Arguments on page 9 46 for more information on specifying and invoking the configuration file in the command syntax The following example illustrates how to save a configuration file that displays six histograms in the Data Display Window 1 Inth
171. ct files Auspex utility to access the PROM Extended Diagnostics menus from UNIX Auspex utility to display disk configuration information Display online system and optional product documentation Auspex UNIX level utility to obtain FRU information stored on Auspex drives Drive statistics maintainer Maintenance Commands a A 7 ax_drvbuild 8 ax_drvmaint 8 ax_enable 8 ax_expand 8 ax_fddistat 8c ax_fsmegr 8 ax_fsutil 8 ax_hot_plug 8 ax_isolated 8 ax_keyenvoyd 8c ax_kill 8 ax_label 8 ax_label_root ax_lfsd 8 ax_lIfstest 8 ax_load_flash 8 ax_loadvpar 8 ax_Islabel 8 ax_mattach 8 ax_mconvert 8 ax_mdetach 8 ax_meminfo 8 ax_memtest 8 ax_mrestore 8 ax_netstat 8c ax_nfsd 8 ax_perfhist 8 ax_perfmon 8 Maintains SCSI disk firmware files downloaded by ax_drvmaint Downloads SCSI disk firmware files Enables Auspex optional products Enlarge a file system on a virtual partition Shows status of FDDI CDDI MLT 3 network interfaces Auspex file system manager Utility for miscellaneous file system services Utility for replacing multiple disk drives while the NetServer is running Daemon for monitoring file system isolation events Auspex daemon that talks to keyserver Command for killing all processes sleeping on an isolated file system Partition a disk and label the drive Repartition and write a label for a root drive Start LFS daemons on Auspex Host Processor Auspex LFS di
172. cts are available at the time of printing this manual Refer to the Auspex Premier Software Series CD ROM for the latest information ServerGuard ServerGuard allows mirroring of an NFS filesystem across a pair of NetServers One NetServer acts as the primary server and the other which may be physically remote anywhere in the world acts as the secondary server With ServerGuard the NFS filesystem is independent of local power failures and natural disasters resulting in uninterrupted NFS service to clients DataGuard DataGuard permits HP only reboots This confines operating system failures to the HP and any services directly provided by it DataGuard makes the NetServer more reliable by reducing the impact of UNIX operating system failures and improving the HP reboot time DriveGuard DriveGuard RAID 5 implementation protects user data in the event of a single device failure If any single disk in a DriveGuard array fails the data can be reconstructed using information on the remaining disks in the array See the DriveGuard documentation on the Auspex Premier Software Series CD ROM for more information NeTservices NeTservices is a UNIX based network operating system that is fully compatible with Microsoft networking technology NeTservices provides file services for the efficient sharing of computing resources among a community of desktop users It incorporates the latest Microsoft technology to deliver powerful new network admini
173. current line Deletes the word preceding the cursor Scrolls the form to the top and moves the cursor to the first field Scrolls the form up Scrolls the form down Toggles a value When you set the time zone in the NSconfig form keep pressing the space bar until the appropriate time zone appears Moves the cursor to the next field Moves the cursor to the first field on the next line Moves the cursor down to the same field on the next line Moves the cursor to the previous field Moves the cursor up to the same field on the previous line NSconfig configures NetServer software only Use the SetupExec and SetupClient commands to install and configure software for diskless SunOS clients 2 10 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Before Running NSconfig Before you begin you need the following information A A The primary host name of your server For each network interface the primary host name Internet address subnet mask and the host part of the broadcast address either all zeros or all ones The X terminal name and Internet address if you are using an X terminal as the server console The NIS service type and NIS domain name Whether or not the route daemon runs on the NetServer fy Note Load balancing through network routers or hubs is supported with the NetServer However the NetServer network interfaces that are designated as belonging to the same host must be o
174. d 0 1 1 After the system boots the following messages appear Remounting and replenishing etc Remounting and replenishing var Remounting and replenishing tmp AUSPEX CD ROM MAINTENANCE MENU OPTION 1 Format label the root drive 2 NSinstall 3 Maintenance shell Select an option 1 2 3 3 Select 3 to enter the maintenance single user mode shell The single user mode prompt appears Check the label on the root drive using the dkinfo command dkinfo ad0 If the label is incorrect apply the correct label using format Determine whether the damaged disk can be repaired using fsck fsck dev rad0a fsck dev rad0g fsck dev radOf Recovering a Damaged Root Disk a 6 27 8 10 11 If the root partition is severely damaged go to step 9 If it can be mounted follow these steps a Mount it and examine the root partition to see what damage would cause the system not to boot For example mount dev ad0a a ed a ls l vmunix boot sbin ls lt etc more The output of these commands helps you determine whether the files crucial for booting are present and whether they have been changed b If you notice the files are modified or missing restore them from the most recent backup If a backup is not available go to Using NSinstall to Install Software From CD ROM on page 6 29 to restore the root partition from the CD ROM After restoring the partition modify
175. d NSinstall to copy software from the CD to a spare disk no system reboots are necessary after the installation the NSinstall procedure is now complete Remove the spare disk from the slot and save it in a safe place The system may or may not reboot after NSinstall depending on whether you enabled automatic reboot in the NSinstall form If automatic reboot occurs go to the next step If the automatic reboot brings the server up in multiuser mode use the shutdown command to enter single user mode before going to the next step If you reinstalled files to the current root disk and specified no automatic reboot enter the system maintenance commands that are appropriate to your server After you finish with the commands reboot the system in single user mode Using NSinstall to Install Software From CD ROM a 6 33 AUSPEX D installing system software from the CD you must reinstall the optional products before rebooting the server NSinstall does not preserve optional products during installation because older versions may not be compatible fy Note If optional software products were installed on your server before The key and license are preserved during the NSinstall procedure After the system boots in single user mode run NSconfig Follow the procedures in NSconfig on page 2 10 to complete the configuration When NSconfig is finished type exit to go to multiuser mode If you enabled newfs in the NSinstall form you ma
176. d partition consists of only one member Moving a Striped or Concatenated Partition This section describes how to move a striped or concatenated partition from one group of drives to another The drives must still be managed by the same SP The following procedure uses vp1 as the name of the partition to be moved 1 Use ax_mconvert to convert vp1 to a one membered mirrored partition The ax_mconvert command creates a new virtual partition for example vp2 to be the only member of vp1 vp2 comprises the physical partitions that were in vp1 2 Create a new virtual partition for example vp3 consisting of the group of physical partitions to which you want to move the old striped or concatenated partition Defining and Reconfiguring Virtual Partitions a 5 21 3 Use ax_mattach f to attach vp3 to the mirrored partition created in step 1 This step starts the ax_mrestore command which copies the contents from vp2 to vp3 Go to the next step only after ax_mrestore is finished 4 Use ax_mdetach to detach vp2 from the mirrored partition After this procedure vp1 becomes a one membered mirrored partition with its contents stored on physical disk partitions that make up vp3 Moving a One Membered Mirrored Partition This section describes how to move a one membered mirrored partition from one group of disks to another In the following procedure the mirrored partition is vp1 the member of vpl is vp2 1 Create a new virtual partition
177. d routes new gateways due to redirects destinations found unreachable 14 uses of a wildcard route SO Orne If you do not specify the i s or r flag in the ax_netstat command the flag defaults to s and protocol information displays The netstat 8C command which is part of SunOS displays information similar to that printed by ax_netstat The major differences between the commands are listed as follows a Statistics gathered by ax_netstat pertain to packets processed by a network processor board netstat when used with the I interface option displays statistics for packets that are received or sent out by the interface but are processed by the HP A If the I interface option is used in a netstat command and the specified interface is down or absent nothing prints except a header The ax_netstat command however displays zero counts for various types of statistics and places an asterisk next to the interface name to show an interface that is down If the interface is not installed in the server an error message appears as follows ax_netstat device ael7 is not an installed network interface 10 18 a Version 1 10 System Manager s Guide W AUSPEX Displaying NFS Statistics The nfsstat 8C command in SunOS displays NFS statistics such as client information and RPC information The command is modified by Auspex to include the I Interface option for displaying information about a selected network interface such
178. dded to etc fstab specifying the file system you want to enable For example to enable write caching for the file system home dvlp create the following entry dev ad30c home dvlp lfs fs 4 2 we 0 2 In this example we write cache enables write acceleration for the file system home dvlp 7 8 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Note The wc option is ignored if no Write Accelerator is installed on the SP in i which case NFS write operations are handled conventionally To enable write acceleration on SPO from the OFF state enter the following command ax_write_cache s 0 enable The SP automatically disables write acceleration when the cache state switches to BADCHECKSUM or DIRTY Refer to Managing the Write Accelerator on page 7 9 for error recovery procedures Disabling the Write Accelerator The write cache is automatically initialized and switched to the ON state enabled at boot time Disable the write cache using any of the following methods a Use the ax_write_cache command to switch the cache state to OFF For example when the write cache state on the SP0 is ON enter the following command at the root prompt to change it to OFF ax_write_cache s 0 disable a To disable the write cache function for a specific file system remove wc from the specific file system entry in the etc fstab file Then unmount and mount that file system Additionally the SP automatically di
179. de cache Hit Access Rate Rate of inode cache hit access FP fp inode cache Hit Access Count s Number of inode cache hits per second FP fp name cache Eject Age Average age of items dropped from the name cache FP fp name cache Eject Count s Number of items dropped from the name cache per second FP dat_bstats data cache Hit Access Rate of data cache hit access Rate 9 44 a Version 1 10 System Manager s Guide Average age BQ_LRU ctl bufs dropped from cache Average age BQ_AGE ctl bufs dropped from cache AUSPEX W W AUSPEX FP dat_bstats data cache Hit Access Number of data cache hits per second Count s FP ctl_bstats meta cache Hit Access Rate of meta cache hit access Rate FP ctl_bstats meta cache Hit Access Number of meta cache hits per second Count s FP nc name cache Hit Access rate FP nc name cache Hit Access Count s SP SP cpu busy SP nvram io_count SP nvram overrides SP nvram coalesce SP nvram buf_used PAGE_8K SP nvram buf_used PAGE_512B SP vpar_total vp SP Disk disk sctr_rd SP Disk disk sctr_wr SP Disk disk ops_rd SP Disk disk ops_wr SP Disk disk ops_total SP Disk disk total_time_per_op SP Disk disk queue_time_per_op SP Disk disk service_time_per_op SP Disk disk drive_utilization SP all disks sctr_rd SP all disks sctr_wr SP all disks ops_rd SP all disks ops_wr SP all disks ops_total Rate of name cache hit access Number of name cache hits per second
180. ding section follow this procedure to add a drive to the server 1 Enter the ax_hot_plug command specifying the slot number of the drive to be added or replaced For example if you wish to add or replace a drive in slot 6 enter the following ax_hot_plug add 6 If a tape drive on the same SP is active the server displays the following error message ax_hot_plug a tape drive is in use on the same SP as the device to be added or changed cannot replace drive This message ensures that no active tape drives are rewound because of the power surge in the drive rack when a drive is added If the message appears you must invoke ax_hot_plug again after all tape drive activity on the same SP has stopped Note The ax_add_device command is also used for adding drives to the D NetServer Refer to the ax_add_device man page for more information When ax_hot_plug executes successfully the system suspends SP activity to prevent I O operations on the SCSI bus and prompts you to insert or replace the specified disk drive The command then tells you to type a carriage return after inserting the new drive Note On a quiescent SP all I O operations are halted NFS requests received i from the NetServer s NPs are halted until the SP restarts This may cause timeouts on client workstations making NFS requests so it is important to install the new drive as quickly as possible 2 Install the new drive 3 Type a carriage return immediate
181. disk drives to reduce disk latency because of request queueing a define mirrored partitions to protect file systems from disk or media failure and provide high data availability a back up file systems without taking the file systems offline 5 2 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Maximum Size of a Virtual Partition The Auspex virtual partition breaks the 2 GB limit traditionally imposed by UNIX systems A virtual partition can have up to 16 members with each member partition limited in size only by the size of the disk holding that member As an example 16 members taking up 16 9 GB disks give a virtual partition of 144 GB Note The size of an indivitual file in an LFS file system is limited only by the i size of the file system when accessed through NFS Version 3 Clients accessing such files must have large file capabilities Advantages of Using a Greater than 2 GB Virtual Partition The benefits of using a virtual partition depend on the type of partition The various types of virtual partitions are described in the following sections Regardless of the partition type defining partitions that exceed 2 GB offers these advantages a Easier file system management You now have the flexibility of grouping a large number of files in one file system which allows you to use fewer dump and restore commands for system backups Also there are fewer file systems to export a More efficient use of di
182. e booting an NS 7000 250 the unit number is the same as the disk slot number For example to boot from a disk drive in disk slot 5 determine the SP number 0 4 and enter the following command HP gt b ad 0 5 0 This example specifies a boot from SPO the first SP in the NetServer drive 5 and partition 0 see the part argument below Here are some other typical boot examples HP gt b ad 0 0 0 Boot the NetServer from the root drive HP gt b ad 0 2 0 Boot the NetServer from the back up drive in disk slot 2 HP gt b ad 0 1 1 Boot the NetServer from the CD ROM drive in disk slot 1 part 1 which is described as follows part Disk partition number if booting from disk or HP identification number if booting from CD ROM for the boot device For disk drives this number ranges from 0 to 7 where 0 corresponds to partition a and 7 corresponds to partition h The default is 0 for partition a For CD ROM drives this number is 1 path The full path specification of the file to boot The default is vmunix options A list of options for the boot program Booting the NetServer a 3 5 a Includes the ask me option on the boot command This option is required when booting from nonstandard partitions When using the ask me option the system asks you to specify the root and swap partitions explicitly Normally you do not use this option when booting the NetServer s Boots in single user mode If you do not specify
183. e dev vp3 ax_fsutil checks etc mtab to verify the file system you try to release is indeed mounted If etc mtab has no entry for the file system ax_fsutil prints out this message ax_fsutil release mountpoint filesystem If ax_fsutil is successful activity on the file system resumes and the following message is generated ax_fsutil Successfully released filesystem mounted on mnt Note Invoke ax_fsutil without an argument to list any file systems that have been isolated You are not required to be root to run ax_fsutil in this form ax_fsutil works only on a clean file system Therefore be sure to run fsck on an isolated file system before using ax_fsutil to release it If you try to release an unclean file system messages similar to the following appear on the console Dec 20 15 34 42 cabot IOPO FPO CPU B fc_unfence File system test ad2f isn t clean and can t be released Dec 20 15 34 42 cabot IOPO FPO CPU B Can t release got error 16 6 14 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX On the terminal where you enter the command the following error message appears ax_fsutil test ad2f isn t clean Cannot release If a disk problem caused an isolated file system using fsck to repair the file system does not solve the problem Note that the fsck command may not alert you to the disk problem For example if the problem is intermittent fsck may mark the file system clean afte
184. e 4 36 for format Using CD ROM Drives on page 4 5 for eject Adding or Replacing a Drive on page 4 25 Man pages for MAKEDEV auspex dkinfo ax_lIslabel ax_label Naming conventions for storage devices Auspex disk drives are named adn Tape drives are named rastn or nrastn CD ROM drives are named acdn nis the slot where the drive is installed DriveGuard arrays are named ardn where nis the number of the array SunOS Enhancements Device Names on page 4 2 A 1 13 Table 1 2 Changes to SunOS Continued Types of changes Description For further information see File systems An added file system type LFS acts as an interface between the operating system and file systems mounted on the File Processor In general LFS remains transparent The NetServer uses the standard 4 2 file system type for its root usr and var file systems exports 5 is modified to add an option allowing asynchronous writes to exported file systems Many other standard utilities including df and du are modified to support LFS and 72 GB file system capacity A file system on a virtual partition can exceed 2 GB Up to 72 GB file systems can be created on an Extended Virtual Partition EVP ax_clonefs 8 creates a clone partition of a specified physical or virtual partition to facilitate online backups When the server detects a corrupted LFS file system the File Processor isolates the file s
185. e 5 7 Figure 5 8 Figure 5 9 Overview of NetServer Features Hardware implementation of Functional Multi processing architecture 1 9 NetServer Functional Multi processing architecture 1 10 Software Configuration Flowchart for configuring the server for the first time 2 3 Sample NSconfig TORN essen hs stir e aady ce eee eae ites 2 13 Flowchart for configuring a TTY port c esisaddia lapagetedariaes 2 18 The SetupTty forn 2 otra si ae us Ges wa eta erie Git ooh a 2 19 Configure client information on the server using SetupExec 2 20 Sample SetupExec form for a local tape device onnan nannan 2 23 Sample SetupExec form for a local CD ROM device 2 23 Sample SetupExec form for a remote tape device 0 2 24 Sample SetupExec form for a remote CD ROM device 2 24 Configure client information on the server using SetupClient 2 27 The SetupClient forms carci ore irent ern winien eik EEEE E ER a 2 29 Booting Up and Shutting Down the NetServer NetServer Storage Devices and File Systems Partitioning a 9 GB disk using the default partition table 4 9 Partitioning a 9 GB disk using the stripe partition table 4 9 Partitioning a 4 GB disk using the root partition table 4 9 The Auspex root file system gia ive oe eit eee eae 4 13 The Auspex var file system yore naktoth arr eela ds Laveredyas 4 14 The Auspex usr file
186. e Data Display Window select 6 for Qty Hists 2 In the Load Window select the type of statistics for each histogram using the Parameter button 3 Configure each histogram and the global characteristics of the histograms according to Defining the Appearance of a Histogram on page 9 24 4 In the Data Display Window click on the Print button to display a pull down menu Select Print Properties in the menu The Print Options window appears which allows you to determine how the histograms are printed For information on how to specify the print options refer to Configuring the print options on page 9 39 5 Inthe Data Display Window click on the File button to display a pull down menu Select Config files in the menu The Configuration window appears 6 In the Configuration window type in the name of the configuration file and the directory that contains the file 7 Click on the Save Config button in the Configuration Window see Figure 9 18 When the system asks you to confirm click on the Save button The configuration file contains information about how each histogram is displayed how the histograms are arranged in the Data Display Window and what print options are used if the histograms are sent to the printer or a file All this information is loaded when you apply the configuration file Configuration Window 2histogram cnf a 6histogram cnf co Directory Config File File Ext
187. e Data a 9 15 Using a Filter File with ax_perfmon The ax_perfmon command syntax allows you to specify a filter file listing performance parameters about which you want to save statistics Use a filter file if you want to A keep the output file small The variables in the filter file define the types of statistics stored in the output file For example if you are interested only in statistics about Ethernet interfaces create a filter file that filters out all ax_perfmon data not related to Ethernet networks a convert performance data to ASCII format Without the filter file the ax_perfmon output file contains ASCII data of decimal numbers in an undocumented format It cannot be used as input for databases or spreadsheet software Specifying a filter file in with ax_perfmon automatically converts all the statistics to decimal numbers in an easily understood format Note In pre 1 6 releases ax_perfilter converted performance data to ASCII D format In version 1 6 and later the filter file specified with ax_perfmon eliminates the need for a separate command for converting data formats As a result ax_perfilter is removed in Auspex software Version 1 6 and later The following sections describe how to create a filter file and provide a sample output file Creating a Filter File These sample filter files are provided in the usr auspex directory a summary_filter for summary statistics a np_filter for NP statistics a fp_filt
188. e NetServer and their characteristics HSFS You can mount an HSFS file system but only on the HP not as an LFS file system on an FP For example you can mount a file system on the SunOS CD ROM as an HSFS file system UFS You can mount a file system of type 4 2 or type lfs with an fs 4 2 option For example you can mount a file system on a Catalyst CD ROM as a UFS file system The file system must be mounted read only any attempt to write to the CD ROM generates an error message The following instructions provide an example for mounting a CD ROM 1 Edit the etc fstab file to add an entry for the file system using the read only ro option In the following example the CD ROM in slot 1 is in UFS format and is mounted as an LFS file system dev acdl library lfs fs 4 2 ro In the following example the CD ROM is in HSFS format dev acd1 library hsfs ro Caution If you include the CD ROM in etc fstab be sure that the CD is in the drive each time the system boots Otherwise fsck fails at reboot and the system cannot boot to multiuser mode Using CD ROM Drives a 4 5 AUSPEX D Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems wi
189. e SP fails to attach the drive Is the Problem Caused by a Malfunctioning Drive Slot If the procedure in the preceding section does not succeed that is the SP still cannot detect the drive after you reinstall it a SCSI interface error message appears Follow these steps to install the drive in another slot 1 Remove the drive from its current location It is not necessary to enter the ax_hot_plug command because the drive is not recognized by the system software Caution Do not remove a disk while it is spinning down Removing a disk drive while it is still spinning can cause permanent damage to the drive 2 Select a new slot for this drive 3 Use the ax_hot_plug or ax_add_device command to add the drive in the new slot If you need additional information on these commands refer to Adding or Replacing a Drive on page 4 25 If the drive is detected in the new slot continue using the disk at the new location If file systems are mounted on the disk edit the etc fstab file to reflect the new disk location Also if the disk is a member of a virtual partition edit the etc vpartab file and then run ax_loadvpar to load the revised virtual partition table 4 Run fsck p to check all the file systems and then press Ctrl D to go to multiuser mode Note The slot in which the drive was originally installed probably is not D working report the problem to Auspex If the drive is still not detected in the new drive slot locat
190. e file systems to the FPs A VP256 Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades ENABLED do not delete 4 20 a Version 1 10 System Manager s Guide 4 2 swap 4 2 4 2 lfs lfs lfs hsfs slot 2 is the Ifs lfs lfs lfs lfs lfs rw ro rw fs 4 2 fs 4 2 fs 4 2 1 0 1 0 1 ag 1 note slot one is usually occupied by the CDROM ro noauto backup fs 4 fs 4 fs 4 fs 4 root drive 2 nosuid 2 WC 2 nosuid 2 We see etc vpartab fs 4 2 we Mirrored virtual partition see etc vpartab fs 4 2 we Caution The following line appears at the top of the etc vpartab and etc fstab files on systems running Version 1 10 AUSPEX W while W AUSPEX Using etc fstab to assign file systems to FPs To specify the FP on which a particular file system should be mounted use the fp n option in etc fstab The n in this option is the FP number The default FP number is 0 you do not need to use this option if you want to mount a file system on FPO For example to mount the cad file system on FP1 include the following entry in etc fstab dev adl10c cad l1fs fs 4 2 fp 1 1 2 Automatic file system mapping If no fp n option is specified in etc fstab the
191. e is assigned to partition h The following example shows partitioning ad5 with only the c partition ax_label c all ad5 The following example shows partitioning three disks ad7 ad9 identically with 200 MB a partitions and 300 MB b partitions with the remaining sectors assigned to the h partition ax_label a 200 b 300 h all ad7 ad8 ad9 Using ax_label to Partition and Labela Disk a 4 41 AUSPEX W Displaying a Disk s Partitioning Information This section describes the ax_lslabel command which displays information about physical disk partitions If you want to see comprehensive reports about file systems or virtual partitions use the ax_diskconf command which is described in Displaying Disk Configuration Information on page 5 23 The syntax for ax_Islabel is as follows ax_lslabel sc all disk The c option prints out the disk information in a concise format the sc option prints out the information in superconcise format The following examples show the ax_Islabel commands and their outputs hostname ax_lslabel ad0 ad0 has 3818 cylinders 16 heads and 64 sectors track per cylinder 32768 sectors 32 cyls 0 31 204800 sectors 200 cyls 32 231 3909632 sectors 3818 cyls 0 3817 131072 sectors 128 cyls 488 615 624640 sectors 610 cyls 616 1225 204800 sectors 200 cyls 1226 1425 262144 sectors 256 cyls 232 487 2449408 sectors 2392 cyls 1426 3817 rOomhoana oO Capacity
192. e l f e g h a b f a b g h See Disk Partitioning on page 4 7 for partition descriptions If the label has a problem or if an existing label needs to be changed use ax_label or format to correct the problem Run newfs 8 on the desired partition s on the disk Enter the following command newfs dev radnp In this command n is the disk drive number and p is the partition letter The device name ad has an r prefix specifying that it is a raw character device newfs can only be run on raw devices For more details see the newfs 8 man page Create the directory that will serve as the mount point for the partition For example mkdir home src Add an entry to etc fstab to create a permanent mount entry for the partition Using the previous directory as an example add the following entry dev ad8c home srce lfs fs 4 2 0 2 If your NetServer has multiple FPs you can specify the FP on which to mount the file system For example you can create the following entry dev ad8c home sre lfs fs 4 2 fp 1 0 2 If you want to use write acceleration for this new partition the entry reads as follows dev ad8c home src lfs we fs 4 2 fp 1 0 2 The server automatically distributes file systems to FPs if you do not specify any FPs in etc fstab and if you execute the mount a command For more information on mapping file systems to FPs refer to Mapping LFS File Systems to File Processors
193. e partition 8 12 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The b option logs errors to syslogd The p option shows the percentage of the file system that has been copied when ax_clonefs is in progress You can specify the file system name or partition name for the source partition for example usr openwin or vp3 Specify the partition name for the clone partition for example vp2 or ad4a Important Guidelines for Creating a Clone Partition Read and remember the following guidelines before you attempt to create a clone partition 4 Aclone must be at least the same size as the partition that it duplicates Figure 8 2 shows a sample clone of a mirrored partition on SPO The clone consists of six concatenated physical disk partitions on SP1 which duplicate all write operations on the mirrored partition 4 Before selecting a clone partition always use ax_diskconf to make sure the clone partition is not a member of a two membered mirrored partition or an underlying physical partition of any virtual partition For example to back up vp1 to vp3 using ax_clonefs make sure vp3 is not a member of a mirrored partition If vp3 is a member of vp2 ax_clonefs overwrites the data currently in vp3 and the members of the mirror no longer contain the same data A The clone partition is for backup purposes only Do not mount the clone partition or its underlying physical partitions and modify its contents after ax_clonefs
194. e partition map and label to the disk quit partition gt To use a predefined partitioning scheme on the disk enter the select option A list of supported disk partition types appears If the NetServer identifies a disk drive that has a partitioning scheme not specified in etc format dat format denotes this scheme Using the Format Command a 4 39 AUSPEX W as original adn where n is the disk number You can use this scheme to partition another drive of the same size as adn For more information on these partition types refer to Disk Partitioning on page 4 7 The following is the Select menu for a 4 GB disk drive partition gt select 0 Seagate 4GB 1 Seagate 4GB STRIPE 2 Seagate 4GB ROOT 3 Original ad4 Specify table enter its number 0 4 Select a disk partition type from the list displayed The partition table is created Specify table enter its number 0 2 partition gt 5 Use the label option to label the current disk partition gt label Ready to label disk continue y format gt q If you want to examine the partition table you have created enter print For further information on the print option see Chapter 6 of Sun s System and Network Administration 4 40 a Version 1 10 System Manager s Guide W AUSPEX Using ax_label to Partition and Label a Disk To save time use ax_label instead of format to partition and label a disk if you know the exact size of each physical par
195. e systems Then press Ctrl D to enter multiuser mode 6 24 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX If Other Disk Blocks Are Unreadable If an unrecovered media error occurs ona sector that is neither a disk label nor a file system superblock you see error messages from these sources besides the SP error message a The FP if the media error occurs in an LFS file system The following is an example of an FP message for an error on a virtual partition Nov 17 17 47 36 server_name FPO vp2 fatal error Permanent drive error on read block 9382416 The following is an example of an FP message for an error on a physical partition Nov 17 17 47 36 server_name FPO ad2c fatal error Permanent drive error on read block 9382416 fy Note The block number printed by the FP is relative to the beginning of the partition whether the error occurs in a virtual or physical partition That is the block number printed by the FP always matches the block number in the corresponding SP error message a The UNIX device driver if the media error occurs when the server tries to access a UFS file system In addition you may see an error message from the application program that tried to access the bad sector i Note If the SP error message indicates that the error has been recovered you can ignore the message unless it happens frequently Alternatively you can use the ax_sputil command to reassign t
196. e the procedure that requires single user mode and are ready to exit to multiuser mode run the fsck p command to check all file systems After all file systems are checked type Ctrl D to go to multiuser mode You do not have to reboot the NetServer to return to multiuser mode Booting From a CD ROM This section describes how to boot from the CD ROM in single user mode to perform system maintenance You cannot boot the server in multiuser mode from a CD ROM This section does not describe how to install the server software from the CD ROM for the 3 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX software installation procedure refer to Using NSinstall to Install Software From CD ROM on page 6 29 G Note When booting from a CD ROM the CD ROM must be in slot 1 Also if a root disk is installed in slot 0 the server uses partition b of the root disk as swap space when booting from a CD ROM If you keep the CD ROM in slot 1 after booting and the etc fstab file has dev ad1 entries for example entries for dev ad1a dev ad1c and so on for the disk that the CD ROM replaced remember to modify the dev ad1 entries Loading the CD Depending on the type of CD ROM drive follow the appropriate steps to load the CD If the CD ROM uses a CD caddy 1 4 Open the CD ROM caddy and place the CD in the caddy with the CD artwork facing out Close the CD caddy completely Gently push the caddy in
197. e vp5 and vp6 Each of the member partitions consists of a single physical disk partition ad3e for vp5 and ad3f for vp6 a Drives ad0 and ad1 both contain unmounted partitions indicated by rows of decimal points Although all this information is available from other sources such as fstab 5 vpartab 5 and dkinfo 8 no other utility makes it available in a single report 5 24 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Capacity in 1000 megabytes 0 1 2 3 la gqgdeeefFFFFFF FAM a ri fusr fusr openwin fvar crash var home leccccseccccccccesccccoeel lececc c fexport mfg leeeeeeeceeeceefFFFFFFFFFFF e vpdtypS fdisks usr local fF ss vpBrvp fdisks opt local ceCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC c fhome attic2 lecceecc c vpll 1 fhome docs Figure 5 11 Example ax_diskconf second table with p option Displaying Disk Configuration Information a 5 25 W AUSPEX Recovering From Disk or File System Failures About This Chapter This chapter describes server problems that cause a file system to be inaccessible It also suggests recovery procedures for making the file system available to clients again with minimum interruption of service This chapter also describes error recovery on a root disk using the NSinstall command The procedures provided in this chapter assume that you are familiar with basic file structure conc
198. echo 35 35 message responses generated a To show the routing table on an interface enter ax_netstat I interface Managing Network Interfaces a 10 17 AUSPEX W The r flag displays the routing table used on the specified interface In this command form I interface is mandatory The n option displays IP addresses in the output instead of names For more information about the routing table refer to the ax_netstat man page The following is an example of an ax_netstat r command and its output host0 gt ax_netstat r n I ael Routing tables Destination Gateway Flags Refs Use Interface T27 0 0 1 L260 30 4 UHO 0 0 144 48 216 0 144 48 252 65 UG 0 86 ael 92 42 160 0 144 48 254 1 UGO 0 70620 44 48 248 0 144 48 248 4 UO 0 0 44 48 217 0 144 48 252 65 UG 0 55 ael 44 48 153 0 144 48 252 126 UG 0 2 ael 44 48 9 0 144 48 9 1 UO 1 818999 44 48 218 0 144 48 252 65 UG 0 63 ael 44 48 170 0 144 48 252 37 UG 0 3600 ael 44 48 10 0 144 48 254 1 UGO 0 957 44 48 250 0 144 48 250 4 UO 0 37980 44 48 219 0 144 48 252 65 UG 0 60 ael 44 48 11 0 144 48 254 1 UGO 0 244 44 48 220 0 144 48 252 65 UG 0 61 ael a To show routing statistics not the routing table for a network processor board enter ax netstat r s I interface The following is an example of an ax_netstat r s command and its output host0O gt ax_netstat r s I ae3 routing bad routing redirects dynamically create
199. ecimal number variable length x a hexadecimal number variable length nx an n digit hexadecimal number in the actual error message n is a decimal number s a variable length string E 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table E 2 Auspex Processor Error Messages Error Priority code Message text 1001 1 cm_bread size 0 1002 1 cm_bread 1003 1 cm_breada 1004 1 cm_breadrabp 1005 1 cm_bwrite 1006 1 cm_getblk size too big 1007 1 cm_geteblk size too big 1008 1 cm_brealloc 1009 1 cm_alloc bad size 1010 3 file system full 1011 3 out of inodes 1012 3 bad block 1013 1 ialloc bad size 1014 1 realloccg bad size 1015 1 realloccg bad bprev 1016 3 s optimization changed from SPACE to TIME 1017 3 s optimization changed from TIME to SPACE 1018 1 realloccg bad optim 1019 1 ialloc dup alloc 1020 1 Out of memory in structure allocation 1021 1 Value in fc_conf bufhash must be a power of two 1022 1 Value in fc_conf cm_bufhash must be a power of two 1023 1 Value in fc_conf inodehash must be a power of two 1024 1 Value in fc_conf ndquothash must be a power of two 1025 1 Value in f _conf ncachehash must be a power of two 1026 1 User requested File Processor panic from console 1027 3 fc_read attempt to read from non file 1028 3 fc_readlink attempt to readlink from non link 1029 1 getdiskquota 1030 3 dquot is full 1031 3 inode table is full 1032 1 diskquota 1033 1 dqput 1034 1 dqupdate 103
200. ectories in export1 and another group in export2 Distributing client files among multiple disks in this way reduces disk contention Under each exportN set up a root file system for example on the a partition and a swap file system for example on the g partition Having separate file systems for swap and root allows you to back up root directories without backing up swap directories Analyze the client root and swap needs of your environment before deciding how to partition the disks on your server If the partitions defined at the factory do not meet your needs use ax_label or format to customize the disk partitions Disk Partitioning a 4 11 AUSPEX D Table 4 8 Example of disk partitioning among multiple disks Partition Mount point File system type Description adda Z 4 2 root ad0b swap swap ad0g usr 4 2 usr ad0d usr openwin lfs OpenWindows 3 0 adoOf var 4 2 spool and log files ad2a backuproot 4 2 copy of root partition ad2g backupusr 4 2 copy of usr partition ad2h home Ifs home directories ad4c home dvlp Ifs client data ad5c home rise lfs client data ad6c home src lfs client data ad7c export Ifs client exec share kvm vp1 for example a striped export root Ifs client root directories partition consisting of ad8a and ad9a vp2 for example a striped export swap lfs client swap space partition consisting of ad8g and ad9g ad9h home sysadmin Ifs client data ad10c home rcs Ifs client dat
201. ectory may be part of any mounted filesystem or may itself be a mount point In general it is unwise to select a base directory which already contains other files and or directories Now enter the path name to the destination directory for the package Enter path to package base directory q docs Using lt docs gt as the package base directory AXdocs requires the following disk space 11572 kbytes Checking if docs has enough free space for install Processing package information Processing system information Verifying disk space requirements Checking for conflicts with packages already installed t Checking for setuid setgid programs This package contains scripts which will be executed with super user permission during the process of installing this package Do you want to continue with the installation of this package y n Online Documentation a _ B 3 7 Answer y if you wish to continue Answer n if you want to find another source directory for the package Installing Auspex System Documentation as lt AXdocs gt Installing part 1 of 1 documents FSG 17_FRU PDF documents FSG 181_FRU PDF A long listing of files is deleted in this example but will appear during your installation verifying class lt none gt Executing postinstall script In s docs usr auspex docs Installation of lt AXdocs gt was successful The following packages are availab
202. ed filesystem dropping request to mount home host1 steve Mar 15 14 05 04 host1 ax_isolated 295 Filesystem dev vp77 mounted on disks home is isolated Disk problems When the FP tries to write to a disk that the SP cannot reach the FP also isolates the file system so you can correct the disk or media error The following is a list of errors caused by hardware A PSA_ERR_DRIVE_OFFLINE This error occurs when the disk is spun down and not on line A PSA_ERR_PERM This error is any disk error other than PSA_ERR_DRIVE_OFFLINE which occurs when the SP cannot send commands to the drive a Unrecovered media errors The FP does not isolate a file system if a media error is recovered If the error is not recovered however the FP isolates the file system so you can repair the hardware The following are sample messages in var adm messages caused by a permanent drive error ar 14 07 14 12 hostl FPO ad52c fatal error Permanent drive error on read block 1983184 ar 14 07 14 12 hostl FPO bp gt b_un b_addr 21e22000 bp gt b_attr 1 ar 14 07 14 12 hostl FPO isolating filesystem mounted on disks eng ar 14 07 15 21 hostl rquotad 368 rpc rquotad Skipping isolated filesystem disks eng ar 14 07 15 58 hostl ax_isolated 215 Filesystem dev ad52c mounted on disks eng is isolated Unrecoverable Disk or File System Problems in Multiuser Mode a 6 11 What Does the
203. ed histogram 3 Repeat the previous steps to remove other histograms Defining the Appearance of a Histogram This section describes how to define the appearance of an individual histogram and control the overall properties of the histograms in the Data Display Window Configuring the properties of a histogram You can customize a histogram by changing its size scale title and so on Although a single Data Display Window contains multiple histograms each histogram can assume different characteristics To define the appearance of a histogram click on the Properties button to display the pull down menu Select Each Histogram in the menu The Histogram Properties Window is displayed see Figure 9 14 In this window specify a which histogram to modify a the values for a set of fields that determine the appearance of the selected histogram Refer to Table 9 3 for information about the fields After modifying the parameter values in the Histogram Properties Window click on the Apply button The new parameter values take effect immediately 9 24 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX After modifying a parameter value click on the Reset button to undo the change The Reset button restores the value of the parameter if you have not applied the new value ke Histogram Properties Window niste 1 2 s a s s 7 8 Input Filename acdelco dat Color Black Total Samples 12 Firs
204. ed making sure the file is associated with the inode reported by fsck In this example the file with inode number 26 is cshrc so enter the following to add the file to the list of files to restore restore gt add cshrec cshrc added 4 Enter the extract command and specify the volume from which to extract the file restore gt extract Which volume 1 26 extracted The restore program uses the inode number as the name of the file extracted You can change the filename back to cshrc Restoring a File With a Damaged Inode a 8 17 Restoring a File from 1 5 1 produced Tapes This section describes how to restore from 1 5 1 produced dump tape Software Release 1 5 1 contains a bug that generates an additional End Of File EOF marker at the beginning of the tape and at the beginning of each file on the tape This bug was corrected in later releases but can create a compatibility problem for 1 5 1 produced tapes To perform a restore of a particular file system on a Release 1 5 1 produced dump tape advance the tape past the extra EOF mark at the beginning of each file as well as past the normally occurring EOF at the end of each file For example to reach the third file on the tape you must advance through five EOF markers with the following command mt fsf 5 This command advances the tape beyond the extra EOF markers to start the restore from the correct position At this point you can invoke the restore command For example
205. ed a member of a mirrored striped or concatenated virtual partition in Before You Remove a Drive on page 4 31 re create the member 4 34 a Version 1 10 System Manager s Guide W AUSPEX Attaching Drives to the Host Processor SCSI Port The SCSI port on the HP supports the standard SunOS devices listed in Table 4 9 not Auspex devices The system kernel by default allows you to attach up to seven Sun supported devices to the SCSI port which include four disk drives two tape drives and one CD ROM drive Note Devices attached to the HP SCSI port cannot be used as boot devices i Only Auspex devices attached to an SP are bootable You must set the appropriate SCSI ID for each device on the HP or the system kernel cannot recognize the device The SCSI IDs that you can use are listed in Table 4 9 Be sure each device has a unique ID Table 4 9 SCSI IDs and names for devices on the HP Device SCSI ID Device name Disk 0 3 sd0 Disk 1 1 sdi Disk 2 2 sd2 Disk 3 0 sd3 Tape 0 4 rstO Tape 1 5 rst CD ROM 0 6 sr0 Follow these steps to attach a SCSI device to the HP SCSI port 1 Shut down and halt the NetServer 2 Verify that the device being attached has the correct SCSI ID Refer to the device manufacturer s documentation for instructions on verifying and changing the SCSI ID 3 Attach the device to the port using the SCSI cable described in the hardware manual for your NetServer 4 Power on the NetServ
206. ed by ax_netstat are cumulative they do not include the per second rates as in the ax_perfmon screens Also ax_netstat can display routing tables and various networking statistics on a per protocol basis The ax_netstat commands have three forms depending on the command line s first flag a To show the statistics for a specified interface enter ax_netstat i n I interface interval The information displayed includes the MTU value number of packets received number of packets sent number of errors and number of collisions You can display only information about interfaces that were configured and detected at boot time The i flag displays the interface statistics The n option displays IP addresses in the output instead of names The I interface option specifies the interface for which statistics are displayed These are examples of interface names ae0 afddil If no interface is specified the command displays statistics gathered from all network interfaces in the system If the interface specified is not enabled the interface number in the screen output is followed by an asterisk for example ae0 Note When ax_netstat prints statistics for all interfaces it may not list them in numerical order but in the order they were detected when booting the system The interval option specified in number of seconds displays statistics gathered since boot time and incremental statistics for each time interval The statis
207. em is for a specific application unexporting it only aborts the application on clients that have mounted it Clients trying to mount it are denied the use of this application but they do not hang Use fsck to check the file system Add the file system back online with the following command ax_fsutil release wrld apps frame_3 0 Successfully released filesystem mounted on wrld apps frame_3 0 Re enable the write accelerator if one is used ax_write_cache s 0 enable If there is lost data restore it from the most recent backup Export the file system again so clients can mount it The file system recovery procedure is now complete Recovering from Permanent Disk Errors Without File System Isolation When the server is running in multiuser mode it is possible for a disk to become invisible without triggering file system isolation If the file system on the disk is managed by the HP or if the disk error is not one of those causing file system isolation replace the disk as soon as possible to prevent a system crash Replacing a Disk Containing Unmirrored Partitions If the disk experiencing problems contains file systems that are not mirrored follow these steps to replace the disk 1 Unexport and unmount the file systems on the inaccessible drive For example if ad3 is the inaccessible disk and the file system mounted on ad3c is home src enter the following commands exportfs u home src umount home sre Unrecoverab
208. emoving a Drive When the Operating System is Shut Down 4 33 Using ax_hot_plug to Add and Remove Multiple Drives 4 33 Attaching Drives to the Host Processor SCSI Port 00 see eee 4 35 Using the Format Cominatidi jc y 4m toca saute ee Os 1a ee ashe Ria te a 4 36 Running the Format Command in Single or Multiuser Mode 4 36 Accessing the Format Menu 43 0 eee eee REA 4 36 Defining a Disk Type ics pen ian nurru surrar ruru aie oo heat 4 37 Formatting a Disk Using the Format Option 055 4 38 Changing the Partitioning ona Disk 0 00 cece eee 4 39 Using ax_label to Partition and Label a Disk 0 000 000 0000 4 41 Displaying a Disk s Partitioning Information 00 000 eee 4 42 Chapter 5 Virtual Partitions About This Chapter a ta al as hi Goes Ok aie Aline ie haehe NOda paid PS grea Sees eah autho ecicy 5 1 Virtual Partition Overview s nirice ayate gud dial aaratiguntenarerd airmen pee a ae 5 2 Maximum Size of a Virtual Partition i655 nein Seok Si ae aa aetoeg anda er 5 3 Advantages of Using a Greater than 2 GB Virtual Partition 5 3 Different Types of Virtual Partitions sussuru ssa cece eee ee 5 4 Concatenated EATEN ONY aisha spatea cord aces dea gs rand ath da Sedo pane ants 5 4 Striped Partition peri Aig ad Soaps eai aa aed Sed ia eona ES gs te deh ord igh 5 4 Mirrored Partition 0 6 0 0 eee eee eens 5
209. epts for example the functions of inodes superblocks and so on Also familiarize yourself with virtual partitions by reading Chapter 5 carefully before recovering a disk that is part of a virtual partition Note Several of the procedures involve unmounting or unexporting file systems on the failed drives You can use the ax_diskconf command to get a report that maps all file systems to the physical partitions or virtual partitions Also some procedures require copying data from the failing drive to a spare drive Be sure the spare drive is at least the same size as the failing drive and both drives are formatted with the same formatting scheme Recovering From Disk or File System Failures a 6 1 When Does a File System Become Unavailable A file system may become unavailable for one of the following reasons a Hardware problem which can be a disk error or a media error Disk errors The entire disk becomes invisible to the system Error messages are generated by the SP vmunix or the FP when the system tries to read from or write to this disk Disk errors can be either permanent or intermittent Media errors Only a sector of a disk is damaged Error messages are generated by the SP or vmunix when the system tries to read from or write to the failing sector The sector number is reported in the SP or vmunix error message a File system problem which can be a metadata corruption within the file system The symptom of an una
210. er and boot it to multiuser mode During the boot process observe the messages printed on the console to verify the new device is detected by the system The following sample messages show a tape device attached to the HP esp0 at SBus slot 1 0x800000 pri 3 st0 at esp0 target 4 lun 0 st0 lt Archive QIC 150 gt Before you remove a device from the HP SCSI port be sure to check that the server is halted Attaching Drives to the Host Processor SCSI Port a 4 35 AUSPEX D Using the Format Command This section explains how to use the format command which is a menu driven command for setting up your disks Under normal circumstances such as when installing a new system you do not need to use format Auspex disks are shipped formatted and labeled with the default partition tables You generally need to use format only if a disk is corrupted or if you want to change the drive partitioning Also if you need to partition a drive Auspex recommends using the ax_label command which is a more straightforward command for achieving the same purpose as partition and label in the Format menu For more information on ax_label refer to Using ax_label to Partition and Label a Disk on page 4 42 Caution Low level formatting of a disk is normally not necessary Drives are A preformatted at the factory before shipment When formatting a drive use extreme care as existing data can be erased from the disk Be sure to back up any data on the dis
211. er for FP statistics a sp_filter for SP statistics a vpar_filter for virtual partition statistics Use a text editor to view each of these sample files to see if they fit your needs You can create your own filter files by deleting or adding parameters from an existing filter file For example if you want to capture data about both SP and NP statistics you can create anew filter file that concatenates sp_filter and np_filter Types of Statistics Displayed by ax_perfhist on page 9 40 lists the filter file parameters and definitions for ax_perfhist and ax_perfmon Look at the file usr auspex screens std to see which filter file parameter displays in each field of the statistics screen Specifying a Filter File on the ax_perfmon Command Line The following example shows an ax_perfmon command with a filter file named npOstats and an output file named out ax_perfmon o out f npOstats In this example the command creates an output file named out which contains statistics related to NPO in decimal format You can use this file as input for ax_perfhist in which 9 16 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX case you can only display histograms for the parameters defined in the filter file For more information on ax_perfhist refer to the next section applicable to the configuration of your server ax_perfmon displays a list of messages indicating some parameters are not found before displaying the
212. er at the same time Run fsck on the cloned partitions The server prompt appears when the data copy procedure is finished You can now dump the contents of the clone partition to tape If you plan to do incremental backups on the partition we recommend you issue the dump command on the same command line as ax_clonefs Refer to Dumping a Clone Partition to Tape for information on using ax_clonefs and dump Caution Never use the same clone partition for backing up different source A partitions simultaneously For example if you execute ax_clonefs vp3 vp2 before ax_clonefs vp1 vp is finished the second ax_clonefs generates an error message and the data backed up in the first ax_clonefs may be corrupted Dumping a Clone Partition to Tape If you use ax_clonefs to back up a file system online remember the following regarding the subsequent dump command that copies the clone partition to tape a Always issue the dump command immediately after ax_clonefs unless you use dump with the T option see the next bullet item For example ax_clonefs adic ad2c dump Odsbfu 141000 11500 126 dev rast4c dev ad2c In this example the date string that dump inserts in etc dumpdates is also the time when ax_clonefs is finished When you do an incremental dump next time all changes after the snapshot was taken are backed up A time gap between ax_clonefs and dump causes a loss of some changes at the next incremental dump Normally you do n
213. es S Avoid calling sync 2 so that reboot does not hang if the machine from which the NetServer has mounted a file system cannot be reached File systems are still synced during the multiuser reboot p processor If processor is the HP then reboot only the HP A boot argument is a string passed as an argument to the boot command in the PROM monitor The string must be preceded by an option terminator string C Perform a cold boot If you do not specify c the system performs a WarmStart which skips several POST tests and speeds up the time required to boot This option does not modify WarmStart s default enabling Subsequent restarts skip POST tests unless the c argument is included in the command Refer to Using the Boot Command on page 3 5 for more information on the arguments to boot command fasthalt and fastboot The fasthalt and fastboot scripts perform the same functions as the halt and reboot commands respectively The only difference is that fasthalt and fastboot disable testing of the system cache I O memory upon reboot Use reboot if you suspect the system is experiencing problems with system cache I O memory The following are the fasthalt and fastboot syntaxes usr etc fasthalt halt options usr etc fastboot boot options The fasthalt options are the same as those used for the halt command The fastboot options are the same as those used for the reboot command Commands for Shuttin
214. es Because the mail message contains the shutdown message Auspex Technical Support can more easily diagnose your server problems halt The halt command stops the operating system more quickly than shutdown but its syntax does not allow you to send out a message to system users warning them of the impending loss of service Use halt only when the system is currently in single user mode or when you are sure that no users are logged in to the system or have file systems mounted from the system The halt command brings the system down to monitor mode without delay The following is the halt command syntax usr etc halt dnqs p processor The options are as follows d Dump system core before halting n Avoid running etc rc shutdown and executing sync rc shutdown is an Auspex script that attempts to unmount all the mounted LFS and UFS file systems except usr and Because this option avoids unmounting the file systems if an LFS file system was active before the reboot it is not marked clean and is checked by fsck at reboot This option is used during emergencies q Reboot the system quickly and ungracefully without first shutting down the running processes S Avoid calling sync 2 at reboot so reboot does not hang if the machine from which the NetServer has mounted a file system cannot be reached File systems are still synced during the multiuser reboot p processor Shut down only the named processor if processo
215. es and programs added or modified in SunOS see the auspex 8 man page or Appendix A To see a printed version of the man pages refer to the Command Reference Guide For each of the system management procedures described in this guide the Sun Microsystems documentation can provide additional information Table 1 3 lists basic system management topics and indicates which chapter to consult in Sun s System and Network Administration manual Table 1 3 Pointers to Sun documentation Topic Sun chapter number System administrator s role Overview of SunOS Booting and shutdown File maintenance System backup and restore Maintaining disks with format Monitoring file system usage Administering security Reconfiguring the system kernel System accounting fsck Adding a modem Line printer Modifying termcap System crash and accompanying messages Description of networking concepts The Sun network environment 1 16 A Version 1 10 System Manager s Guide 1 N DO DO oO oO ff N E ke ae a a a EES N N O fF WD WO YO O AUSPEX W W AUSPEX Table 1 3 Pointers to Sun documentation Continued Topic Sun chapter number NFS 18 The Name Information Service NIS 20 Domain Name Service 21 Electronic mail and communications 23 UUCP 22 sendmail 23 Monitor and boot program error messages Appendix E SunOS Enhancements A 1 17 W AUSPEX Software Configuration About This Chapter This chap
216. estats on a busy device the message Device busy appears Using ax_hot_plug to Remove a Drive Run ax_hot_plug only on the local console If you try to run it from a remote login session the following error message appears ax_hot_plug must be on dev console to use this command Caution Never run ax_hot_plug on an SP that is formatting a drive or A running ax_mrestore or ax_clonefs Doing so may hang the system After completing the preceding procedure follow these steps to remove a drive from the server 1 Enter the ax_hot_plug command specifying the slot number of the drive to be removed For example if you wish to remove the drive in slot 6 enter the following ax_hot_plug remove 6 If a tape drive on the same SP is active the server displays the following error message ax_hot_plug a tape drive is in use on the same SP as the device to be removed cannot remove device This message ensures that no active tape drives are rewound because of the power surge in the drive rack when a drive is removed If the message appears invoke ax_hot_plug again after verifying that no tape drive activity is taking place on the same SP The system suspends the activity on the SP so no I O operations in progress on the SCSI bus spins down the drive prompts you to remove the specified drive and tells you to type a carriage return when you finish from the NetServer s NPs are halted until the SP restarts This may cause timeouts o
217. etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 1 Ifa mirrored partition has a damaged member the following message is sent to var adm messages Warning a mirrored partition has a dirty or damaged component vp Note You can also use ax_vpstat to determine which mirrored partition has a bad component 2 Run ax_mdetach to detach the bad disk from the mirrored partition This step is required for each mirrored partition that has a member on the damaged disk For example if the mirrored partition is vp3 and the member partition on the bad disk is vp4 enter the following ax_mdetach vp3 vp4 6 20 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX This change is automatically recorded in etc vpartab ax_mdetach appends a comment to the end of the file for the edited entry For example ax_mdetach vp4 removed from vp3 on Fri May 5 22 18 50 1995 The mirrored partition now with one member never stops performing read and write operations even when a member of the virtual partition is being detached Edit etc vpartab to comment out any remaining references to the virtual partition or the failed disk drive The following exam
218. etc vpartab each time you modify the per file If one of your NetServer drives fails having a hard copy record of the virtual partitions and their members that reside on the failed drive makes it easier to recover from the failure Defining a Virtual Partition Use this procedure to create a new virtual partition Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 1 If there is an active file system on the physical disk partition you are going to define as a virtual partition unexport and unmount the partition to prevent access Change the etc fstab and etc exports files as necessary 2 Back up the file system to tape or to another drive If necessary use format to repartition the disks 4 Edit etc vpartab to add an entry for the new virtual partition Refer to Table 5 1 on page 5 10 for the list of valid partition numbers and corresponding SPs 5 Run ax_loadvpar If ax_loadvpar detects errors in etc vpartab the command reports the errors and does not load the new table If this happens edit etc vpartab again to correct the errors and run ax_loadvpar again 6 Run newfs 8
219. etically takes up the whole disk Therefore a copy operation of a c partition from a single partition disk to a multipartition disk of the same size fails Restrictions Applicable to Concatenated and Striped Partitions A The disk partitions in a concatenated partition need not be of equal size However the members of a striped partition must be the same size Refer to Disk Partitioning on page 4 7 for more information You can specify the size of a stripe in a striped partition but the size you choose must be a multiple of 8 KB 8192 bytes up to a maximum size equal to the partition size The recommended size of a stripe is 128 KB The size of each member of a striped partition should be a multiple of the stripe size if not the partition size is truncated automatically to the nearest multiple of the stripe size For example in Figure 5 4 part of ad7g ad8g ad9g and ad10c is truncated because the length of each of these physical partitions is not a multiple of the stripe size Virtual Partition Restrictions a 5 7 Striped partition vp2 ad7g ad8g ad9g ad10c Truncated because length of the partition is not a multiple of the stripe size Figure 5 4 Disk space truncated from members of a striped partition If a member of a virtual partition starts at the beginning of a disk that is if the member is partition a c d or e the first 8 KB of disk space is not used as part of the virtual partition to preserve t
220. evice connected 0 no device connected logins disabled 1 dial out modem logins disabled 2 dial in modem logins enabled 3 local terminal logins enabled 4 serial printer logins disabled 3 Complete the fields in the form and execute it by typing Ctrl F The command then returns you to the system prompt This completes the procedure for setting up the serial ports NetServer Configuration Commands a 2 19 SetupExec Figure 2 5 is a flowchart describing the procedure for configuring client information on the server using SetupExec No Will server boot diskless clients Do nothing Do you have the Sun architecture dependent executables either on CD or tape Purchase executables from Sun Enough disk space for executables in export Move export to a larger partition SUNBIN tape CD on No on local host remote host Yes Verify server is in remote host s rhost file with root permission No on remote tape SUNBIN on remote CD Yes Mount CD ROM Run SetupExec in multiuser mode Figure 2 5 Configure client information on the server using SetupExec Do You Need to Run SetupExec SetupExec is required only if your installation includes diskless SunOS client workstations that boot from the NetServer Client workstati
221. exec kvm ARCH sunos VERS____ Path to usr share files CtEXPORT share sunos VERS Distribution medium CDROM Tape gt CDROM Drive location Local Remote gt gt Local CD ROM device name Cacdi_ e 9 acdl acd2 acd209 Mount point of CD ROM cedrom Tse TAB CR or arrow keys to move among fields when done hit CTRL F to execute HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VAL Figure 2 9 Sample SetupExec form for a remote CD ROM device 2 24 a Version 1 10 System Manager s Guide W AUSPEX Table 2 8 The SetupExec form fields Field Description Architecture type SunOS version Path to executables Path to kernel executables Path to usr share files Distribution medium Drive location Tape device name or CD ROM device name Media host IP Addr Mount point of CD ROM Architecture to install for example sun4 aushp For further explanation of application architecture see the arch 1 man page or Table 2 7 The SunOS version to install for example 4 1 3 4 1 3_U1 4 1 4 Path name of the directory in which to install the application executables When SetupExec is complete mount this directory as usr by the diskless SunOS clients of architecture specified by Architecture type Path name of the directory in which to install the kernel specific executables When SetupExec is complete mount the path to executables as us
222. f 1 2 hardware implementation 1 9 meaning of 1 5 Font files for X terminals 4 16 Form configuration 2 9 NSconfig 2 13 NSinstall 6 31 SetupClient 2 29 SetupExec remote devices 2 24 SetupTty 2 19 format command difference from SunOS 1 13 format option 4 38 invoking 4 36 label option 6 24 repairing adrive 6 26 repartitioning adisk 4 7 verifying if disk label exists 6 5 virtual partition 5 7 Formats date and time 2 17 Formatting adisk 4 38 Formatting time 4 39 FP Statistics screen 9 10 9 12 Free disk space on the server 5 3 fsck command 1 14 3 8 4 22 6 26 10 10 FTP on the Network Processor 1 8 Full backup 8 2 G getcores sh command 1 15 go goto command 10 11 Greater than 2 GB virtual partitions 5 3 H halt command 1 15 3 2 3 10 10 12 Hardware subsystems of the NetServer 1 5 Hardware upgrade tracking 10 5 Hayes compatible modem 2 18 HELP function in configuration forms 2 10 High Sierra File System HSFS 4 5 4 18 Histogram 2stepload 9 24 clearing 9 22 continuous plot 9 28 defining its appearance 9 24 displaying average statistics 9 29 printing 9 38 Properties Window 9 25 to 9 27 removing 9 24 selecting samples 9 32 server data 9 18 summing parameters 9 35 Home 2 10 Home directories for clients 2 30 Host information worksheet D 2 Host name alias 2 14 server 2 11 2 14 Host Processor functions of 1 5 mounting file systems on 4 18 PROM monitor 3 3 SCSI port 4 35 Host Processor VII HP VII architec
223. f the directory in which the kernel specific executables reside This is the directory that the client mounts as usr kvm Path name of the directory in which the architecture independent operating system files reside This is the directory that the client mounts as usr share Path name of the directory in which various home directories reside It is the path name of the directory that the client mounts as home This field is optional if omitted no home mount point is created in the client s etc fstab 5 Fill in all fields of the SetupClient form and execute it by typing Ctrl F The following messages appear updated ethers pushed ethers updated hosts pushed hosts Creating root for client CLIENTNAME Creating swap for client CLIENTNAME Updating etc exports to export CLIENTNAME info Updating etc bootparams updated bootparams pushed bootparams Completed creating ARCH client CLIENTNAME Now reboot the client machine 2 30 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Root Login and File System Isolation When a file system managed by the FP an LFS mounted file system becomes corrupted the FP isolates the file system so you can repair it However you cannot log in to the server as root after a file system is isolated if the root s login or any shell startup file contains references to that file system For example if usr local is LFS mounted and u
224. fer to Sun s System amp Network Administration manual Note If you experience problems with rebuilding the kernel contact Auspex D Technical Support Dealing with Server Problems a 10 9 System Panics and Crashes The NetServer can hang or crash causing a failure to respond to commands This section discusses what happens when the NetServer encounters a system problem contacting Auspex Technical Support before rebooting In some cases the system is not hung and Technical Support can bring back the system without rebooting fy Note When a NetServer hangs with no response to commands consider When the NetServer crashes it performs the following steps 4 Attempts to execute a sync to force changed blocks to be written to the disks 4 Prints a short message describing the cause of the crash The message is in this format panic error message a Writes all kernel panic messages to usr adm messages If you cannot obtain the crash error message on the system console look in this file for the information You can also retain crash messages in this file in order to keep a record of system behavior over time a Attempts to save a crash dump by writing a core image of memory from each processor into the primary swap partition on disk After reboot the savecore program writes this image to the directory var crash hostname Upon a system crash an NP produces two core dumps one for each CPU on the board For NP III an NP IV
225. fied suffix when you click on Scan Dir Another way to start scanning the directory is to press Return after typing the filename extension In Figure 9 11 the file list displays four filenames With a long list use the scroll bar to move up and down the list Click on the file name you want to load The filename appears in the Data File field 4 Each data file contains more information than one histogram can show You must select one type of statistics to be displayed in a histogram To select a parameter hold down the right mouse button when pointing at the Parameter button to display a pull down menu that lists the statistics types The pull down menu lists all the processor boards Each board has two layers of submenus For further information on selecting parameters refer to Types of Statistics Displayed by ax_perfhist on page 9 40 Displaying Performance Data in Histograms a 9 21 Note If the data file contains only filtered statistics you can choose only from the parameters included in the data file The pull down Parameters menu displays the parameters without being categorized by boards By default the statistics you selected appear immediately in the Data Display Window If you do not want the statistics to appear immediately and prefer to display them after all the histograms have been defined follow the procedure in Loading Statistics Using 2 Step Load on page 9 24 oz Recommendation Displaying both the
226. figuring a New Disk Drive 1 Determine your partition scheme refer to Disk Partitioning on page 4 7 Also plan ahead for whatever virtual partitioning scheme you use The configuration worksheets in Appendix D may be helpful as you gather disk drive information Note Disk drives are labeled at the factory so a partition table is already i assigned to each disk Unless you are changing the standard partition configuration you only need to install a UNIX file system using the newfs 8 command as described in step 3 2 Check to see if the disk drive label is correct all disk drives are shipped already labeled To do this enter the following command dkinfo adn where n is the disk drive number This command displays the disk label which lists the size of the usable UNIX partitions on the disk For a 4 GB disk in slot 8 the dkinfo command and the resulting display look like this 4 28 a Version 1 10 System Manager s Guide W AUSPEX dkinfo ad8 ad8 Auspex SP at addr 1180 unit 8 4094 cylinders 16 heads 128 sectors track a 2095104 sectors 1023 cyls 0 1022 b 2095104 sectors 1023 cyls 1023 2045 c 8384512 sectors 4094 cyls 0 4093 d No such device or address e 4190208 sectors 2046 cyls 0 2045 f 4190208 sectors 2046 cyls 2046 4091 g 2095104 sectors 1023 cyls 2046 3068 h 2095104 sectors 1023 cyls 3069 4091 Capacity in 100 megabytes 0 5 10 15 20 25 30 35 40 c
227. g Down the NetServer a 3 11 W AUSPEX NetServer Storage Devices and File Systems About This Chapter This chapter discusses various features of the NetServer storage subsystems In addition it describes a file system type unique to Auspex the Local File System LFS This file system type is a key feature of the NetServer software architecture in which the UNIX operating system is separated from the file system The following lists the major topics covered a The Auspex device naming conventions a Default disk partitions a Distribution of file systems a LFS a Adding new drives a Removing drives a Formatting drives Version 1 5 1 If you are upgrading your server from a pre 1 5 1 version be sure to read the Tape Drives section on page 4 2 fy Note The naming convention for 4 mm tape drives changed in software NetServer Storage Devices and FileSystems a 4 1 AUSPEX D Device Names This section describes the conventions for assigning device names to Auspex disk CD ROM and tape drives Disk Drives Disk drive devices are named adn where n corresponds to the drive slot The maximum number of slots depends on your NetServer model DriveGuard arrays are named ardn See the DriveGuard Manager s Guide for array numbering information CD ROM Drives How you refer to a CD ROM drive depends on where the device name is used Table 4 1 describes the name for the CD ROM drive in various commands The let
228. g file systems to the current root drive boot the system from the CD ROM as described in Booting the server from the CD on page 3 7 If you are installing file systems to a spare disk you can run NSinstall in multiuser mode For example you can invoke NSinstall to install file systems from a CD ROM in slot 5 by entering the following command NSinstall cdrom 5 Note If you want to run NSinstall in multiuser mode from a root disk running a system software release earlier than Version 1 7 first mount the disk before running the installation program After entering the NSinstall command the following message appears AUSPEX CD ROM MAINTENANCE MENU OPTION 1 Format label the root drive 2 NSinstall 3 Maintenance shell Select an option 1 2 3 The target drive for the NSinstall must have the correct root partitioning scheme which was updated with Release 1 7 If your target drive has not been properly formatted select Using NSinstall to Install Software From CD ROM a 6 29 1 from the Maintenance menu and follow the instructions Refer to Using the Format Command on page 4 36 for additional information If your target drive is properly formatted select 2 from the Maintenance menu to begin NSinstall If the TERM variable for your terminal is not set the system displays a numbered list of terminal types and prompts you to enter the terminal type Select the number from the menu that corresponds t
229. g from a media error in various situations After repairing the bad sector add the file system back online if a media error caused file system isolation Unrecoverable Disk or File System Problems in Multiuser Mode a 6 23 If Sector 0 Disk Label Is Unreadable Sector 0 which is the disk label is the most important disk sector because it contains drive partitioning information If the label is missing on a disk the server cannot check the file system on this disk during reboot nor can it reboot to multiuser mode To make the disk label readable again follow these steps 1 If the SP does not generate an error message the label probably has been erased accidentally Follow these steps to relabel the disk a Invoke the format command If the disk is not listed by format go to step b If the disk is listed use the backup command under format to recover the spare disk label and copy it to the primary label If this problem happens more than once reassign sector 0 as described in step 2 b If the disk is not listed by format that means the disk cannot be accessed by the system at all Go to Replacing a Drive in Single User Mode on page 6 5 for more information about replacing an inaccessible disk 2 Ifan SP error message indicates that a media error occurs in sector 0 use the ax_sputil command to reassign the label to another location For example if sector 0 of disk 2 is bad enter the following command ax_sputil reassign
230. go to step 3 If a message indicates that the state of the write cache is still DIRTY you may have a disk problem that prevents some data from being flushed Go to Recovering From a Disk Media Error DIRTY State Re enable the write cache by entering the following ax_write_cache s 0 enable Recovering From a Disk Media Error DIRTY State If the write cache state is DIRTY you may have a malfunctioning disk that prevents the data from being written This section provides an example for recovering a disk media error through the ax_write_cache command on SP0 1 If the ax_write_cache flush command returns a message indicating the write cache is still dirty look for the SP error message logged in the var adm messages file that reports the disk error Use ax_sputil reassign to reassign the sector that caused the problem and then try to flush the data again If the flush succeeds go to step 2 If it fails and you want to continue using the write cache follow these steps a Enter the following command to purge all data from the cache ax_write_cache s 0 purge b If you cannot flush data to disk it is likely that the disk is bad You should replace the disk use newfs to create a file system on the new disk and restore the contents of the file system from the backup tape Then go to the next step Re enable the write cache by entering the following ax_write_cache s 0 enable Recovering From a Disk Hardware Error
231. guration Worksheets a D 1 Host Information Worksheet Network Information Primary Hostname NIS Type Master Slave Client None NIS Domain SNMP Daemon Yes No Route Daemon Yes No Quiet Time Zone Host Information Host name Internet Address Subnet Mask Ethernet Address S CO Gl al a B wl YN BRIO ra oO j fai ja N jak w j a ol ja a ja N j ee ry O N ae N N N 69 D 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Notes Host Information Worksheet a D 3 Client Information Worksheet Server Information Hostname Client Information NIS Domain Client Name Arch SunOS Swap Addresses Names and Paths Version Size Ethernet addr Notes Internet addr Domain name Export path Root path Swap path Executables path Kernel executables path Home path usr share files path Ethernet addr Notes Internet addr Domain name Export path Root path Swap path Executables path Kernel executables path Home path usr share files path D 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Client Name SunOS Swap Addresses Names and Paths Arch Version Size Ethernet addr Notes Internet addr
232. h 8 growfs 8 halt 8 hpboot 8S hphalt 8 hpreboot 8 hpshutdown 8 icheck 8 ifconfig 8c inetd 8c installboot 8S looptest 8c mailmessages 8 mailmessages sh 8 mkfs 8 mount 8 mountd 8 netstat 8c Control the SP write acceleration feature Start the system kernel or a standalone program Build system configuration files Gather information from Auspex core dumps Generate disk accounting data by user Report information about disk geometry and partitioning Collect system diagnostic messages to form error log Incremental file system dump Dump file system information EEPROM display and load utility Export and unexport directories to NFS clients Reboot or halt the system without checking disks Sets the delayed metadata write capability of local file systems File system consistency check and interactive repair Compress and tar core files to tape Enlarge a file system Stop the Host Processor Start the system kernel or a standalone program Stop the processor Restart the operating system Close down the system at a given time File system storage consistency check Configures network interface parameters Internet services daemon Install bootblocks in a disk partition Tests an ATM interface for cell reception and transmission Inform Auspex Customer Service of a reboot Construct a file system Mount and unmount file systems NFS mount request server Show network status for Auspex Host Processor
233. has started Auspex recommends you record the correspondence between each pair of source and clone partitions to avoid accidentally mounting a clone partition a Cloning a partition that spans other partitions such as c on a root drive is not supported Online Backup a 8 13 Read and write operations Write operations only vp6 source partition is a mirrored partition vp65 clone of vp6 is a concatenated partition Copy Dump to e tape Concatenated partition Concatenated partition J j iA vp4 vp5 ad30a ad31b ad32b ad33c ad34a y y y y Physical partitions adic ad2c ad3c ad4c ad5c ad 6c Physical partitions Figure 8 2 Example of a clone of a mirrored partition Procedure for Cloning a Partition This section describes cloning a partition You can use the procedure to back up the entire root drive to another drive Before you start use the dkinfo command to verify that each partition on the target drive is at least the same size as each partition on the root drive Follow these steps to clone a partition 1 The clone partition must be at least the same size as the source partition If you need a larger clone partition create the clone as a concatenated or striped virtual partition The clone and source partitions can be on different SPs Refer to Defining and Reconfiguring Virtual Partitions on page 5 14 for information on creating a virtual partition 2 Attach the clone partition to the source partiti
234. hat run the Routing Information Protocol RIP If set to Quiet the route daemon still receives broadcasts from other machines but does not forward routing information from the NetServer If set to NO the route daemon is disabled See routed 8C for more details Yes or NO default Table 2 4 on page 2 17 lists the various date and time formats that are accepted in the NSconfig form Yes or NO default Refer to Appendix C for a list of time zones When entering the time zone name type the initial letter of the name to display the first time zone starting with that letter Pressing the space bar displays the next value Yes or NO If set to Yes the NetServer sends an email message to Auspex upon reboot Auspex recommends that you set it to Yes if your site can communicate with Auspex using email AUSPEX QD W AUSPEX Table 2 3 The NSconfig form fields Continued Field Description Possible values Customer Site The name of your site and the NetServer serial number If specified this information appears in the mail message your server sends to Auspex upon reboot Table 2 4 Date and time formats Date format Time format dd mm yy hh am pm dd mm yyy hh mm am pm dd mm yyyy hh mm dd mm yyyy hh mm am pm dd mm yy hh mm ss am pm month dd yyyy dd month yyyy hh mm ss hh mm ss am pm hh mm ss name service NSconfig does not configure the NetServer to be a client a fy Note Alth
235. he file systems use the v verbose option in the mount command The following is an example of mount av and its output mount av mount mounting export on FPO mount mounting home on FP1 mount mounting home export on FPO You can also see the FP assignment after you have mounted the file systems Simply enter mount without any options The following is an example of the mount command and its output mount dev ad3e on export type lfs fs 4 2 fp 0 dev ad3h on home type lfs fs 4 2 fp 1 dev ad0e on home export type lfs fs 4 2 fp 0 Mounting File Systems on the FP and HP a 4 21 AUSPEX D Enabling Write Acceleration in etc fstab Another Auspex modification to etc fstab is related to write acceleration To enable write acceleration for a file system modify the entry for the file system in etc fstab by adding we The following is an example showing the we option for the home dulp file system dev ad30c home dvlp lfs we fs 4 2 0 2 For more information on write acceleration and its requirements refer to Chapter 7 Other Commands Affected by LFS In addition to the LFS file system type and the fp n option to etc fstab an Ifs option has been added to the t argument of mount 8 df 8 and umount 8 to specify the LFS file system type File Processor Constraints The following list shows File Processor constraints a Each FP supports a maximum of 256 mounted LFS file systems a If the number of
236. he NetServer s software which is based on the Sun Microsystems SunOS 4 1 4 offers Auspex specific system administration commands for minimizing system down time Procedures such as drive replacement and file system backup can be completed while the server is in multiuser mode All models of NetServers have the features described in Table 1 1 1 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table 1 1 NetServer features System architecture a Functional Multi processing FMP architecture with separate dedicated processors for network file UNIX and storage processing a Host Processor HP VII and HP VIII which incorporates a Ross Technology HyperSPARC Mbus module running at 125 MHz The HP offers full compatibility with SPARC Compliance Definition 1 1 a WarmStart Upon booting the system automatically bypasses certain diagnostic tests reducing the time required to return to service a 100Base T Ethernet full duplex software is an integral part of the system software ready for configuration and use with no additional installation after system software installation 100Base T Ethernet runs in full duplex mode with the appropriate adapter card 4 Optional nonvolatile write cache for improved Network File System NFS write operation performance a Optional ServerGuard software which allows mirroring of an NFS filesystem across a pair of NetServers One NetServer acts as the primary server and the ot
237. he application that initiated the write request However if the write cache fails before it can write the data to both members of the partition there is no way for you to retrieve the data from the write cache to update the mirrored partition Therefore if you want to increase data reliability of a file system mount it on a mirrored partition without using the write cache Figure 5 3 illustrates a mirrored partition vp3 whose two members are concatenated partitions Each of the concatenated partitions consists of one physical partition Different Types of Virtual Partitions a 5 5 However the member of a mirrored partition can be a striped partition and it can be made up of several physical partitions In this example if ad9 fails the server can continue to write to and read from the file system on vp3 using ad8 Mirrored partition vp3 Read from member with least seek time Write to both members Concatenated partition vp4 Concatenated partition vp5 ad8c ad9c Physical partitions Figure 5 3 Example of a mirrored partition 5 6 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Virtual Partition Restrictions When you set up a virtual partition remember the restrictions described in the following sections Restrictions Common to All Kinds of Virtual Partitions A A A The root and usr partitions cannot be members of any virtual partitions A physical disk partition ma
238. he directory var crash hostname according to the savecore procedure described previously Message Logging Auspex s error message handling resembles that of SunOS in most respects The major difference is that messages are recorded from the multiple system processors instead of from a single processor The savecore program logs messages from the processors see the preceding section Some processor error messages are logged by the daemon syslogd to the file var adm messages according to the configuration parameters in the syslog conf configuration file see the syslog conf 5 man page All kernel panic messages are logged to the file Appendix E contains a list of Auspex specific messages including those related to the boot function and those related to the Auspex processors Note Each error message includes the name of the Auspex processor that i reports the error For example the following message is generated by the third SP in the system SP2 1203 Recovered drive write fault slot 44 If the File Processor on an NP generates the error the processor is identified as TOPn FPn CPU B if the network processor on an NP generates the error the processor is identified as IOPn NPn CPU A In both cases n is the number of the NP board For example in the following messages the first message is generated by the File Processor on the first NP board and the second one is generated by the Network Processor on the second NP b
239. he disk label As a result when deciding whether the size of this kind of member partition is a multiple of the stripe size remember to consider the 8 KB of disk space For example if the stripe size is 32 KB and the member partition ad1a is 128 KB the virtual partition uses only 120 KB on ad1a and 24 KB of disk space is truncated to make the size of ad1a divisible by 32K The data in a concatenated or striped partition may be spread over a maximum of 16 disk partitions Each member of a concatenated or striped partition must be a valid physical disk partition A striped partition must have two or more members a concatenated partition with only one member is permissible Restrictions Applicable to Mirrored Partitions A Each member of a mirrored partition must be a valid virtual partition either concatenated or striped a member cannot be a physical disk partition or another mirrored virtual partition A mirrored partition may not have more than two members A virtual partition may not be a member of more than one mirrored partition The members of a mirrored partition should be of equal size If you create a mirrored partition with members of unequal size the portion of the larger member that exceeds the size of the smaller member is ignored For example if one member contains 50 MB and the other contains 52 MB only the first 50 MB of the larger member will be recognized the last 2 MB is never used Each member of a mirrored
240. he file system available for export 12 Use exportfs to export the file system so it is accessible to users Expanding a Concatenated or Striped Partition Through the ax_expand command you can enlarge an LFS file system on a concatenated or striped partition without unexporting or unmounting the file system Network clients using the file system on the virtual partition may not even know that the file system is being expanded Users might however notice a pause or an NFS server not responding message while trying to access the file system during the expansion To expand a concatenated partition use this syntax ax_expand virtual_partition physical_partition To expand a striped partition use this syntax ax_expand s virtual_partition new_virtual_partition All the partitions involved in one ax_expand command must be controlled by the same SP You must be superuser to invoke ax_expand The following sections describe how to expand a file system on a virtual partition Gf Note A command related to ax_expand named growfs enlarges a file system to a specified size Because ax_expand is a front end to growfs always use ax_expand to increase the size of a file system on a virtual partition Defining and Reconfiguring Virtual Partitions a 5 15 Expanding a Concatenated Partition The procedure for expanding a concatenated partition involves adding one or more physical partitions to the virtual partition The following example sh
241. he sector to a new location For example if the SP reports a media error on sector 2718 of disk 5 enter the following ax_sputil reassign 5 2718 Always note the result of the reassign operation as a success or a failure If the error occurred in a sector other than the disk label or a file system superblock the recovery procedure is as follows 1 a FY N Use ax_sputil reassign to reassign the bad block to a new location If the reassign operation is successful the procedure is complete If not go to the next step Unexport and unmount the file system affected by the bad block Run fsck on the file system Mount and export the file system again Restore from the most recent backup tapes any files that were erased by fsck Unrecoverable Disk or File System Problems in Multiuser Mode a 6 25 Recovering a Damaged Root Disk This section outlines procedures for recovering from a failure on the root disk The procedures range in complexity from booting from a backup root disk the easiest and quickest means of recovery to reinstalling the operating system from CD ROM The choice of procedure depends on whether you have a backup root disk or a recent level 0 backup of the root partitions on tape and on the nature of the root disk damage Auspex provides the NSinstall command for installing the operating system from the distribution CD ROM to a formatted disk in any slot However installing the operating system could be a time cons
242. her which may be physically remote acts as the secondary server With ServerGuard the NFS filesystem is independent of local power failures and natural disasters resulting in uninterrupted NFS service to clients a Optional DriveGuard software RAID 5 implementation which protects user data in the event of a single device failure If any single disk in a DriveGuard array fails the data can be reconstructed using information on the remaining disks in the array See the Auspex DriveGuard documentation on the Auspex Premier Software Series CD ROM for more information a Optional DataGuard software allowing the HP to reboot without affecting other system components or NFS service to clients This feature improves overall reliability of the NetServer by isolating any failures to the HP and services directly provided by it a Optional EtherChannel combines multiple 100Base T Ethernet interfaces into a single channel All the interfaces in the EtherChannel load balance the traffic among themselves giving the concept of one fat pipe a Optional NP ATM Release 2 software is Auspex s Network Processor NP resident implementation of Asynchronous Transfer Mode ATM It includes LAN Emulation LANE for Ethernet clients and allows you to configure three types of network interfaces FORE IP Classical IP and LEC a Optional NeTservices software a UNIX based network operating system is fully compatible with Microsoft networking tech
243. hist 9 39 SCSI adding drives to host port 4 35 channel management 1 6 ID of drive 4 34 interface error message 6 4 port on Host Processor 4 4 4 35 Sector 0O 6 24 Serial port baud rate 2 19 ports 2 18 10 12 printer 2 19 Serial number of drives 7 4 of NetServer 2 17 Server architecture 1 5 Server host name 2 11 2 14 Server problems 10 9 Setting SCSI ID on a drive 4 34 Setting the date and time on the server 2 16 SetupClient 2 27 command 2 27 flowchart 2 27 form 2 29 form fields 2 30 functions of 2 28 how to run 2 29 loading root and swap directories 4 11 SetupExec 2 21 flowchart 2 20 form remote devices 2 24 form fields 2 25 how to run 2 22 loading files to export exec 4 11 SetupTty form 2 19 form fields 2 19 shutdown command 3 8 3 9 10 12 Shutting down the NetServer 3 9 Silent mode of ax_perfmon 9 19 Single user mode 3 4 3 6 6 3 6 5 8 4 Site specific configuration files 6 26 6 28 Size of a clone partition 8 13 Size of a mirrored partition 5 8 Slave server 2 15 Slot number in an SP error message 6 8 Software architecture 1 10 configuration 2 1 2 9 Software architecure of the NetServer 1 10 SP problems that cause file system isolation 6 10 SP retries 6 8 SP Statistics screen 9 13 Space Bar 2 10 Spare root drive 6 31 A index 13 Stale file handle during file system isolation 6 16 Standard boot messages E 2 Starting ax_perfmon 9 3 Static RAM NVRAM 1 6 Static Table Format 4 22 Statistic
244. histogram title after you sum two parameters in the histogram NPO e0 packets sec NPO el packets sec In addition to summing parameters you can add statistics from different input files For example you can combine statistics from server1 dat and server2 dat which contain data collected from two NetServers Refer to Pull Down Menu Displayed in the Input File List on page 9 37 for more information on adding statistics from different input files The example in Figure 9 22 shows a Data Display Window containing three histograms The bottom histogram is the sum of the first two Auspex Systems Data Display Tool aty nists 3 Hist1 Fri Feb 3 15 25 57 1995 Servertacdelco File acdelco dat 432 4 378 3 324 3 Statistics showing 270 2 packets per second on 216 2 interface 0 of NPO ee Fri 15 26 43 Fri 15327327 Fri 15328308 NPO e0 packets sec Ave 107 3 Fri 15328348 Fri 15329329 Hist2 Fri Feb 3 15 25 57 1995 Servertacdelco File acdelco dat 214 4 187 64 160 8 134 0 107 24 80 4 53 6 Statistics showing packets per second on interface 1 of NPO 0 0 Wh bds aats sr sen o ika aah ah hlii L L m L Fri 15 26 02 Fri 15 26 43 Fri 15327327 Fri 15 28 08 Fri 15 28 48 Fri 153293 NPO el packets sec Ave 15 5 Hist3 482 2 421 9 361 64 301 4 241 1 180 84 Ninh agidhlah bi Fri 15326302 Fri 15326343 Fri 15327327 Fri 153283 Fri 153283 NP
245. ia sch aaa ara EA AE ee uns EA KEE a a 9 13 Virtual Partition Statistics Screen 4 6 4 esr g cde ees weg ts 9 15 Using a Filter Pile with ax_perfmon 0451 0 ae achat heey eee ei eee ata 9 16 Creating a Filter Pile ssis ceinion Mek ete eens eae ey dae aes 9 16 Specifying a Filter File on the ax_perfmon Command Line 9 16 Displaying Performance Data in Histograms 6 0 0 e cece eee 9 18 OVERVIEW 6 28 inca Satta Mit poe diane th ate oa gan bal satngithe RE ose 9 18 Before Invoking ax perthist 0 02 filet ieeg anh Pooh eee es 9 19 Starting ax_perfhist Intetactively c c 00 236s 0d Pio aed a Pie 9 19 Loading Statistics From a Biles ovis shania ii S ih ae ta et 9 21 Clearing a BUStO statis sien Seacoast changes Hoe Be He eda 9 22 Displaying Multiple Histograms 6 66 rsrsr rrer 9 23 Loading Statistics Using 2 Step Load 0 00 cee eee eee 9 24 Eliminating a Histogram from the Data Display Window 9 24 Defining the Appearance of a Histogram 000 0000 9 24 Saving and Applying a Configuration File 004 9 29 Examining Samples in Histograms ein aie ou Rae i nee 9 32 Menu for Zooming In and Zooming Out 600 e ee eee 9 34 Summing Parameters in HistOgrams sicuadd ein ook aa 9 35 Deleting an Input File si 5 cee gio pes ee apes Vk Re 9 37 Pull Down Menu Displayed in the Input File List 9 37 Printing Histograms sgi ise es Cirenano eens 9 3
246. icates that if disk errors occurred they occurred on ad47 Examine the SP error messages in var adm messages by entering the following command grep SP1 var adm messages Nov 20 16 06 37 host1 SP1 1233 Recovered SCSI bad bus phase at data xfer slot 47 drive 1 Nov 20 16 06 50 host1 SP1 1196 Disk Media Error slot 47 drive 1 sense key 3 sense code 12 sector 196224 Ignore the first message because it describes a recovered error The second message indicates that an unrecoverable error happened in sector 196224 of ad47 Use the ax_sputil verify command to see if the disk can be verified ax_sputil verify 47 196224 ax_sputil Read write sector error status 02030000 sense data F0000500 3BB1EC14 00000000 21000000 00000000 00000 000 Because the ax_sputil reassign command can return the verification result try mapping out the bad block Try reassigning the sector to a new location as follows ax_sputil reassign 47 196224 ax_sputil Read write sector error status 02030000 sense data F0000500 3BB1EC14 00000000 21000000 00000000 00000 000 ax_sputil restoring data to sector 196224 is needed ax_sputil Confirm reassign sector y n y ax_sputil Read write sector error status Oe03ffe0 sense data 70000500 00000014 00000000 21000000 00000000 00000 000 ax_sputil sector 196224 reassigned and data restored Unrecoverable Disk or File System Problems in Multiuser Mode a 6 17 If the SP c
247. ied Tape is write protected Eject the tape and open the write protection slot tar xvf dev rast2 vmunix ast2 sense key 0x8 blank check vmunix sense f0 0 48 00 fc 01200000000 0 070 T 20 00 On 22 Fe 27 error Tape is not formatted or tape is formatted for the 8500 drive but read in an 8200 drive Use the tape on the correct tape drive or use the correct tape on the current drive Common Tape Drive Errors a 8 19 Memory Error Command tar tvf dev rast13 Error messages tar read error Not enough memory Analysis Data on the tape is not in tar format or an incorrect block size is specified Solution Load the correct tape that contains the tar file End of tape Error Command dd if dev ad0e of dev nrast3 bs 16384 Error messages dd write Error 0 vmunix EOM on forward spacing command Analysis Use mt f dev nrast3 status to see if it returns the following message sense key 0x13 EOT If so the tape has reached the end and the tape drive is unable to write Solution Use a higher capacity tape Tape Media Errors This section discusses errors caused by bad spots on a tape These errors are called Error Checking and Correcting ECC errors and are correctable Tapes often contain unusable spots which the tape drive simply skips over during read or write operations If the number of ECC errors exceeds 10 per megabyte of data transferred vmunix generates a warning message
248. if the Control option in the Print Options Window is set to Auto The screendump and pssun commands accept several arguments for example the X and Y arguments for screendump set the X and Y coordinates of the upper left corner of the area to be dumped the X and Y arguments set the width or height of the area to be dumped The Print Options Window allows you to specify these values individually thereby eliminating the need for memorizing the print command syntax Follow these steps to configure the print command arguments 1 Two commands can capture a screen xwd and screendump If you are unsure which command to use refer to Capturing the screen using xwd or screendump now If you use xwd to capture the screen go to step 2 If you are using screendump to capture the screen assign values to other print options as follows a Specify the screendump and pssun arguments For more information about the arguments refer to the man pages for these commands b Select Monochrome or Color for the Screen Type option c Select Portrait or Landscape for the View option to determine the screen image orientation in the output 2 Select Printer or File for the Output option The Output command is described in detail in Print Command Output on page 9 40 As you configure the print options the print command changes accordingly After setting all the print options click on the Apply button Capturing the screen using xwd or scree
249. iguring the NetServer to Be an NIS Slave 0 2 4 NetServers and UDP Checksumming 6 60 e eee ee nna 2 5 Modifying Site Specific Files can isn h i ah ee ae eee 2 5 Kernel Parameter s evieioviiys s t e Rae eee Oh ae eae ee 2 6 Additional TTY Sicsigjccctak ton n a ee aurea hos 2 6 Mailing Site Reports to Auispexiass cuancteiuas kia an cen seal e ae 2 7 NetServer Configuration Commands india coer cara ene ee chee res 2 9 Using Configuration Command Forms 000 0 e ee eee eee 2 9 IN SCORES r eaa ware ta te os Sel einer A PE A it Wace a arg ie 2 10 Before Running NSconfig wis sah p rad dat aU ae aS 2 11 Contents A v R nning NSCOMES sii putes Meise Bila Ak ay EEE DATER ee ie 2 11 SetUpTty eriari kaci ie aaa aia Lion naka Ea EE Ein E EET aa near nate TSS 2 18 SetUpEXEC n eai a kes PE AEE EE a E TEE ET EE ETEN ae nag iA 2 20 Do You Need to Run SetupExec 0 0 0 0 2 20 What Does SetupExec Do 0 ccc eee ees 2 21 Things to Remember When Running SetupExec 4 2 21 R nni g DETIPEKOG te ow scaanehurettweies du ahh eh GAT AIS ahaa 2 22 SetupClent ssicimuchapeens Sexes Ea E Wl oe ee A E ES ay 2 27 Do You Need to Run setiipChent orig 26 fc iG eeoadiatan coeds aie 2 28 What Does SetupClient Do 2164 0 B S Shs CaN ESAs 2 28 Running Det pCEN bac oss Rae ch gee hate eee Bare hase 2 29 Root Login and File System Isolation 00 cece eee eee 2 31 100Base T Ethernet Pu
250. in 100 megabytes 0 2 4 6 8 10 12 14 16 18 c abl g idl e f h hostname ax_lslabel c ad0 device start_cylno nblks dev rad0a 0 32768 dev rad0b 32 204800 dev rad0c 0 3909632 dev rad0d 488 131072 dev rad0e 616 624640 dev rad0f 1226 204800 dev rad0g 232 262144 dev rad0h 1426 2449408 hostname gt ax_lslabel sc ad0O device sizeA sizeB sizeC sizeD sizeE sizeF sizeG sizeH ad0 32 200 3818 128 610 200 256 2392 4 42 a Version 1 10 System Manager s Guide W AUSPEX Virtual Partitions About This Chapter This chapter describes the Virtual Partition Manager which comprises these commands and files that create and manage different types of virtual partitions supported by the NetServer a vp a vpartab a ax_diskconf a ax expand a ax_loadvpar a ax_mattach a ax_mdetach a ax_mconvert a ax_mrestore a ax_vold a ax_vpstat This chapter also enumerates the advantages and limitations of using virtual partitions on your server Virtual Partitions a 5 1 Virtual Partition Overview The Auspex NetServers offer the capability to group one or more physical disk partitions from one or more disk drives into a single virtual partition Virtual partitions allow you to a create disk partitions larger than the maximum size of a single physical disk a use all available disk space by combining all unused runt partitions into a usable size a distribute high use file systems over multiple
251. ing Up With a Large Number of Disks The procedures for automated backups for a NetServer with fewer than 50 GB of storage are the same as the procedures for other servers with a comparable amount of disk storage However for a server with more than 50 GB of storage implementing a scheme for unattended automated backups requires planning This section provides some recommendations for backing up large capacity NetServers You may wish to modify the scheme to suit your environment and needs You need to create the cron 8 scripts that execute the backups the scripts are not supplied by Auspex The scheme for unattended backups shown in Figure 8 1 on page 8 10 makes the following assumptions A The NetServer has 40 4 GB disk drives arranged in 80 2 GB file systems and one or more tape backup systems attached to the HP SCSI File systems average less than 10 change per week Backups are performed five nights per week In Figure 8 1 below 0 represents a level 0 dump and the numbers 1 5 and 9 represent different levels of incremental dump A Level 0 dumps are performed biweekly on each file system level 0 dumps are distributed over five nights not performed all on the same night Each night s level 0 dump of a single file system fits easily on one tape Incremental dumps alternating levels 1 5 and 9 are performed on each file system on all nights when level 0 dumps are not performed Week 1 File System Sun Mon Tue Wed Thu
252. interval host0 gt ax_netstat i I ael Name Mtu Net Dest Address Ipkts Ierrs Opkts Oerrs Coll ae0 1500 192 42 160 0 192 42 160 245 1470 0 3607 0 0 To show statistics on a per protocol basis enter I interface ax_netstat s The s flag displays statistics about the UDP IP and ICMP protocols gathered since boot time The statistics pertain to the entire processor board to which the specified interface is attached In this command form I interface is mandatory 10 16 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The following example shows the ax_netstat s command and its output host0 gt ax_netstat s I ae3 udp ips 0 0 0 0 0 0 0 icmp 0 Or 7O O incomplete headers bad data length fields bad checksums 65516383 total packets received bad header checksums with with with with 165188 size smaller than minimum data size lt data length header length lt data size data length lt header length fragments received fragments dropped dup or out of space 911 fragments dropped after timeout 9684974 packets forwarded packets not forwardable redirects sent calls to icmp_error 0 errors not generated cuz old message was icmp Output histogram echo reply 35 messages with bad code fields messages lt minimum length bad checksums messages with bad length Input histogram echo reply 62 destination unreachable 9 routing redirect 489
253. interval between screen updates In the previous example the screen is updated every second The title of the screen follows the time interval The title can be one of the following System Summary NP Stats FP Stats SP Stats Virtual Partition Stats gt gt gt gt gt The header also includes the name of the host from which statistics are gathered and the current date and time If you are playing back statistics the host name and time information show where and when the data were collected The header for the System Summary screen shows additional information on the server type The server type which is displayed before the host name can be one of the following a NS7000 200 for an NS 7000 200 Series NetServer a NS7000 600 for an NS 7000 600 Series NetServer IS THIS SUPPORTED a NS7000 700 for an NS 7000 700 Series NetServer a NS7000 800 for an NS 7000 800 Series NetServer The following sections include screen examples generated by different server types System Summary Screen The format of the System Summary screen is different among server types Figure 9 1 is a sample System Summary screen for the NS 7000 200 Series NetServer Figure 9 2 is a sample System Summary screen for the NS 7000 600 Series NetServer The NS 7000 800 NS 7000 700 NS 7000 500 and NS 6000 System Summary screens are similar to the NS 7000 600 The statistics on the screens are described after Figure 9 2 9 6 a Version 1 10 System Manager s G
254. inue to use write acceleration ax_write_cache s 0 enable 6 Unmount the file system so you can replace the disk 7 Use ax_hot_plug to replace the drive For example ax_hot_plug add 36 Refer to Installing or Replacing a Drive After Booting on page 4 26 for more information on how to replace a disk 8 After replacing the disk run newfs and fsck to create and check the new file system 9 Restore the file system from the most recent backup Run fsck to check the file system after the restore 6 16 a Version 1 10 System Manager s Guide W AUSPEX 10 Mount and export the file system so it is available to clients again 11 Reboot diskless clients that crashed when the disk was replaced This step is required for this example For other file systems the clients would not have crashed and would not require a reboot The file system recovery procedure is now complete File System Isolation After Media Errors That Can Be Fixed The example in this section shows how to recover from an isolated file system by repairing a bad sector that caused file system isolation 1 Enter the following command to determine which file system was isolated grep isolate var adm messages Nov 19 17 05 02 host2 syslog ax_isolated Filesystem dev vp277 mounted on home is isolated Determine which disk underlies vp277 by entering the following command grep vp277 etc vpartab dev vp277 concat ad47c The screen output ind
255. ion go to the next section Is the Problem Caused by a Disk Without a Disk Label If the disk cannot be detected in another slot or if the system still cannot reboot to multiuser mode after the disk has been detected the problem may be caused by a missing disk label Although the SP can recognize a disk without a disk label during an ax_hot_plug or ax_add_device command ax_loadvpar and fsck fail if they cannot see the label If an error 6 4 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX message during reboot indicates that there is no disk label on a disk use the format command in single user mode to relabel it Refer to If Sector 0 Disk Label Is Unreadable on page 6 24 for instructions on fixing the problem If format does not list the disk you can conclude that the problem arises from a malfunctioning disk You must remove or replace this bad disk before you can boot the system to multiuser mode Replacing a Malfunctioning Disk To replace a drive refer to Replacing a Drive in Single User Mode or Entering Multiuser Mode and Replacing the Drive Replacing the disk allows users to have continuous access to the file systems on the drive although the data on the drive may not be up to date Follow the instructions in Using the System Without the Failing Drive on page 6 7 only if you do not have a spare drive and need to wait for a replacement Replacing a Drive in Single User Mode This
256. ion of disk I Os Given the large I O memory cache provided by an NFS file server most client NFS read operations get their data from the I O memory cache As a result the majority of disk I Os are write operations The write cache design helps NetServer throughput in two ways First it discards redundant writes to the same disk location as happens with inode and indirect block updates Second it coalesces sequential data blocks that remain in the cache and have not been written to disk Each of these design features helps reduce the number of write operations to disk The Write Accelerator III on the SP V has 8 MB of NVRAM The Write Accelerator includes a battery backup Note Because the data written to the cache is not written immediately to disk i a disk or power failure can occur before the data is permanently stored When this happens the contents of the write cache are preserved until the error is corrected While the write cache is disabled NFS writes are written to disk in the conventional manner Caution Although the NVRAM in the Write Accelerator can prevent data loss A during a power outage the integrity of data in the cache is not guaranteed Before writing cached data to disk the SP checks to see if the control data in the cache is valid If not it changes the state of the Write Accelerator to BADCHECKSUM Purge the data from the cache when this happens because the SP cannot determine the disk location to which to w
257. is always 1 because each drive slot contains only one drive It is not necessary to replace a drive that has experienced recoverable errors However if recovered errors occur often or if you believe the read and write retries are slowing down the system use ax_clonefs or ax_sputil copy to copy the disk to another disk When you finish copying replace the disk with the newly created disk 6 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Unrecoverable Disk or File System Problems in Multiuser Mode When a system is running in multiuser mode a file system can become unavailable because of a hardware or software problem The following subsections discuss how you can resume the use of a file system after an unrecoverable disk or file system error A Error Messages Indicating Disk or File System Problems discusses the symptoms of an unrecoverable disk or file system error File System Isolation describes a NetServer feature that takes an inaccessible or corrupted file system off line without affecting the operation of the entire server This section discusses the types of errors that cause file system isolation and outlines the major steps to follow to add the file system back online Examples of Recovery Procedures After File System Isolation provides examples in which file systems are isolated under different circumstances In each example a detailed procedure shows how to determine the cause of
258. isolated If you want messages sent to additional users enter user names after the m flag Use a comma to separate multiple user names For example ax_isolated m bsmith lchung smueller When a file system is isolated the ax_isolated daemon automatically calls fsck to check the file system If fsck is successful ax_isolated then releases the file system If you wish to disable automatic calling of fsck as well as ax_fsutil invoke ax_isolated with the n flag An additional option for ax_isolated is f which is automatically placed in var spool cron crontabs root Do not invoke this flag from the command line Note Upon system startup ax_isolated is invoked without the m n or f D flag To start ax_isolated with the m or n flag you must first kill the currently running ax_isolated daemon Refer to the ax_isolated man page for more information If you suspect file system isolation use the following command to determine which file system is isolated ax_fsutil 6 12 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Running ax_fsutil without any arguments prints a list of isolated file systems ax_isolated 8 also places file system isolation messages in var adm messages The ax_isolated daemon generates the following message showing which file system was isolated Mar 14 07 15 58 hostl ax_isolated 215 Filesystem dev ad52c mounted on disks eng is isolated The procedure for recovering from
259. ist of all exported file systems Check the output to verify that the new file system was exported properly The disk drive is now ready for use 4 30 a Version 1 10 System Manager s Guide W AUSPEX Removing a Drive If you need to remove a drive and do not plan to replace it with another drive right away use the following procedures Remove a device in one of two ways depending on whether the operating system is booted Note For information on adding removing or replacing HDDA drives refer i to your hardware manual Removing a Drive When the Operating System is Booted This procedure does not apply to the root drive To replace the root drive follow the procedure for replacing a drive while the operating system is shut down on page 4 28 Before You Remove a Drive Follow these steps before you remove a drive from the server 1 Refer to the hardware manual for your NetServer for guidelines on handling drives 2 Log in as root on the system console 3 If you are replacing an existing drive make sure no partition on the drive is open for read or write access For example verify that a no file system is being dumped a no process such as a database is using a raw partition a no swapping is being done to the disk a no partitions are members of other virtual partitions Use the ax_diskconf command to get a report that maps all file systems to the physical partitions or virtual partitions they reside on De
260. ites existing files A If the system reboots the tape automatically rewinds The restore command has the following syntax usr etc restore option filename Using the Dump and Restore Commands a 8 3 Option is a character that can be followed by a modifier or argument It determines how the system restores the files for example whether to restore files interactively whether to list files on a tape using a table of contents and so on For more information on available options see the restore 8 man page For more information on the strategy to restore a file system refer to Sun s System and Network Administration Recommendation For HP mounted file systems you can speed up restores pe by using the F modifier with the restore command This modifier enables delayed writes of file system information to the disk allowing restore to run about three times faster You can also query and enable delay writes with the fastfs command For more information on available options for delaying file system writes see the fastfs 8 man page System backups and restores with dump and restore can be run with the system in one of three different modes a Ona quiescent file system in multiuser mode a Ona quiescent file system in single user mode a Onan active file system although this is not recommended The dump and restore commands used for each mode are similar See Sun s System and Network Administration for dump and restore examp
261. ition active status 5 12 advantages of 5 2 cloning 8 12 concatenated 5 4 converting to a mirrored partition 5 11 creating 5 14 damaged status 5 12 defining 5 14 dirty status 5 12 disk errors at reboot 6 3 error recovery 5 13 expanding 5 15 files drivers and commands 5 9 for export root and export swap 4 11 4 17 information worksheet D 1 D 7 larger than 2 GB 5 3 managing 5 14 maximum size of 5 3 mirrored 5 5 5 9 6 10 8 1 numbering 5 9 reconfiguring 5 14 recovering disk errors 6 20 restoring status 5 12 restrictions 5 7 running newfson 5 14 status 5 11 striped 5 4 5 9 synced status 5 12 typesof 5 4 Virtual Partition Manager 1 4 components of 5 1 copying data between disks 6 22 Virtual Partition Statistics screen 9 15 VME 1 4 4 18 VME bus 1 6 vmunix 3 6 4 4 6 9 8 20 Volume daemon 5 11 vp device driver 1 13 5 9 vpartab table 1 13 5 9 5 10 5 14 W WarmStart 1 3 Warning message from the SP 6 10 Worksheet client information D 1 D 4 configuration D 1 disk drive information D 1 D 6 host information D 2 virtual partition information D 1 D 7 Write acceleration accelerator board 7 2 analyzing the need for 9 48 badchecksum status 7 6 dirty status 7 6 disabling 7 9 enabling 4 22 7 8 error recovery 7 9 examining statistics with ax_perfmon 7 8 how to manage the cache 7 9 none status 7 6 NVRAM 7 3 off status 7 6 on status 7 6 overview 7 2 A Index 15 AUSPEX QD restrictions 7 5 X console termina
262. ition is offline The information includes using the dump and restore commands calculating the amount of tape to use for a backup asuggested scheme for backing up a system with 40 drives backing up the root disk 4 Backing up a file system while it is online that is cloning an active file system a Restoring a file with a damaged inode Remember that no matter how effective your backup strategy is if you are not following the recommended procedures for cleaning and maintaining your tape drives and tape media you risk losing valuable data For more information refer to Preventive Maintenance in the hardware manual for your NetServer For a more detailed description of backups and how to perform them see Sun s System and Network Administration Recommendation Although a mirrored partition protects data on the per partition from loss because of disk or media errors mirroring cannot protect the data from loss because of other causes such as accidental deletion Follow a regular backup routine for mirrored partitions as for other partitions Backing Up the NetServer a 8 1 Using the Dump and Restore Commands The dump command is a flexible and effective command for performing backups at several levels of completeness A full system backup copies everything on the file system while an incremental one backs up only selected parts of the system The dump command allows you to specify a dump level from 0 to 9 Dump
263. itions If you are only interested in the physical partitions use the ax_Islabel command to display the information Refer to Displaying a Disk s Partitioning Information on page 4 42 for more information on ax_Islabel The ax_diskconf report consists of two tables each displaying disk information in different ways The first table is organized by file system For each file system the following information is provided a Name of the mirrored virtual partition if the file system is a mirrored virtual partition a Names of the members of the mirrored virtual partition if the file system is a mirrored virtual partition Name of the virtual partition if the file system is a concatenated or striped partition a Name of the DriveGuard array if applicable a Names of the physical disk partitions on which the file system resides If the file system is a virtual partition this column also indicates whether the physical partitions are concatenated or striped In the list concatenated partitions are separated by a plus sign and striped partitions are separated by a vertical bar The second table is organized by drive For each drive the following information is provided in graphic form a The layout of partitions on the drive indicated by the partition identifiers a The location of unused partitions indicated by decimal points In addition if you enter the ax_diskconf command with the p option the second table is exp
264. itor To start collecting server performance data execute ax_perfmon 8 The syntax of the ax_perfmon command is shown as follows ax_perfmon s o fil f filterfile i file p period t time The optional arguments perform the following functions S Executes ax_perfmon in silent mode Does not display data as it is gathered This option cannot be used with the i option o file Stores gathered data in a specified output file The data in this file can be displayed using the i argument This option is required to save the Capturing and Displaying Real Time Performance Data a 9 3 statistics for display later using the ax_perfhist command which organizes the statistics in histogram format for easy viewing For further information on ax_perfhist refer to Displaying Performance Data in Histograms on page 9 18 Note The o option uses a lot of space A file can grow to a few megabytes in D size in as little as 30 seconds if the update period see the p option is left at the one second default Be sure that the partition receiving the output file has adequate space and use the p option to control the number of updates that go to the output file f filterfile Specifies the name of the file containing a list of variable names which determine the types of performance data for example statistics concerning network interfaces to save in the output file Specify the filter file only if you are writing perf
265. itten license agreement from Microsoft Auspex Systems Inc 2300 Central Expressway Santa Clara California 95050 Phone 408 566 2000 Fax 408 566 2020 Internet info auspex com World Wide Web http www auspex com Protection Against Electrostatic Discharge To prevent damage to the system due to electrostatic discharge always wear the antistatic wrist strap provided with your network server when you come in contact with the system Publication Change Record The following table records all revisions to this publication The first entry is always the publication s initial release Each entry indicates the date of the release and the number of the system release to which the revision corresponds Part Number Date Description 850517 001 March 1998 Release 1 10 W AUSPEX Contents Preface Chapter 1 Overview of NetServer Features About This Chapter seica saa aot haa eae twa ea tae eR A he Sas aoe od 1 1 Product Overview vimus ee eh ae i oe BA hee ad Ce eee Pia aes 1 2 NetServer Hardware Architecture 0 cece eee eee 1 5 Host Processor HIP i eis on ea ead Gag Dae tine Gua Gaede ead 1 5 Network Processor NP 0 2 00 eee eee etn ee eet eee eens 1 6 Storage Processor SR ress ewe ciate agg RE abe eee BEE Atay ae 1 6 TCC ache Memory a waxad ciucag a i aad Gates eee wae cae e 1 6 Enhanced VME B s ovat gass teal ey tad AGE GOs Qa ted E 1 6 Optional Products cntonn aca wits tat bias
266. ive d 1225 3 Recovered SCSI selection timed out slot d drive d 1226 3 Recovered SCSI handshake timed out slot d drive d 1227 3 Recovered SCSI bad status byte slot d drive d 1228 3 Recovered SCSI request sense failed slot d drive d 1229 3 Recovered SCSI reconnection error slot d drive d 1230 3 Recovered SCSI error in message in slot d drive d 1231 3 Recovered SCSI error in message out slot d drive d 1232 3 Recovered SCSI message reject error slot d drive d 1233 3 Recovered SCSI bad bus phase at data xfer slot d drive d 1234 3 Recovered SCSI bad bus phase at command xfer slot d drive d 1235 3 Recovered VME data transfer timed out slot d drive d 1236 3 Recovered AlC6250 fifo status error slot d drive d 1237 3 Recovered AFC chip status error slot d drive d 1238 3 Recovered AFC status error at disconnection slot d drive d 1239 3 Recovered SCSI inbound data parity error slot d drive d 1240 3 Recovered drive disconnected at odd byte slot d drive d 1241 3 Recovered AlC6250 status error slot d drive d 1242 3 Recovered AlC6250 error at disconnection slot d drive d 1243 3 Recovered DMA illegal operation slot d drive d E 14 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1244 3 Recovered DMA VME bus timed out slot d d
267. ive partition tables lt 5 so6 ied Ree eee cee poe anes 4 8 Root drive partition table in MBs 0 eee cece eee ees 4 8 Example of disk partitioning among multiple disks 4 12 SCSI IDs and names for devices on the HP 0 eee eee 4 35 Disk drive format verification time eee eee ee eee 4 39 Virtual Partitions Numbering virtual partitions 22ie 3 cicladee5 Bie dee eu ieee Reed 5 10 Recovering From Disk or File System Failures The NSinstall form fields 0 0 0 cece eens 6 31 Write Acceleration ax_write_cache command options 0 0 6 eee eee 7 8 Backing Up the NetServer Arguments for the dump command voi vey tid jae orgnrtaar eae 8 2 Tables A xv Table 8 2 Tape drive data capacities c5s shies la ia awe eer ee oes 8 6 Table 8 3 Suggested dump and restore arguments 00 000 eee eee eee 8 7 Table 8 4 Dump and restore command examples 00000 e eee eee eee 8 9 Table 8 5 Differences between cloning and mirroring a partition 8 12 Chapter 9 Measuring the NetServer s Performance Taple 9 1 NPStatisties i4cc4 h40 fiend wet ase ten we ae anche eek se ga lode ta 9 9 Table 9 2 NP Protocol Statistics iba ie aaah ate iar g 8 okt NE eae tase 9 11 Table 9 3 Fields in the Histogram Properties Window 00000 ee eee 9 25 Table 9 4 Fields in the Overall Properties Window eee e eens 9 29 Table 9 5
268. ization of the CPU for network processing Nfs work que busy Number of NFS workers running on the NP number of NFS jobs queued waiting for free NFS workers to process number of busy NFS workers An NFS worker is an NFS server daemon servicing remote procedure calls from clients Multiple NFS workers allow simultaneous NFS operations to take place on a server MBUF stats Statistics for MBUF interfaces 0 1 2 and 3 are for BUFE slots on the NP f is for FDDI E is for half duplex 100Base T Ethernet O is for offboard buffers taken from another NP T is for TCP ais for ATM large buffers and small buffers H is for full duplex 100Base T Ethernet Frames Number of frames received and sent by the Ethernet interface Bytes Number of bytes received and sent by the Ethernet interface IBytes Number of bytes received by the Ethernet interface Obytes Number of bytes sent by the Ethernet interface Capturing and Displaying Real Time Performance Data a 9 9 Table 9 1 NP statistics Continued Statistics Explanation InErr Number of frame errors received by the Ethernet interface OutErr Number of frame errors sent by the Ethernet interface Collis Number of Ethernet collisions on the Ethernet interface OutDisd Number of output packets discarded by the Ethernet interface InBig Number of oversized frames received by the FDDI interface InUndrsz Number of undersized frames received by the FDDI interface InInval Number of frames with inv
269. k sp_msg 16 retries sp sp_scsi_cmd_error boot_simple_send err_code X err_codes X X X X err_code X TIMI EOUT System Startup Error Messages a E 7 Auspex Processor Error Messages Auspex processor error messages appear in the system log in the format processor type and no error no message text For example for error 1209 on the first SP board the message is SPO 1209 Recovered drive seek failure slot 10 drive 1 In the example SPO is the processor type and number 1209 is the error number and the remainder is the message text If you need to report an error message to your service representative please have all relevant information available including the entire message recorded in the system log the state of the server and network and the type of activity occurring at the time of the error Table E 2 provides the following information for each message a Error number A Priority code which is a number in the range 1 3 The following list explains the meaning of each number 1 System panic or isolated file system report these messages to your Auspex service representative immediately 2 Recovered errors report these messages only if they recur frequently 3 Notifications these messages are generated by routine conditions such as user errors or recovered errors on a peripheral device A Message text In the table the following conventions are used to represent variables d a d
270. k before formatting it Running the Format Command in Single or Multiuser Mode The format command is usually run in multiuser mode However in multiuser mode if the disk drive you want to format is ad0 you must run format after booting from the CD ROM or an alternative root device Recommendation For faster recovery from a disk or media error on the root ie disk keep an up to date backup of key partitions of the root disk on a separate disk for example ad2 Additionally make tape backups of all key data files including root file systems This procedure is described in Backing Up the Root Disk on page 8 11 Caution Do not use the format command for DriveGuard arrays See the A DriveGuard Manager s Guide for using ax_label 8 with DriveGuard arrays Accessing the Format Menu 1 Make sure the disk on which you want to run format is installed in the system For installation instructions see Adding or Replacing a Drive on page 4 25 2 Execute format and select the disk you want In the following example the NetServer has a total of 8 disk drives the user selects ad4 Enter the disk number in the form of adn where n is the slot number format Searching for disks done AVAILABLE DISK SELECTIONS ado adl ad2 ad3 ad4 ad5 ad6 ad7 Specify disk number enter its name ad4 selecting ad4 lt SEAGATE 4GB gt disk formatted no defect list found 4 36 a Version 1 10 System Manager
271. l 4 16 states and commands 7 6 X terminal Storage Processor statistics 9 14 address 2 11 uninitialized status 7 6 as system console 2 15 used with a mirrored partition 5 5 name 2 11 used with mirrored partition 5 5 use of usr openwin 4 16 Write cache X windows analyzing data 9 48 on line documentation B 2 control datain 7 3 xwd command 9 38 9 39 hit percentage 9 14 on state 7 6 Y purging unwritten data 7 10 states and commands 7 6 ypinit command 2 4 statistics 9 3 Z X Zooming in on samples in a X console 2 15 histogram 9 32 to 9 34 Index 16 A Version 1 10 System Manager s Guide
272. le 1 AXEC1 etherchannel HP VII HP VIII 1 10 2 AXNTBios Auspex NetBIOS HP VII HP VIII 1 10 3 AXNeTsrv Advanced Server for UNIX Systems HP VII HP VIII 1 10 4 AXatm2 ATM 2 HP VII HP VIII 1 10 5 AxXbackup FastBackup HP VII HP VIII 1 10 6 AXdgrd DataGuard HP VII HP VIII 1 10 7 AXdocs Auspex System Documentation HP VII HP VIII 1 10 8 AXdrvgrd DriveGuard HP VII HP VIII 1 10 9 AxXftp NP Resident FTP HP VII HP VIII 1 10 10 AXsrvgrd ServerGuard HP VII HP VIII 1 10 Select package s you wish to process or all to process all packages default all q Enter q to quit the pkgadd routine Note that the pkgadd command copies the session into two log files as shown in the following display pkgadd session finished on Tue Jan 14 15 28 50 PST 1997 A log of this session of pkgadd is at tmp pkgadd log on the current root disk and at var log pkgadd log on the target disk B 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Starting the Online Documentation To open the online document from an X display on the network use the following procedure In the procedure the machine foobar has the local X display screen and the server dochost1 holds the ax_docs 8 utility and files 1 Verify your path contains the path entry usr auspex 2 Add the server dochost1 to your xhost list foobar xhost dochostl 3 Use rlogin 1C or tip 1C to login to d
273. le Intermittent disk errors are harmless unless they happen frequently for example several times a week Replace a disk with a high rate of intermittent errors by using one of the procedures described in the next two sections Unrecoverable Disk or File System Problems in Multiuser Mode a 6 21 Note Both procedures involve copying data from the drive with errors to a i spare drive However if the drive with errors becomes invisible to the system before data copy is complete follow the procedure in Recovering from Permanent Disk Errors Without File System Isolation on page 6 19 Replacing a Drive that Is a One Member Mirrored Partition If the drive that experiences intermittent errors is a one member mirrored partition use commands in the Virtual Partition Manager to copy data from the bad disk to the spare disk The following example assumes the bad disk contains a single virtual partition vp3 and vp3 is the only member of a mirrored partition vp2 The entry in etc vpartab for vp3 is as follows dev vp2 mirrored vp3 1 Obtain and install a spare drive the new drive that is the same size as the one to be replaced the old drive Choose an available drive slot for the new drive Use ax_hot_plug to install the new drive Insert the new drive and then type a carriage return a FY N Edit etc vpartab to create a virtual partition on the new disk For example if the spare drive is in slot 8 you can add the fol
274. le 2 7 NetServer Configuration Commands a 2 21 Table 2 7 Sun system architectures Application Kernel architecture architecture Sun system models supported sun3 sun3 3 50 3 60 3 75 3 110 3 140 3 160 3 260 3 280 sun3 sun3x 3 80 3 470 3 480 sun4 sun4 4 110 4 260 4 280 SPARCsystem 330 sun4 sun4m SPARCstation20 SPARCstation10 SPARCstation5 SPARCclassic sun4 sun4c SPARCstation 1 4 60 SPARCstation 1 4 65 SPARCstation 2 4 75 SPARCstation SLC 4 20 SPARCstation IPC 4 40 SPARCstation IPX 4 50 SPARCstation ELC 4 25 Running SetupExec The following procedure describes how to start SetupExec and fill out the SetupExec form 1 2 Make sure the system is in multiuser mode Log in as root Insert the SUNBIN CD or tape into the appropriate drive If the SUNBIN medium is a CD ROM and is installed on a remote host mount the CD ROM see Things to Remember When Running SetupExec on page 2 21 Execute SetupExec SetupExec The command displays the SetupExec form Sample forms appear in Figure 2 6 Figure 2 7 Figure 2 8 and Figure 2 9 Table 2 8 explains the fields in the SetupExec form In some cases if you select a certain value for a field an additional field appears on the form For example if you select Remote for Drive location you are prompted for the media host and the host s IP address The additional fields are indented in the table Note If you specify a remote tape
275. le Disk or File System Problems in Multiuser Mode a 6 19 2 If the disk is a member of a concatenated or striped virtual partition follow these steps a Edit out the entries for this drive in etc vpartab b Run ax_loadvpar to load the new virtual partition table which no longer includes the bad disk Use ax_hot_plug to replace the failing drive with a spare drive 4 Verify the new drive has the correct disk label The new disk should have the same partitions as the replaced disk Obtain the partitioning information about the replaced disk from your records or worksheets Use ax_Islabel to check the partitioning scheme for the new disk and use format to repartition it if necessary All partitions referenced in etc vpartab must exist in the disk label and their size must be greater than 0 MB 5 If the disk is part of a concatenated or striped virtual partition follow these steps a Open etc vpartab and add entries for this drive b Run ax_loadvpar to load the new virtual partition table 6 Run newfs to create the file systems on the new disk 7 Restore data to the file systems on the new drive from the most recent backup tapes 8 Mount and export the file systems you unmounted and unexported in step 1 Replacing a Disk Containing Mirrored Partitions If the disk experiencing problems contains mirrored file systems follow these steps to replace it Caution The following line appears at the top of the etc vpartab and
276. len 0 1079 1 ufs_readdir dir reclen gt 512 1080 1 ufs_badop 1081 1 vn_rele 1082 1 vfs_unlock 1083 1 bread size 0 1084 1 bread 1085 1 breada 1086 1 breadrabp 1087 1 bwrite 1088 1 getblk size too big 1089 1 geteblk size too big 1090 1 brealloc 1091 1 dup biodone 1092 1 binval_vp_ctrl not implemented 1093 1 dnic_purge zero vp 1094 3 dnic_purge1 no entries to purge 1 ufs_readdir_nfs dir reclen 0 1095 2 Warning Proc 0x x received bogus message 1 ufs_readdir_nfs dir reclen gt 512 1096 1 sleep chan 0 1097 1 _subr_on_stack Stack overflow for func 1098 1 _stack_depth_used Stack overflow detected 1099 2 nfs_server bad proc number 1100 2 nfs_server bad version number 1 ufs_readdir_nfs dir reclen not word aligned 1101 2 nfs_server Can t allocate args struct 1 ufs_readdir_nfs dir reclen not word aligned 1102 2 nfs_server bad getargs 1103 2 nfs_server weak authentication source IP address s Auspex Processor Error Messages a _ E 11 Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1 dirlook dir reclen not work aligned 1104 2 nfs_server Can t allocate results struct 1 dircheckforname dir reclen not word aligned 1105 2 nfs_server bad freeargs 1 ufs_readdirplus dir reclen not word aligned 1106 2 nfs_server bad sendreply 1 ufs_getdirattrs dir reclen 0 1107 2 NFS request from unprivileged Channel 1 ufs_getdirattrs dir reclen gt 512
277. les for each of these operating modes Also refer to Restoring a File With a Damaged Inode on page 8 17 for an example of restoring a file whose inode is damaged 8 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Tape Drive Capacity This section describes the capacity of each type of tape drive supported by Auspex Use the information in this section to estimate the number of tapes needed to back up your server The maximum capacity of an Auspex tape drive depends on the following factors a Type of tape drive 8 mm DLT4000 and so on i Note The DLT4000 tape drive requires a DLT4000 drive rack 4 Length of tape used in the drive 60 meter 90 meter 112 meter and so on a Whether data compression is enabled which is determined by the following For 4 mm drives the switch 2 setting on the drive the device name used in software commands for example rast5 and rast5c and the prerecorded data format on the tape For 8 mm and DLT4000 tape drives the device name used in software commands Data Format on 4 mm Tapes Switch 2 on the 4 mm tape drives affects the data format DDS or compression used on the tape When the switch is OFF the drive can be used in low density or high density mode when it is ON the drive can be used in high density mode only For the switch 2 location and information on how the switch setting works with device naming to determine the data format refer to your NetServe
278. let wnt ined a get are tid au kaGae pen ae Re aoe ae 8 1 Using the Dump and Restore Commands 00 c cece eee eee 8 2 Dump Command esris acera iiaee aa aa E EE EA EEEE 8 2 RESTORES Commands siei okee i a sh tal pb A ee E yak Poe eee 8 3 Tape Drive apacity wins cs esta macnn aus wie G0 Lana whee AY Ahaha are 8 5 Data Format on 4 mm Tapes 0 6 eens 8 5 Tape Drive Capacities sese inii iG A eat TEES Nba ea eonn hans 8 5 The Size Density and Blocking Factor Options for Dump 8 7 Backing Up With a Large Number of Disks 0 0000000 e eee 8 10 Backing Up the Root Disk cana ayde tye non to8 ava ded rere 8 11 Online Backup erenn eee a eee eT ne ee ween nee eee ceo 8 12 Cloning a Virtual or Physical Partition suasanana esnea 8 12 Important Guidelines for Creating a Clone Partition 8 13 Procedure for Cloning a Partition ussas aie eee ines ahs 8 14 Dumping a Clone Partition to Tape snaa oe eng APRs es ea eae 8 15 Restoring a File With a Damaged Inode n n elias sak g ond Dace 8 17 Restoring a File from 1 5 1 produced Tapes 0000 e eee 8 18 Common Tape Drive HrrOts 4 6 gc haa edie ihe ww ALG ce rrr 8 19 Operator Errors oreraa ina aA a a a a aae lg ids ted aaa 8 19 MO ERO reae a tte Moet Aa one ean E a AA 8 19 Writer Error si a esas ON Peed A E ht G aN aE a A 8 19 Blank Tape Errori uipi s Bene i n E EEEE E OI EGEE 8 19 Memory Error siir ear e
279. lhcO ecccooccoccococeoceoceoceceo c bed ee bd kol bd kd bad bl bd bd ee kl bd ee had bad khad bd Oc Oc Ol ol Ol OI oc Oc Ol ol Ol oc Oc Oc ol ol oc oc Oc ol ol Ol Figure 9 7 Example of a Virtual Partition Statistics screen Each SP supports up to 256 virtual partitions Each SP board has four virtual partition screens The first screen displays statistics for SPO partitions vp0 vp63 For information on subsequent virtual partitions on the same SP type v The next screen for the same SP displays statistics for vp64 vp127 The third screen displays statistics for vp128 vp191 The final screen displays statistics for vp192 vp255 For information on subsequent SPs type s The partition numbers for each SP match the partition numbers set up in the etc vpartab file so the first number on the SP1 display is 256 Figure 9 7 shows an example of the first screen for SP1 The system supports up to five SP boards The number of I O operations follows the partition number The operations bar shows which member of the virtual partition actually did the operation A 0 represents I O going to the first member 1 represents I O to the second member and so on Each write to a mirrored virtual partition generates a write to each member so the I O operations count may be higher than expected A read to a mirrored virtual partition is satisfied by a read to just one member Capturing and Displaying Real Time Performanc
280. lizes the write cache and if all data successfully writes to disk switches it to the OFF state enable Puts the write cache in the ON state You can set the write cache to ON only from the OFF state disable Puts the write cache in the OFF state You can set the write cache to OFF only from the ON state This option flushes all the cached blocks If an error occurs while flushing these blocks the write cache enters the DIRTY state instead of the OFF state purge Discards all unwritten data in the write cache and then switches the state from DIRTY or BADCHECKSUM to OFF flush Disables the write cache and writes all data in the write cache to disk This only applies if the write cache state is DIRTY If the flush is successful the write cache state becomes OFF If the flush is unsuccessful the write cache state remains DIRTY state Displays the write cache states UNINITIALIZED OFF DIRTY ON BADCHECKSUM or NONE errors Prints a list of cache pages that cannot be written to disk This only applies if the write cache state is DIRTY To examine write cache performance statistics use the ax_perfmon command Enabling the Write Accelerator Enabling the Write Accelerator involves invoking the ax_write_cache command and modifying etc fstab The cache memory is automatically initialized and switched to the ON state enabled at boot time Write acceleration is enabled only after the cache is turned on and an entry a
281. ll Duplex cos 14 hese ane oat aiee Riise 2 32 Chapter 3 Booting Up and Shutting Down the NetServer About This Chapters iss seheng ara terse Lath kere teont digs alent ae ah ante aM 3 1 Booting the NetServet coos coug acy ren ee AP EES Ket a ead eee ees 3 2 Single User Mode vs Multiuser Mode 0 00000 e eee eee 3 2 Accessing the PROM Monitor n ass naan sae Made Lada aed eas 3 3 Entering Monitor Mode When the System is Powered Off 3 3 Entering Monitor Mode When the System is Running 3 4 Booting the Server at the Monitor Prompt 10 06 00 cian eo ya eed cs 3 4 The Default Boot Device 22 onda es va bake ged ave ee dae as 3 5 Using the Boot Command ccsiia a lonay gies Grogan ea eae Oe IA eine 3 5 Exiting Single User Mode to Multiuser Mode 5055 3 6 Booting From a CD ROM gn ws etic ay alba nh enol ph renren 3 6 Rebooting the Server to Multiuser Modes iy sacs ci dole ae in ee cee yas tals 3 8 DataGuard and ReDOGHING a6 igsn shyt leas act Acaecniaial Sacnrdeda whe aigbod alae 3 8 Commands for Shutting Down the NetServer 6 0 0 cece eee ee eee 3 9 SIA OW M sane aoaaa i Salar leet Sa bli ta sella at E A a tea Gn Bae Laka er dant 3 9 VG T ied ir AAG OE E E E E Daa eee heed 3 10 PEDOOE ori yten e aaae i aa r hail aan RL dieu edas Aisle aTe a a Maia SEN oe Sine dapat 3 10 fasth alt and fastboot cesor ieni Ca wile Gian gtk ty Set ee ute Gerda knoe 3 11 Chapter 4 NetServer St
282. lly disconnected from the ring before attempting the firmware download If a failure occurs the download process may be repeated until it succeeds ax_load_flash The following processor boards and SBus cards support the ax_load_flash utility a SPV a NPIV a FDDI and MLI 3 SBus adapter on the NP IV A Caution Downloading firmware to SP NP and MLT 3 boards with ax_load_flash requires you to place the NetServer in single user mode and reboot the NetServer once the download is complete and verified All other ax_load_flash requests can be done in single or multiuser mode Flash PROM Download Utility a 10 7 The following are examples of using the ax_load_flash utility a To display the boot and Flash firmware revision levels for a specified interface such as afddil type the following ax_load_ flash d afddil where d is the display only option fy Note Flash firmware is not downloaded with the d option 4 To download the contents of a specified file for example fddi 4 4 flash to a specified interface such as afddi0 enter the following ax_load_ flash afddi0 fddi 4 4 flash When a major release or patch release requires a processor board s Flash PROM to be upgraded the release note includes specific instructions that identify the board s to upgrade and provides the pathname to the file containing the PROM code Follow those instructions carefully For additional information refer to the ax_load_flash man page
283. lowing etc vpartab entry to create vp4 dev vp4 concat ad8c Run ax_loadvpar to load the virtual partition table 7 Use ax_clonefs to copy data from the bad drive to the new drive For example ax_clonefs dev vp3 dev vp4 8 Edit etc opartab to change the definition of vp2 and to comment out the reference to vp3 For example if the old drive is in slot 6 the new entry for vp2 and vp3 should read as follows dev vp2 mirrored vp4 dev vp3 concat ad6 c 9 Run ax_loadvpar to load the new etc vpartab 10 Use ax_hot_plug or ax_remove_device to remove the old drive Replacing a Drive that Has Multiple Partitions If the drive generating errors consists of multiple partitions instead of a one member mirrored partition follow this procedure to replace it 1 Obtain a spare drive the new drive that is at least the same size as the one to be replaced the old drive 6 22 a Version 1 10 System Manager s Guide AUSPEX D W AUSPEX 2 Choose an available drive slot for the new drive same SP as the old drive If you use ax_clonefs in step 6 the drive can be in any slot Gf Note If you use ax_sputil in step 6 the new drive must be supported by the Use ax_hot_plug to install the new drive 4 Insert the new drive and then type a carriage return Unexport and unmount all the file systems on the old drive so they cannot be modified while you copy data from the old drive to the new drive 6 Use ax_sputil
284. lowing line CORE_STAFF dl corescreener auspex com You can also direct the system core dump to a local user by adding his or her email address to the line For example CORE_STAFF dl corescreener auspex com jsmith abc com 2 8 a Version 1 10 System Manager s Guide W AUSPEX NetServer Configuration Commands Auspex provides a set of configuration commands that simplify the configuration process a NSconfig configures information about the NetServer its network interfaces and many of the services it supports You must use this command after installing the NetServer for the first time Note For specific information regarding ATM interfaces refer to the i appropriate documentation provided on the Auspex Premier Software Series CD ROM a SetupTty configures a TTY port on the HP for a serial device a SetupExec installs architecture dependent executables for diskless SunOS clients a SetupClient adds one or multiple SunOS client workstations to the list of clients that can boot from the NetServer or removes clients from the list you use the commands is important Use the commands in the order they are listed previously fy Note When you customize a NetServer for the first time the order in which The commands reside in usr etc and may be invoked from any directory However you must be root to invoke these commands Note If you use the NetServer as a router then after adding a new inte
285. ltiple mirrored partitions are being restored simultaneously To reduce the slowdown use the m option to assign a medium priority to the ax_mrestore process After lowering the priority for ax_mrestore the SP services other system requests faster but takes longer to restore a mirrored partition You cannot run ax_mattach or ax_mdetach while ax_mrestore is in progress ax_mrestore copies data in 192 KB blocks If there is an NFS request for writing to a block being copied by ax_mrestore the SP places the request in a queue and performs the NFS write after ax_mrestore finishes copying the 192 KB block Caution Do not hot plug devices using ax_hot_plug ax_remove_device or A ax_add_device on the SP that is running ax_mrestore Doing so may hang the system ax_vold 8 is the daemon that maintains a table showing the current state of all members of all mirrored partitions based on messages received from the SPs vp LFS and ax_mrestore ax_vpstat 8 displays information about all virtual partitions It shows the type and components of a virtual partition and the stripe size of a striped virtual partition If the partition is a member of a mirror it also shows the status of the partition See Figure 5 6 for an example of using the ax_vpstat command Virtual Partition Driver File and Commands a 5 11 ax_vpstat vp65 Concatenated ad2lc ad22c vp68 Mirrored vp66 Striped size 128K AC ad34c ad35c ad36c vp67 S
286. ly after installing the new drive Adding or Replacing a Drive a 4 27 AUSPEX D The system restarts the SP and attempts to attach the drive which takes approximately 20 seconds If the drive does not respond if it is not readable or if it is not properly connected the system generates an error message Caution If you see an error message indicating that the newly installed drive A is not found or that a SCSI interface error has occurred remove the drive from the slot immediately The sooner you remove the drive with the wrong SCSI ID the less likely the server will crash Repeat the drive installation procedure only after verifying or correcting the drive s SCSI ID 4 Follow the procedure for configuring a new disk drive Installing or Replacing a Drive When the Operating System is Shut Down 1 Refer to your NetServer s hardware manual for guidelines on handling disk drives 2 If the operating system is booted use etc shutdown lh to shut down the system 3 When the NetServer console displays a PROM monitor prompt and if you are replacing an existing drive remove the existing drive from the slot into which you plan to install the new drive 4 Install the new drive and reboot the system During the boot process observe the boot messages to verify that the system detects the presence of the new drive Unless the drive label has been corrupted the drive appears in the list of devices during the boot process Con
287. mber 5 21 attaching a member 5 9 automatic error recovery 5 13 changing from a concatenated partition 5 19 definition 5 5 detaching a member 4 26 expanding a one membered 5 18 moving 5 19 5 22 need for backup 8 1 removing a member 5 21 replacing a drive containing 6 20 unrecovered disk errors 6 10 used with write cache 5 5 with one member 6 22 Mirrored reads 5 5 Mirrored writes 5 5 Missing disk label 6 24 mkdir command 4 29 Mode echo 10 13 multi user 2 22 2 29 3 2 3 4 8 4 single user 3 4 3 6 8 4 Modem dial in 2 19 dial out 2 19 for remote diagnostics 10 12 Hayes compatible 2 18 port 2 18 Monitoring CPU utilization 9 2 I Os 9 2 network utilization 9 2 processor utilization 9 2 mount command 4 2 4 22 4 30 Mounting a CD ROM 4 5 Mounting file systems FP 4 18 HP 4 18 Moving a one membered mirrored partition 5 22 Moving a striped or concatenated Partition 5 21 Moving a two membered mirrored partition 5 22 mtcommand 8 7 8 22 Multiple histograms displaying 9 23 Multi user mode backups and restores in 8 4 booting 3 2 entering monitor mode 3 4 for running SetupClient 2 29 for running SetupExec 2 22 replacing adrivein 6 6 N Name conventions for disks and partition tables 4 10 devices 4 2 host 2 11 NIS domain 2 15 2 30 X terminal 2 11 Name cache 9 12 Naming a statistics file 9 19 Naming Exabyte 8200 tape drives 4 2 8 6 ncheck coomand 8 17 Netinterfaces kernel parameter 2 7 NetServer documentati
288. meter values Following the delimiter are the parameter values that are in decimal format Each line contains one ax_perfmon sample and the date and time the sample was collected Each number in a sample has a one to one correspondence with the parameter names Note An output file containing filtered statistics cannot be used as an input file to the ax_perfmon i file command Using a Filter File with ax_perfmon a 9 17 Displaying Performance Data in Histograms To display performance data collected by ax_perfmon in histograms execute the ax_perfhist command This section provides the following types of information about ax_perfhist a Overview of ax_perfhist 4 How tostart ax_perfhist and load statistics using the default settings 4 How to configure the histograms and windows to suit your needs A Information on parameters that control the types of statistics displayed in the histograms 4 Suggestions on how to use ax_perfhist efficiently Overview The ax_perfhist tool presents server statistics collected by ax_perfmon in an easy to read format ax_perfhist offers these advantages a The statistics are shown in histograms which are easier to read than the character oriented ax_perfmon displays For example at a glance you can identify the time periods during which the HP CPU load exceeded a predefined threshold a You can see the trend of server usage when playing back statistics using ax_perfhist Because the ax_perfmon sc
289. might help improve performance ease of use and ease of installation or configuration R Tools The tools icon identifies the tools you need to complete a task Terminology Throughout this manual certain terms refer to NetServer families generically Where distinctions are necessary the actual machine model is used The generic term Describes these models NS 7000 800 NS 7000 700 NS 7000 600 NS 7000 200 NS 7000 150 NetServer or server NS 7000 Model 800 NS 7000 Model 700 710 and 720 NS 7000 Model 600 602 610 612 and 650 NS 7000 Model 200 210 220 and 250 NS 7000 Model 150 All models Host Processor PROM monitor prompts are represented with HP gt throughout this manual Since this system prompt varies depending on the Host Processor model installed in the NetServer the prompt that appears on your system might be different A X X Getting Help Customer support and online help are described in this section Email Support Customer service and help through electronic mail are available to North American and International customers at the following address support auspex com North America Telephone Support To reach Auspex customer service for North America at any time dial 1 800 328 7739 International Telephone Support Customers within the countries and territory listed as follows can reach Auspex customer service at any time by dialing the appropriate telephone number Co
290. mmand 1 13 5 10 ax_mconvert command 1 13 5 11 5 20 5 21 ax_mdetach command 1 13 5 11 ax_mrestore command 4 27 4 32 4 33 5 11 5 13 ax_netstat command 1 12 10 14 10 15 ax_nfsd daemon 1 12 ax_perfhist options and arguments 9 46 starting 9 46 ax_perfhist command 1 12 9 1 Index 2 a Version 1 10 System Manager s Guide AUSPEX WD W AUSPEX 9 18 to 9 46 ax_perfmon command 1 12 9 1 9 2 to 9 17 ax_remove_device command 1 13 4 33 5 11 ax_sputil command 6 16 6 27 ax_startup daemon 1 12 ax_statd daemon 1 12 ax_statd2 daemon 1 12 ax_tapestats command 4 27 4 32 ax_timed daemon 1 12 ax_vold daemon 1 13 5 11 ax_vpstat command 1 13 5 11 ax_write_cache command 1 15 7 6 7 8 ax_write_cache command options 7 8 B b boot command 3 2 4 2 Backing up a clone partition to tape 8 15 a virtual partition 4 11 by cloning 8 12 full 8 2 incremental 6 26 8 2 more than 20 drives 8 10 on line 8 12 Backspace 2 10 Backup copy of root disk 6 26 Backup superblock 6 24 Bad spots on a tape 8 20 BADCHECKSUM 7 6 Battery for NVRAM on the Write Accelerator 7 3 7 6 BATTERY_LOW write cache status 7 6 Baud rate 2 19 Blank tape error 8 19 Blocking factor of tape 8 7 Boot alternate device 3 6 arguments 3 4 ask me option 3 6 command 1 13 default boot device 3 5 from backup root disk 6 26 from CD ROM _ 3 6 4 5 6 27 initializing the Write Accelerator 7 6 manual 3 4 messages 4 28 6 3 E 2 multi user mode 3 2 procedure
291. mmand to collect and save server statistics ax_perfmon o filename Alternatively use the s option to run ax_perfmon in silent mode in which ax_perfmon collects statistics without displaying them Note Use a consistent naming convention when saving ax_perfmon statistics per to a file We recommend you end a filename with a period followed by a constant text string for example server1 dat When choosing the input file for ax_perfhist specify listing only filenames with that particular text string suffix for example dat Refer to Loading Statistics From a File on page 9 21 for more information on selecting a file within ax_perfhist You can run ax_perfhist interactively or non interactively The non interactive mode is useful mainly when you include ax_perfhist in a script file Starting ax_perfhist Interactively Because ax_perfhist displays statistics graphically you can only invoke it on an X terminal or a workstation running OpenWindows or X11 Motif It does not work on an ASCII terminal All the sample screens in this guide are from the OpenWindows environment The appearance of the windows may be slightly different with X11 Motif To view the histograms on an X terminal or workstation on the network first set the DISPLAY variable at the shell prompt as follows setenv DISPLAY displayname 0 where displayname is the name of your display To invoke ax_perfhist on a host other than the one you are logged in to
292. n and then run ax_loadvpar so the system reads the revised table See Virtual Partition File on page 5 9 a For a mirrored virtual partition edit etc vpartab to comment out the entry for the member on the drive being replaced and redefine the mirror so it has only one member Finally run ax_loadvpar so the system reads the revised table See Virtual Partition File on page 5 9 4 26 a Version 1 10 System Manager s Guide W AUSPEX Caution When you hot plug a drive make sure no active tape drives are on A the same SP Trying to add a drive while a tape drive on the same SP is open for read or write access generates an error message To determine if a tape drive is busy use the ax_tapestats command If you try to run ax_tapestats on a busy device the message Device busy appears Using ax_hot_plug to Add a Drive to the Server Note For information on adding removing or replacing HDDA drives refer i to your hardware manual The ax_hot_plug command allows drives to be hot plugged that is added or removed without shutting down the NetServer You can run ax_hot_plug only on the local console If you try to run it from a remote login session the following error message appears ax_hot_plug must be on dev console to use this command Caution Never run ax_hot_plug on an SP that is formatting a drive or A running ax_mrestore or ax_clonefs Doing so may hang the system After completing the steps in the prece
293. n cnf Scan Dir Save Config Figure 9 18 Configuration Window 9 30 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX To apply a configuration file follow these steps 1 In the Configuration window type in the name of the directory that contains the configuration file Click on the Scan Dir button in the Configuration Window The file names of the configuration files appear in the scroll list Click on the configuration file you want to use The histograms which contain no data appear in the Data Display Window The number of histograms displayed depends on the configuration file Open the File Load Window While pointing to the input file name hold down the right mouse button A pull down menu appears In the pull down menu select Load All Histograms Statistics are loaded to the histograms which pertain to the parameter types specified in the configuration file For example if the configuration file includes two histograms one for HP hp usr_time and one for NPO cpu busy these two types of statistics from the selected input file are loaded to the histograms Figure 9 19 shows a Data Display Window that uses a predefined configuration file In this configuration file six histograms are displayed the second histogram uses continuous plot and the rest use discrete plot Auspex Systems Data Display Tool File 7 Properties v Print 7 Qty Hists 6 Histl Fri Feb 3 14 48 52 199
294. n ax_expand refer to Expanding a Concatenated or Striped Partition on page 5 15 a ax_loadvpar 8 reads vpartab and informs vp the virtual partition driver and the SPs of the current virtual partitions ax_loadvpar checks the validity of each entry in vpartab If errors are found ax_loadvpar reports the errors and does not load the table a ax_mattach 8 attaches a second member partition to a mirrored virtual partition that currently has one member It is useful for adding a temporary member to a mirrored partition for backup purposes The new member is automatically recorded in etc vpartab Note Before version 1 8 software the new member was manually entered D in etc vpartab Now that it is done automatically do not manually edit etc vpartab Editing the etc vpartab file to add a member partition such as vp6 to the mirrored partition vp41 and then running ax_mattach returns a warning message similar to the following WARNING vpartab edit not done partition already has members Please verify state of vp41 vpartab entry using ax_vpstat When this warning occurs ax_mattach attaches the new member but does not edit etc vpartab 5 10 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX ax_mconvert 8 converts a striped or concatenated virtual partition into a one member mirrored partition For more information on ax_mconvert refer to Expanding a One Membered Mirrored Partition
295. n be Printer which prints out the Data Display Window or File which saves the captured screen image to a specified file Saving a screen to a file If you save the data to a file two additional fields appear in the Print Options Window Dir Specifies the directory containing the destination file By default the directory is which is the directory in which you started ax_perfhist File Specifies the name of the destination file that contains the captured screen After specifying the output file click on the Apply button in the Print Options Window The file name appears in the Print Command field Printing a screen to a printer If you select xwd to capture the screen the screen image is piped to the xpr command which prints out the image If you select screendump to capture the screen ax_perfhist automatically uses the lpr command for printing Regardless of the command you use for screen capture always verify that the commands included in the Print Command field are in your path Types of Statistics Displayed by ax_perfhist As discussed in Loading Statistics From a File on page 9 21 you must specify a parameter when loading statistics to the Data Display Window The Parameter button in the File Load Window displays a pull down menu that organizes the statistic types by processor types The processors for which you can display statistics are HP NP FP and SP This section describes the parameters by processor boards F
296. n client workstations making NFS requests so it is important to remove the drive as quickly as possible fy Note On a quiescent SP all I O operations are halted NFS requests received 2 When prompted remove the drive Caution Do not remove a disk while it is spinning down Removing a disk A drive while it is still spinning can cause permanent damage to the drive Wait until the system console prompts you to remove the device 4 32 a Version 1 10 System Manager s Guide W AUSPEX 3 Press the Return key immediately after you have removed the drive The system restarts the SP Note Once you remove a drive from a given slot as described in this procedure the SP does not recognize any drive inserted into that slot until you use the ax_hot_plug command to inform the SP that a drive is to be added 4 Edit etc fstab to comment out entries in the table that refer to the removed drive If you detached a member of a mirrored striped or concatenated virtual partition as described in Before You Remove a Drive on page 4 31 you should recreate the member on another drive Note You can also use the ax_remove_device command for removing drives from the NetServer Refer to the ax_remove_device man page for more information Removing a Drive When the Operating System is Shut Down The following procedure describes how to shut down the operating system and remove a drive 1 Refer to the hardware manual for your NetSer
297. n is modified to contain the directory auspex which contains files specific to Auspex systems The root file system is defined in etc fstab as type 4 2 root auspex bin gt usr bin boot cdrom dev etc export export export2 home kadb lib gt usr lib lost found mnt sbin sys gt usr kvm sys tftpboot tmp usr var vmunix vmunix AX_GENERIC Figure 4 4 The Auspex root file system Disk Partitioning a 4 13 AUSPEX D var file system Figure 4 5 shows the server s var file system By default this file system is UFS mounted that is its file system type is defined as 4 2 in etc fstab for these reasons a var contains the var adm messages files to which system messages are written Mounting this file system as a UFS instead of an LFS file system ensures that the server continues to write system messages to var adm even when the File Processor has failed a If you mount var as an LFS file system using make or yppush to make a new NIS map might generate an error message and corrupt the maps that you try to build The message is as follows Can t bind master to send ypclear message to ypserv for map mapname on the master By mounting var as a UFS file system you can modify NIS maps without this problem adm crash log var lost found net preserve sadm spool tmp yp Figure 4 5 The Auspex var file system 4 14 a Version 1 10 System Manager s Guide W AUSPEX
298. n that is bigger than the one you want to enlarge This new partition must not be a mirrored partition Refer to Defining a Virtual Partition on page 5 14 for information about creating a virtual partition In this procedure the existing partition is named vp1 and the new partition is named vp2 The following entries are for vp1 and vp2 in etc vpartab vpl striped size 128k ad7va ad8a ad9a vp2 striped size 128k adl0a adila adl2a adl13a Defining and Reconfiguring Virtual Partitions a 5 17 2 Enter the following command to expand vp1 ax_expand s vpl vp2 Remember to use the s option otherwise the following error message appears ax_expand can t grow striped partition directly If the command executes successfully messages similar to the following appear About to mattach vp2 to vpl This will take some time Attach and sync complete about to grow vpl This may take some time dev rvpl1 1953776 sectors in 1908 cylinders of 16 tracks 64 sectors 1000 3MB in 120 cyl groups 16 c g 8 39MB g 3840 i g Initializing cg 119 Isolating tmnt for superblock adjustments tmnt released vpl now consists of vp2 vp3 is now free After this procedure vp1 becomes a mirrored partition with one member vp2 Although the contents of vp1 are copied to new physical partitions access to the file system mounted on vp1 is unaffected because the name of the virtual partition remains unchanged The virtual partition ex
299. n the same NP board and must be of the same network type For convenience in preparing this information you may copy and use the configuration worksheets provided in Appendix D You can explicitly set the primary hostname of one of the interfaces to the server hostname For example if the server name is Auspex and the server s IP address is the same as that of interface 0 then the name Auspex must be specified in the Server Hostname and AEO s Primary Hostname fields of the NSconfig form If the server s IP address is different from any of the interface IP addresses you must first specify the server s address in etc hosts before running NSconfig NSconfig gives a warning if the server hostname does not have a corresponding IP address in etc hosts G Note The server hostname must have a corresponding IP address in etc hosts Running NSconfig You can run NSconfig either in single user or multiuser mode 1 Enter NSconfig the letters NS must be capitalized NSconfig If you have not set the TERM variable for your terminal the server displays a numbered list of terminal types and prompts you to enter your terminal type Select the number from the menu that corresponds to the terminal type you are using NCD X terminal ANSI compatible terminal including VT220 and Link MC5 Wyse Model 30 50 Televideo 925 Sun Shell Tool Sun Command Tool DEC VT510 Other ArANnaA OB WNE kat hat E a wr vr vr YS Select a
300. nables or disables logins using the file etc ttytab and signaling init 8 If you specify ttyb as a modem port the baud rate you specify is set automatically in etc remote Note SetupTty does not set the type and status fields in etc ttytab you need D to modify these fields manually If you specify ttyb as a terminal and you want the port to default to a specific terminal type when you log in you must modify the terminal type in etc ttytab Follow these steps to run SetupTty 1 Log in as root The system must be in multiuser mode for this procedure 2 Run SetupTty SetupTty The SetupTty form appears as shown in Figure 2 4 The fields in the form are explained in Table 2 6 Auspex NetServer SetupTty Form Wed Sep 13 15 17 Which port lt ttya ttyb gt Port baud rate gt 9600 300 1200 2400 4800 9600 19200 Device connected to this port gt 0 no device connected dial out modem dial in modem local terminal Ike TAB CR or arrow keys to move among fields when done hit CTRL F to execute HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT WAL Figure 2 4 The SetupTty form Table 2 6 The SetupTty form fields Field Description Which port Port to set up ttya or ttyb Port baud rate Baud rate 300 1200 2400 4800 9600 or 19 200 By default ttyb is set to 9600 baud Device connected to this port Kind of device connected to the port being configured The default is 0 no d
301. nd Middle European time same as CET Central European time same as MET Poland Eastern European time Turkey Western Soviet Union People s Republic of China Republic of Korea Japan Singapore Hong Kong Republic of China Israel Tasmania Australia Queensland Australia Northern Territory Australia Western Australia South Australia Victoria Australia C 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table C 1 Time zone listing Continued Time zone name Time zone area Australia NSW New South Wales Australia Australia Broken Hill Broken Hill Australia Australia Sturt Sturt Australia Australia Yancowinna Yancowinna Australia Australia LHI Lord Howe Island Australia NZ New Zealand World Time Zones A C 3 Other time zones none include daylight saving time unless otherwise noted Table C 2 Additional time zone listings Time Zone Name Time Zone Area GMT UTC UCT Universal Greenwich GMT 1 GMT 2 GMT 3 GMT 4 GMT 5 GMT 6 GMT 7 GMT 8 GMT 9 GMT 10 GMT 11 GMT 12 GMT 13 GMT 12 GMT 11 GMT 10 GMT 9 GMT 8 GMT 7 GMT 6 GMT 5 GMT 4 GMT 3 GMT 2 GMT 1 Greenwich mean time Same as GMT Same as GMT Same as GMT Same as GMT 1 hour west of Greenwich mean time 2 hours west of Greenwich mean time 3 hours west of Greenwich mean time 4 hours west of Greenwich mean time 5 hours west of Greenwich mean time 6 hours west of Greenwich mean time 7 hours west
302. nd Tab key to move the cursor within the form Type a value in an input field which is enclosed with square brackets Also press the space bar to display the possible values for a select field which is preceded by an angle bracket gt For more information on cursor control refer to Using Configuration Command Forms on page 2 9 The log file for NSinstall NSinstall log is in the var log directory of the target disk A copy of the log also appears in the current root drive s tmp directory 6 30 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Note If the server was booted from the root disk not a CD ROM NSinstall i will not allow you to install file systems to the current root disk In this case you can only install file systems to a target disk in a slot other than slot 0 Auspex NetServer NSinstall Form Mon Sep 23 12 07 NSinstall OPTIONS Distribution source Release name Target disk slot Target Host Processor type gt Current NSinstall PARTITION ASSIGNMENTS Install root gt Yes Partition neufs first Install usr gt Yes Partition neufs first Install var gt Yes Partition newfs first Install usr openwin gt Yes Partition newfs first Set up export gt Yes Partition newfs first Set up home gt Yes Partition newfs first NSinstall SECURITY CONFIGURATION Remove the in etc hosts equiv to restrict rsh access to hosts users Ise TAB CR or arrow keys to
303. ndump You can specify that ax_perfhist uses xwd or screendump to capture the screen before sending the frame buffer image to a printer or saving the image to a file Note Be sure the directory containing xwd or screendump is in your path ax_perfhist does not enter the complete pathname for these commands in the Print Options Window Table 9 7 describes the differences between xwd and screendump Displaying Performance Data in Histograms a 9 39 Table 9 7 Differences between xwd and screendump xwd screendump Used in the OpenWindows or X11 Motif environment Used with SunView and OpenWindows No arguments need to be specified if you print the screen You can configure the x y X and Y dump to a printer If you print to a file you can specify the argument values the Screen Type option destination file name and the View option After clicking on the Print button in the Data Display Simply click on the Print button in the Data Window to invoke the print command the cursor on the Display Window to invoke the print screen changes to a plus sign Click in the window that command you want printed or saved Make sure the window is not obscured by other windows The default xwd command is as follows The default screendump command is as xwd frame xpr device ps lpr follows screendump x 9 y 38 X 661 Y 624 pssun s 7 5 10 0 0 5 0 5 lpr Print Command Output The Output option in the Print Options Window ca
304. ng Hardware Data Performing Software Probe Performing Hardware Probe Devices on SPO 012 3 67 8 9 10 Devices on SP1 012 3 4 5 6 Figure 10 1 illustrates how ax_config obtains the configuration information 10 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Software information Current configuration report Hardware information Primary operating system modules Storage Processor software IOP software VMUNIX Optional products Probes the system software to determine the revision levels of the primary operating system modules var adm contig report NetServer devices with Field Replaceable Unit FRU identifiers Extracts information from the file HP board SP board Write accelerator Disk tape and CD ROM drives NP board Probes the NetServer devices with FRU identifiers to determine their part numbers revision levels and serial numbers Configuration information displayed on the screen when you run ax_config Figure 10 1 How ax_config obtains configuration information Configuration Management a 10 3 Starting ax_config You must log in as root to use ax_config which runs in interactive or non interactive mode Interactive mode The following commands invoke ax_config interactively ax_config which displays the main menu ax_config v which displays status messages when probing the software and hardware before displaying the main menu If the
305. ng SBus interfaces On a system with more than one FP the FP Statistics screen initially display information for the first FP FPO in the system This command displays information for subsequent FPs On a system having more than one SP the System Summary screen the SP Statistics screen and the Virtual Partition Statistics screen initially display information for the first SP SPO in the system This command displays information for subsequent SPs From the System Summary screen it also displays disk activity statistics for disks on the subsequent SP Displays information for the next group of virtual partitions On a system having more than one NP the NP Statistics screen Summary Screen and NP Protocol Stats screen initially display network information for the first NP This command displays information for the subsequent NP From the System Summary screen it also displays disk activity statistics for subsequent NPs The s and N commands have no effects on systems with only one SP or NP Capturing and Displaying Real Time Performance Data a 9 5 Interpreting Statistics Screens This section describes the fields in each Perfomance Monitor screen Screen Header At the top of each Performance Monitor screen is a screen header similar to the following 1 0 System Summary NS7000 500 hostl Thu Aug 21 14 03 50 1997 The is the spinner that moves each time the screen is updated The number following the spinner is the time
306. nnection request Opens an ATM connection endpoint and gets connection information Receives data from an established ATM connection Sends data over an established ATM connection Example server ATM program using AAL 3 4 Example server ATM program using null AAL Sets batch size for null AAL interface ATM device control operations CD ROM control operations Compatible versions file for software packages Software dependencies file for software packages C Shell built in command A 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX iop 4 lfs 4 mtio 4 pkginfo 4 pkgmap 4 nit_if 4m vp 4 Auspex I O Processor Local file system Magnetic tape interface Display software package information Package contents description file STREAMS NIT device interface module Auspex virtual partition driver Devices and Networking a A 5 File Formats acct 5 ax_keys 5 exports 5 and xtab 5 fstab 5 installation report 5 mtab 5 passwd 5 raidtab 5 securenets 5 syslog conf 5 vpartab 5 Execution accounting file Auspex key and license file Directories to export to NFS Clients File system mounting tables List of installation tasks performed Mounted file system tables Password file RAID array table List IP addresses to which the NIS server should respond Configuration file for system log daemon syslogd Virtual partition table A 6 a Version 1 10 System Manager s Guide AUSPEX W W
307. nology NeTservices provides file services for the efficient sharing of computing resources among a community of desktop users Product Overview A_ 1 3 Table 1 1 NetServer features Continued Storage subsystems a A A Auspex system A management tools A A High performance storage subsystems Drives are organized in racks or drawers of seven drives The maximum number of drives supported depends on the NetServer model Software commands allow a Storage Processor to pause and restart if you need to hot plug drives that is remove or insert drives while the NetServer is running Auspex Virtual Partition Manager Mirrored striped and concatenated partitions provide an efficient and reliable way of using the disk space ina NetServer You can enlarge a striped or concatenated virtual partition without interrupting user access to the file system on the virtual partition Very Large File VLF support The size of an individual file is limited only by file system size for LFS files accessable over NFSv3 Both file and file system sizes may be over 2 GB Clients accessing such files must also have VLF capabilities The Auspex Performance Monitor which displays performance data for all processors in the system as well as disk subsystems and nonvolatile RAM NVRAM used for write caching The Auspex Virtual Partition Manager which comprises commands that display and manage the configuration of all virtual partitions Online hard
308. ns only to the client When executed SetupClient performs the following steps a Adds an entry for the client to etc ethers etc hosts etc bootparams and etc exports a Builds a root directory to be mounted by the client machine a Builds a swap file to be used as swap space by the client machine a Uses make and push on the NIS maps etc ethers etc bootparams and etc exports if the server is an NIS master a Runs the exportfs program a Creates a link to the primary boot program in tftpboot Note Analyze the client root and swap needs of your environment before i deciding how to partition the disks on your server Refer to Example of File System Distribution on page 4 11 for more information on setting up partitions for clients root and swap directories After running SetupClient to add a client you can then boot that client machine 2 28 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Running SetupClient 1 Determine the IP address and Ethernet address of the client One way to determine the client s Ethernet address is to power on the client and watch it display the power on messages For convenience in organizing this information copy and use the configuration worksheets in Appendix D If your network uses NIS and if the NetServer on which you run SetupClient is not the NIS master follow these steps a Change the NIS text maps of these files on the NIS master ethers
309. nts can resume access to home without too much delay File System Isolation Caused by Corruption This section provides an example showing how to recover from a file system that was isolated because it was corrupted The disk on which the file system resides is not at fault 1 Enter the following command to determine which file system has been isolated grep isolated var adm messages Nov 21 18 10 33 host3 syslog ax_isolated Filesystem dev ad6 mounted on wrld apps frame_3 0 is isolated Search var adm messages to determine if file system isolation was caused by the SP s failure to reach the disk File system corruption is the cause if var adm messages contains no SP error messages or FP error messages indicating a permanent disk error In this example the FP messages logged are as follows 6 18 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX 8 9 Nov 21 18 10 05 host3 FPO 1012 wrld apps frame_3 0 bad block Nov 21 18 10 18 host3 FPO bad block 872415232 ino 53765 Nov 21 18 10 23 host3 FPO bad block 805306819 If the SP controlling the file system uses write acceleration flush all unwritten data from the write cache ax_write_cache s 0 flush ax_write_cache s 0 purge The write accelerator is automatically turned off after these commands Unexport the file system so clients trying to mount it do not hang For example exportfs u wrld apps frame_3 0 Because the isolated file syst
310. number for your terminal type gt gt For example select 7 fora DEC VT510 terminal or 2 for a Link terminal NetServer Configuration Commands a 2 11 AUSPEX QD If you select 8 Other you are prompted to enter the terminal type as shown in the following example The type must be in etc termcap and the terminfo database Enter terminal type must be in both etc termcap and terminfo gt gt 3 After you enter a valid terminal type NSconfig displays the following notice that warns of changes in NSconfig functionality from previous system software releases With System Software Release 1 10 NSconfig automatically does the configuration file updates and system modifications without requiring a reboot in most cases The exceptions are noted in the message WARNING DO NOT ALTER THE IFCONFIG LINES OF etc rc boot NSconfig relies heavily on them to function correctly PLEASE NOTE NSCONFIG NOW EXECUTES IFCONFIG s OF NETWORK INTERFACES WHETHER RUN IN SINGLE OR MULTI USER MODE A reboot after completion of NSconfig will only be required if run in multi user and when there is a change in IP address or netmask of an interface or when NIS type is changed to master or slave Want to proceed with this y n 4 The NSconfig form displays Figure 2 2 on page 2 13 shows an example of the form Complete the form Table 2 3 on page 2 14 describes
311. ny slot except slot 0 which is reserved for the root disk drive and in an HDDA drawer Before You Start the Drive Installation Procedure Follow these steps before you start drive installation while the operating system is booted Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 1 Log in as root on the system console 2 If you are replacing an existing drive make sure no partition on the drive is open for read or write access For example verify that a no file system is being dumped a no process such as a database is using a raw partition a no swapping is being done to the disk a no partitions are members of other virtual partitions Use the ax_diskconf command to get a report that maps all file systems to the physical partitions or virtual partitions they reside on Depending on the type of partition you may need to do one or more of the following a Ifa virtual partition or a CD ROM partition is mounted first unmount it a For striped or concatenated virtual partitions that are not part of a mirrored virtual partition edit etc vpartab to comment out the virtual partitio
312. o Yes Server goes to single user Server goes to mode single user mode No Rees Can you fix disk error Yes Events that take place automatically Gy Write cache state gt L Operator intervention Fix disk error and flush data to disk Enable write cache Write cache is not enabled automatically if you have purged or flushed data Use ax_write_cache to enable it T On gt Figure 7 1 Write cache changing states after server power on Write Cache States and Commands a 7 7 Note Because write caching is not supported under UFS all write messages to root var and usr file systems bypass the write cache Consequently the state of the write cache does not affect the system boot files A utility command ax_write_cache 8 manages the state of the write cache You must be root to invoke the command The syntax for the command is as follows ax_write_cache v s SP_number option The v argument executes the command in verbose mode If no SP numbers are specified the command applies to all SPs in the system Processor numbering starts from 0 The command options which are described in Table 7 1 are for changing the state of the write cache or printing out information about the write cache You can use more than one option per command Table 7 1 ax_write_cache command options Option Description init Initia
313. o BajaSur Eastern time zone U S Central time zone U S Mountain time zone U S Pacific time zone U S Pacific time zone U S with proposed changes to daylight saving time near election time in presidential election years Yukon time zone U S Eastern time zone U S no daylight saving time Mountain time zone U S no daylight saving time Hawaii Same as U S Mountain Newfoundland Atlantic time zone Canada Eastern time zone Canada Central time zone Canada Central time zone Canada no daylight saving time Mountain time zone Canada Pacific time zone Canada Yukon time zone Canada Baja Norte time zone Mexico Baja Sur time zone Mexico Time Zones A C 1 Table C 1 Time zone listing Continued Time zone name Time zone area Mexico General South America Brazil East Brazil West Brazil Acre Brazil DeNoronha Chile Continental Chile Easter Island Europe GB Eire WET Iceland MET CET Poland EET Turkey W SU Asia PRC ROK Japan Singapore Hongkong ROC Israel Australia and New Zealand Australia Tasmania Australia Queensland Australia North Australia West Australia South Australia Victoria Mexico general time zone Eastern Brazil time zone Western Brazil time zone Territory of Acre time zone Brazil Fernando De Noronha time zone Brazil Continental Chile time zone Easter Island time zone Chile Great Britain and Ireland Western European time Icela
314. o reserve The computation takes the number of NP boards in the system multiplies that by six and then multiplies that by 56 KB For example a system with four NPs would get a netinterfaces working value of four times six or 24 The amount of memory reserved for interfaces would be 56 KB times 24 In general the most efficient value to insert for netinterfaces is the actual number of interfaces in the system A system with four NPs and three interfaces on each would need a netinterfaces value of 12 using only half as much memory as the default computation The maximum netinterfaces value recognized by the system is 64 for each NP up toa total system maximum of 256 For example if you give a netinterfaces value of 100 on a system with one NP the system reserves memory for a value of only 64 A site s use of interfaces may call for more memory than the guideline given here in which case you can increase reserved memory by changing the netinterfaces parameter Remember that any change to this parameter is effective only after rebuilding and rebooting the kernel See the Rebuilding the System Kernel on page 10 9 for rebuilding instructions Mailing Site Reports to Auspex If your site can communicate with Auspex using email Auspex recommends that you edit var spool cron crontabs root to ensure that weekly customer site reports are sent to Auspex These reports contain uptime information NFS statistics and so on Add the following
315. o the terminal type you are using NCD X terminal ANSI compatible terminal including VT220 and Link MC5 Wyse Model 30 50 Televideo 925 Sun Shell Tool Sun Command Tool DEC VT510 Other Arya Ow WN EF Select a number for your terminal type gt gt For example select 7 fora DEC VT510 terminal or 2 for an ANSI compatible terminal If you select 8 Other you are prompted to enter the terminal type as shown in the following example The type you enter must be in etc termcap and the terminfo database Enter terminal type must be in both etc termcap and terminfo gt gt After selecting the NSinstall option and entering your terminal type if necessary a message similar to the following appears xxx WARNING All data will be destroyed on the target disk NSinstall from the boot CD ROM y n After selecting y the NSinstall form appears Refer to the next section for information about the fields in the form NSinstall Form NSinstall displays the form shown in Figure 6 1 Table 6 1 explains the meaning of each field Some fields in the form appear only if you specify Yes for a related field For example the prompts for the partition and newfs appear only if you want to install or set up the related file system These fields are indented in the table Be sure to read the form carefully before executing it some default values in the form may not be applicable and need to be changed Use the Arrow keys a
316. oard IOPO FPO CPU B 1010 disks hweng home file system full IOP1 NP1 CPU A 1114 ae4 No carrier transceiver cable problem For more information on error logging see Chapter 8 in Sun s System and Network Administration Guide In addition to messages being logged to var adm messages Auspex specific notice informational and diagnostic messages are logged to var log auspex messages Dealing with Server Problems a 10 11 Using the Continue Command At the monitor prompt the continue co command resumes execution of the operating system This command is especially helpful if you unintentionally enter a Break on the console because it allows you to continue without rebooting the system Note A Break signal may be generated when you turn the console terminal off and then back on In this case you can also use the co command to continue without rebooting The continue command which can be abbreviated to co is executed in the following format HP gt co virtual_address The virtual_address argument allows you to select the virtual address of the instruction to be executed when the program resumes By default the program resumes at the instruction located at the address pointed to by the program counter Note The co command may not be able to restore the system to the state in effect at the time of the Break depending on how long you waited before entering co If co fails to restart the operating system re
317. ochost1 foobar rlogin dochost1 dochost1 4 From dochost1 set the display environment for the shell type For example for the C shell enter the following command to send the display back to foobar dochost1 setenv DISPLAY foobar 0 For the Bourne shell or Bourne compatible shell follow this example dochost1 DISPLAY foobar 0 export DISPLAY 5 Enter the following on dochost1 to start the online documentation utility o dochost1 ax_docs This script invokes the documentation Main Menu Overview Auspex NetServers support online documentation through the licensing of Acrobat software from Adobe Corporation The online documentation contains system hardware system software and optional software information The software information is in the online version of the System Manager s Guide the hardware information is extracted from the hardware manual that ships with your server Figure B 1 shows part of the Main menu Click on a menu choice to see documentation Online Documentation a B 5 AUSPEX ia a Awe gt gt s gt gt of Bal aa Main Menu System Manager s Guide Premier Software Series Menu System Hardware Menu Auspex Homepage Figure B 1 Main Menu window for Auspex online documentation The entries on the Main Menu are a System Manager s Guide This entry brings up a window with the System Manager s Guide a Premier Software Series Menu This entry brings
318. ocumentation and pull down the File menu when the document appears The Print option allows you to print a particular page a range of pages or the entire document The Print Setup option allows you to select a paper size including A4 Alternately to print the whole lt opf product gt document mount the Premier Software Series CD and use the lpr s command to print 100btatmbackupdguardftpraid ps in the psdocs directory A XXi W AUSPEX Overview of NetServer Features About This Chapter This chapter provides an overview of the NetServer and describes the features specific to the NetServer s hardware and software architectures In addition this chapter points out the distinct advantages such as balanced client traffic ease of administration and scalability offered by the Auspex environment The last section lists changes to SunOS Version 4 1 4 that Auspex has made in the NetServer software Overview of NetServer Features a 1 1 Product Overview The Auspex NetServer is specifically designed to deliver industry leading performance and reliability that meets the demands of large networks in real world production environments With its unique Functional Multi processing FMP architecture the NetServer delivers high speed network I O performance The NetServer is also highly scalable you can improve its performance by adding processor boards and storage devices as the number of network clients grows In addition t
319. ode a If file systems are mounted on the disk fsck fails to check the file systems and lists them as file systems that have unexpected inconsistency The system enters single user mode after displaying the following error message Reboot failed help Note When the system cannot see a drive during reboot the SP does not generate an error message This is the only case in which the SP does not report a disk error a If the file systems are mounted on a physical partition of the disk fsck notes the missing device but the system will reboot successfully a If the disk is part of a RAID array ax_write_cache may display messages related to the RAID array changing state and fsck notes the missing device but the system will reboot successfully Disk Problems at System Reboot a 6 3 Is the Problem Caused by Poor Disk Installation Sometimes the system cannot detect a disk because it is not properly installed The problem can be easily detected and corrected with this procedure 1 In single user mode enter the ax_hot_plug or ax_add_device command 2 When the system prompts you to insert or replace the specified disk drive remove the drive 3 Reinsert the drive in the drive slot and press the Return key If the SP can successfully attach the drive the problem was caused by a poorly installed disk 4 Run fsck p to check all the file systems and then press Ctrl D to go to multiuser mode Go to the next section if th
320. of Greenwich mean time 8 hours west of Greenwich mean time 9 hours west of Greenwich mean time 10 hours west of Greenwich mean time 11 hours west of Greenwich mean time 12 hours west of Greenwich mean time 13 hours west of Greenwich mean time GMT 12 with daylight saving time 12 hours east of Greenwich mean time 11 hours east of Greenwich mean time 10 hours east of Greenwich mean time 9 hours east of Greenwich mean time 8 hours east of Greenwich mean time 7 hours east of Greenwich mean time 6 hours east of Greenwich mean time 5 hours east of Greenwich mean time 4 hours east of Greenwich mean time 3 hours east of Greenwich mean time 2 hours east of Greenwich mean time 1 hour east of Greenwich mean time C 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Configuration Worksheets About This Appendix This appendix contains a set of four worksheets for system configuration tasks a The Host Information worksheet compiles information for host configuration with the NSconfig command and is useful when using the NSinstall command a The Client Information worksheet compiles information for client configuration with the SetupClient commands The Disk Drive Information worksheet may be helpful during disk partitioning The Virtual Partition Information worksheet compiles information about the virtual partitions defined on your server These worksheets are designed to be duplicated for your convenience Confi
321. om the NetServer to check the jumper setting Jumper settings are described in the hardware manual Note If you are using a CD ROM drive that requires a CD ROM caddy and if i for some reason you have inserted an empty caddy into the drive you must use the eject button to remove the caddy you cannot use the eject command 4 6 a Version 1 10 System Manager s Guide W AUSPEX Disk Partitioning Making decisions about disk partitioning and assigning client space are tasks that are vital to the efficient operation of your system This section describes the default disk partition schemes available for Auspex disks and illustrates how file systems are distributed by default among available partitions on the root drive fy Note CD ROM drives are read only devices and cannot be partitioned Disk Partition Tables Each disk can be divided into seven or fewer partitions named partition a b d e f g or h Partition c is reserved for the entire disk The partition sizes are defined in the disk label which is the disk s first sector The NS 7000 800 and NS 7000 700 Series NetServers support 1 GB 1 35 GB 1 76 GB 2 GB 3 GB 4 GB and 9 GB disk drives the NS 7000 200 supports the 1 76 GB and 4 GB drives only Each disk size referenced here refers to the disk capacity after formatting The HDDA drive drawer supports only 9 GB drives Each disk size has three partition tables to choose from default stripe and root The
322. om the source partition to the A read request can be processed by either clone NFS read operations are performed only on member of a mirrored partition the source partition not on its clone Write operations are performed on both partitions The clone and the source partition can reside on Both members of a mirrored partition must reside different SPs on the same SP Both the clone and source partitions can be either Both members of a mirrored partition must be virtual or physical virtual partitions The purpose of creating a clone is to back up a large partition without taking the partition offline Backing up file systems by cloning offers these advantages a You can back up a large virtual partition to another partition on another SP Without cloning online backup requires a mirrored partition to be created on the same SP as the virtual partition that needs to be backed up You may not be able to find enough space on disks that are on the same SP a You can create a clone to take a snapshot of a mirrored partition When you back up the files from the clone partition to tape the members of the mirrored partition continue to function uninterrupted Without cloning you must detach a member from the mirrored partition to back up files to tape and as a result temporarily lose the advantages of mirroring The command used for cloning a partition is ax_clonefs Its syntax is as follows ax_clonefs b p source_partition clon
323. ommands a 2 25 5 After you fill in all the fields in the SetupExec form execute the form by typing Ctrl F The system displays messages similar to the following which shows sun4c executables being loaded 1 WM x O Crac xX crac X crac xX crac xX crac X crac X crac X crac x crac X crac xX crac xX crac xX crac X crac xX crac X crac xX crac xX crac xX crac X trac le x K H crac X crac x x Crac HHP PP Phe PPP PP Pee eee crac Hh H Hi 0 Crace Cract ai ti ti ti ti Gi ti ti Ei ti ti ti ti ti ti ti Ba ti ti ti ti ti ti ti tupExec Ng Ng ng ng ng ng ng ng ng ng ng ng ng ng ng ng ng Ng Ng Ng Ng Ng ng ng ng ng reading distribution CD ROM CD ROM file CE a E o S E v tte tet A EAS E tet tet st co ct Gh Gh ch ach sah E ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne ne SunOS Sun0s Sun0s Sun0s SunOS Sun0s Sun0s Sun0s Sun0s SunOS Sun0s Sun0s Sun0s Sun0s Sun0s Sun0s Sun0s Sun0s Sun0s Sun0s SunOS SunOS SunOS SunOS Sun0s SunOS 4 1 eN ae Ae s S ae KR KR KKK DB BR BR WB WA WWWW WWW WWW WW WWW WwW WwW WwW Ww SN oB BWW WWW WW Ww Poe su su su su su su su su su su su su su su su su su su su su su s
324. on about protocols used in TCP IP See services 5 Defines site specific mail handling procedures See sendmail 8 Defines any special logging requirements for system messages that is assigns special log files for certain messages See syslog conf 5 Defines serial ports etc ttytab is configured automatically by NSconfig and SetupTty See ttytab 5 Defines virtual partitions See vpartab 5 If your site uses NIS these files are used only at boot time before NIS is up Kernel Parameters A kernel has three site configurable parameters The parameters in the generic kernel usr kom sys aushp conf AUSPEX1 are maxusers and netinterfaces Edit the file AUSPEX1 to change the parameters Maxusers determines the approximate size of kernel data structures Read the comments for maxusers in AUSPEX1 for guidance Additional TTYs In order to provide logins for additional TTYs in the generic kernel usr kom sys aushp conf AUSPEX1 change the parameter from pseudo device pty to pseudo device pty255 The default number of TTYs is 16 2 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The netinterfaces value determines how much memory is reserved for the exclusive use of interfaces Each increment in the netinterfaces value adds 56 KB of memory to the amount reserved for interfaces A netinterfaces value of 0 zero which is the default causes the system to compute its own number for the amount of memory t
325. on using ax_clonefs as in this example ax_clonefs vp6 vp265 In this example vp6 is the source partition and vp265 is the clone partition For the source partition you can also use the file system name For example if the usr openwin file system is mounted on vp6 enter the following ax_clonefs usr openwin vp265 To back up your root disk use ax_clonefs for each partition on the root disk For example if you are copying the root disk to a disk in slot 4 enter the following command to copy the a partition ax_clonefs ad0a ad4a 8 14 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The ax_clonefs command causes the NetServer to perform the following a Copy data from the source partition to the clone partition During the copy all NFS write operations are performed on both the clone and source partitions and read operations are performed on the source partition alone Aa Lock the file system and sync the meta and user data to ensure consistency between the source and clone partitions a The clone partition is detached from the source partition Caution When the SP is running ax_clonefs do not hot plug devices using A ax_hot_plug ax_remove_device or ax_add_device on the SP Doing so hangs the system Also never run ax_clonefs and ax_mrestore on the same mirrored partition simultaneously The system might crash if you try to clone a mirrored partition and restore the contents from one member to anoth
326. on xvii NetServer serial number 2 17 NetServer throughput improving with the Write Accelerator 7 3 Network monitoring the utilization of 9 2 server software structure 1 10 Network interfaces managing 10 14 setting memory use 2 7 Network Processor 1 6 functions of 1 6 optional products 1 8 newfs command for a physical partition 5 7 for root partition 6 26 6 28 initializing file systems 4 10 4 29 NFS NetServer compatibility 1 4 Index 10 A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX nfsstat command 1 12 reads and writes on a clone partition 8 15 workers 9 9 write acceleration 7 3 nfsstat command 1 12 10 14 10 19 NIS client 2 15 database 2 28 domain name 2 11 2 15 2 30 hosts database 2 28 master server 2 4 2 15 2 29 service 2 15 2 30 service type 2 11 slave server 2 4 2 15 None write cache status 7 6 nonvolatile RAM NVRAM 1 6 No rewind option 4 2 8 3 North America customer support xx nosuid option in etc fstab 4 20 Notifications E 8 NP statistics 9 9 NP Statistics screen 9 8 NSconfig after running NSinstall 6 34 command 2 10 2 12 2 17 files modified by 2 17 form 2 13 form fields 2 14 how to run 2 11 NSinstall 2 2 6 29 for a spare root drive 6 31 form 6 31 form fields 6 31 NSinstall form 6 30 NSupdate 2 2 6 29 NVRAM nonvolatile RAM 1 4 1 6 7 3 O Off write cache status 7 6 On write cache status 7 6 ONC NFS_ 1 4 One membered mirrored partition
327. ons with a local disk simply mount the exported file systems they need to access Also if clients require usr from the NetServer you must execute SetupExec For more details on this command refer to the SetupExec 8 man page 2 20 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX What Does SetupExec Do SetupExec installs architecture dependent executables for diskless SunOS client workstations from either a tape or CD ROM using either a local drive or a drive ona remote host It adds a supported SunOS architecture to the NetServer For example the SunOS Version 4 1 3 Sun 4 executables are loaded into export exec sun4 sunos 4 1 3 and a Sun 4 client mounts this directory as the client s usr file system Figure 4 8 on page 4 17 provides an illustration of the export directory structures SetupExec updates two database files as follows etc exports SetupExec updates etc exports see the exportfs 8 man page to export the executable directories it has installed Check this file to verify that the directory was exported to the correct groups etc hosts SetupExec updates etc hosts to add any remote CD ROM host or tape host names and IP addresses not currently in the database Things to Remember When Running SetupExec Before running SetupExec be aware of the following a Run SetupExec in multiuser mode only by the root user a Auspex does not supply the architecture dependent executables purchase the exec
328. onsole add or remove the specified drive Caution Do not remove a disk while it is spinning down Removing a disk drive while it is still spinning can cause permanent damage to the drive Wait until the system console prompts you to remove the device 3 Type a carriage return immediately after the specified drive is added or removed After typing a carriage return the system restarts the SP If a drive was added the system attempts to attach the drive which takes approximately 20 seconds When prompted at the system console repeat steps step 2 and step 3 for each drive specified in the ax_hot_plug command The command addresses each drive in the order that it is entered in the command line If the drive or drives do not respond if they are not readable or if they are not properly connected the system generates an error message Caution If you see an error message indicating that any newly installed drive A is not found or a SCSI interface error has occurred remove the drive from the slot immediately Such an error message is possibly caused by an incorrect SCSI ID The sooner you remove the drive with the wrong SCSI ID the less likely the server will crash Repeat the drive installation procedure only after you have verified or corrected the drive s SCSI ID 4 Follow the procedure for configuring a new disk drive on page 4 28 5 Edit etc fstab to comment out entries in the table referring to the removed drive If you detach
329. ontrolling the isolated file system uses write acceleration flush all data from the write accelerator ax_write_cache s 1 flush ax_write_cache s 1 purge The write accelerator is automatically turned off after these commands Use fsck to check the isolated file system which fixes any damage to the file system caused by the bad disk sector For example fsck dev rvp277 Start fsck of dev rvp277 CONTINUE yn y dev rvp277 Last Mounted on home Phase 1 Check Blocks and Sizes x Phase 2 Check Pathnames DIRECTORY CORRUPTED SALVAGE yn y Phase 3 Check Connectivity xx Phase 4 Check Reference Counts xx Phase 5 Check Cyl groups FREE BLK COUNT S WRONG IN SUPERBLK SALVAGE yn y 158427 files 4389754 used 1154922 free 2114 frags 144101 blocks 0 0 fragmentation Add the file system back online For example ax_fsutil release home Successfully released filesystem mounted on home Reenable the write accelerator if one is used on the SP For example ax_write_ cache s 1 enable The file system recovery procedure is now complete Note Whether you unexport the file system while fixing the disk sector i depends on how many clients are affected by the file system and how long the repair takes In this example home is used by all clients which all hang if you do not unexport home However because reassigning the bad block takes only a short time clie
330. or a remote CD ROM drive you must have root permission on the remote host To obtain root permission edit the file rhosts on the remote host and add an entry in this format hostname root where hostname is the name of the Ethernet interface through which you send requests to the remote host For example if your local server is named Engineering and Engineering e0 is the interface attached to the network to which the remote host is connected specify Engineering e0 root in rhosts 2 22 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Auspex NetServer SetupExec Form Fri Feb 2 15302 Architecture type ARCH SunOS version VERS FileServer partition EXPORT export1 Path to executables C EXPORT exec ARCH sunos YERS Path to kernel executables tEXPORT exec kvm ARCH sunos YERS____ Path to usr share files C EXPORT share sunos VERS Distribution medium CDROM Tape TEE Drive location Local Remote gt Local Tape device name rast9_ e g rastl rast2 rast209 Select the distribution medium HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VYAL Figure 2 6 Sample SetupExec form for a local tape device Auspex NetServer SetupExec Form Fri Feb 2 15 00 Architecture type SARCH SunOS version VERS FileServer partition EXPORT C export1 Path to executables EXPORT exec ARCH sunos YERS Path to kernel executables
331. or the NP and SP parameters processor numbers or network interface numbers are used in the display to identify the processors or interfaces from which statistics were collected For example the NPO e0 packets sec parameter pertains to Ethernet interface 0 on NPO 9 40 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX The Parameters menu lists the maximum number of boards Your server may not have all the processor boards listed in the menu Note The ax_perfhist parameters for network interfaces disks virtual partitions NFS operations and the write cache on the SP are per second not per interval as in ax_perfmon HP HP cpu usr_time HP cpu sys_time HP cpu idle_time NP NP cpu busy NP np NFS workers NP np NFS jobs queued NP np NFS busy workers NP np NFS nfs vers 2 over UDP NP np NFS nfs vers 2 over TCP NP np NFS nfs vers 3 over UDP NP np NFS nfs vers 3 over TCP NP np FIP puts NP np FITP gets NP np FIP put bytes NP np FIP get bytes NP np FIP workers NP np UDP dgrams in NP np UDP dgrams out NP np UDP cksum errs NP np TCP segs in NP np TCP segs out NP np TCP segs w error NP np TCP curr estab conns NP np TCP active opens NP np TCP passive opens NP np TCP workers HP CPU utilization time in application HP CPU utilization time in kernel HP CPU idle time NP CPU utilization Number of NFS workers on the NP Number of NFS jobs queued on the NP Number of busy NFS workers on the NP
332. orage Devices and File Systems About This Chapter aeaaaee EA e teh dog Liss a E a ts ai 4 1 Device NATES e E ech cart es ial a ae AA Atl ERE S 4 2 Disk Drives pet Aedes bas con heer ea Ne ah ore ee aon ae RANE ON ie Ae as Alls 4 2 CD ROM Drives cs c04 tne act pagan et ie Bhd ee SA ee ae 4 2 Tape Drives e 2 2 ccc secede hates wine tetas eh en ane aor aetna ata ta S 4 2 Raw Disk Partitiontt c0 02 2fecie facp eee Dad ha A Sa oe ena 4 3 Weviee Special Files riziere tie esa valet eet luathise ani a eth turer lear eae ay 4 4 Usine CD ROM Drives ieser eriari sr Sey ation ke genes anced hol NA aa ie hay 4 5 Mounting or Unmounting File Systems on CD ROM 4 5 Removing a CD ROM From the NetServer 0 000000 0 eee ee 4 6 Disk Paf toNiNg risanke aa E E ee Alsat E EEL hated 4 7 Disk Partition shales vis eiyan ati e E Ghat te Rend EAA A 4 7 Distribution of File Systems on Default Partitions 55 4 10 Default File Systems on the Root Drive 0 00 00 renun 4 10 Default File Systems on Default Drives 0 000000 4 10 Natt File OY StCMN Sc feu etd ie Rae Gon hice te eiren ani eae es 4 10 Example of File System Disttibutonc o itduwnav ie cahidee ence ets 4 11 Auspex File System Structure te cist Cid ashes bus da A hand edn Casta aes 4 12 Mounting File Systems on the FP and HP 000 4 18 File Systems Mounted on the File Processor 0 000 e eee eee 4 18 File Sys
333. ormance statistics out to a file using the o option Saving statistics only for items that interest you reduces the size of statistics files For more information on creating a filter file refer to Using a Filter File with ax_perfmon on page 9 16 i file Displays data from the specified input file instead of gathering it from an active system p period Specifies the interval in seconds between screen updates default is one second If used in combination with the i argument the p argument controls the screen update rate but does not change the rate at which data was originally gathered In this way you can replay saved data quickly If used without the i argument the p argument controls the interval between data retrievals t time Specifies the duration in seconds that ax_perfmon runs Without the t option ax_perfmon runs continuously either on the screen or to a file Some of the most commonly used ax_perfmon commands are listed as follows a To display real time data about the local host at one second intervals ax_perfmon a To display real time data about host netserver1 rlogin netserverl ax_perfmon a Tosilently gather data about the local host at five minute intervals and store the data in the file sysadm perf data ax_perfmon s p 300 o sysadm perf data gt amp tmp myerrfile amp The file myerrfile stores any error output messages gathered by ax_perfmon a To play back the data collected in the previ
334. ot need to supply a date because the current time at which dump starts is automatically added in etc dumpdates Use the T option only when you use a script to dump a file system that was cloned by ax_clonefs for example you might want to clone several partitions before dumping them to a tape In the script create a timestamp using a date string in order to specify the time when ax_clonefs finishes copying the file system Include the timestamp along with the T Online Backup a _ 8 15 A option in the dump command Doing so ensures that dump enters the proper date string in etc dumpdates for each partition The timestamp should use the following date string date a th d T 19 y The following is an example of a dump command with the T option where date is the timestamp created in the script dump OTdsbfu Sdatel 141000 11500 126 dev rast4c dev ad0c Use the m option together with the name of the source partition in ax_clonefs This option forces dump to record in etc dumpdates the partition name you specified instead of the name of the partition being dumped that is the clone partition Using the source partition name ensures that the next incremental dump finds the information from etc dumpdates about the source partition that has been backed up The following is an example dump Odsbmfu 65000 6700 126 dev rvp10 dev rast4 dev rvp11 In this example the partition name specified with the m option is dev rvp10
335. ot to multiuser mode Change password Is NetServer No an NIS master Yes Verify NIS operations on the NIS master Verify files in etc No NetServer is an NIS client R age Is NetServer un ypinit m an NIS slave Yes Run ypinit s Reboot NetServer Modify site specific files in etc Figure 2 1 Flowchart for configuring the server for the first time Changing the Root Password After running NSconfig and exiting to multiuser mode enter the passwd root command to create a root password This step is very important for system security New Installations a 2 3 Configuring the Server to Be an NIS Master or Slave Server If you do not use the NIS name service at your site skip to Modifying Site Specific Files on page 2 5 Configuring the NetServer to Be an NIS Master If you use the server as an NIS master follow these steps 1 4 For more information about NIS refer to Sun s System and Network Administration manual Verify that the current versions of the NIS related files etc ethers etc hosts and etc bootparams are located in etc on the NetServer Verify the domain name using the following command bin domainname Run ypinit to build NIS maps for master and slave configurations usr etc yp ypinit m For more information on ypinit refer to the man page Reboot the NetServe
336. otocols Files Protocols Files Partitions SBus SBus HUTT a CCl Network Connection Independent SCSI Channels Ethernet 100Base T FDDI ATM 6 SCSI Model 200 and 700 Series Figure 1 1 Hardware implementation of Functional Multi processing architecture Hardware Implementation of the FMP Architecture a 1 9 NetServer Software Architecture Figure 1 2 illustrates the software structure of the NetServer Notice that the NetServer s Functional Multiprocessing Kernel FMK executes on each Auspex processor I O Cache Memory e eee eee ee eee eee I I I I A I Protocol Processing 4 FMKCPU J Memory NFS protocol processing Buffer IP routing cae Control puters Flow I I i I a Niy Networks Ethernet FDDI ATM L DMG API HSM File Processing FMK CPU UFS File System Memory UFS meta data Host Processing UNIX CPU Floating Licenses Performance monitoring and tuning Automated backup restore Memory Optional Tape or Optical Disk Library File migration Ancillary DBMS ONC NFS NIS mount lockd authd export aa m SCSI Oia System startup and shutdown Data Flow Sp Control Flow aes I I I t I I mo I I I I I I I I Storage Processing FMK CPU Write
337. ough the NSconfig form includes information related to the NIS master server or a slave server in the NIS environment A separate procedure is required to prepare the server for using the NIS name service Refer to Configuring the Server to Be an NIS Master or Slave Server on page 2 4 for more information 5 After you complete the form execute it by typing Ctrl F NSconfig prints several messages on the screen Follow the instructions printed by NSconfig to complete the configuration procedure The NSconfig procedure is complete Table 2 5 lists thje files modified by NSconfig Table 2 5 Files modified by NSconfig File Description of modification etc contiguration Enables or disables mail messages to Auspex upon system reboot Adds the site name and serial number to the file which are included in the message to Auspex Also adds the server s primary host name interface flags each of which determines whether an interface is enabled or disabled NIS type domain name route daemon flag and console X terminal name etc hosts Sets the IP address of each interface adds the host names etc netmasks Adds the IP subnet masks etc netconf Adds the subnet masks to the ifconfig commands etc ttytab Sets the console terminal type usr share lib zoneinfo lo caltime Links to the local time zone NetServer Configuration Commands a 2 17 SetupTty Your NetServer supports two serial ports referred to by the software
338. ous example ax_perfmon i sysadm perf data Note Although the data was gathered at five minute intervals the screens D display at the default one second intervals thus speeding up data playback 9 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Interactive Performance Monitor Commands After starting the Performance Monitor use the following single character commands to control the display Remember that the commands are case sensitive n AL Displays the ax_perfmon help screen Displays the next data screen For example if the current display is System Summary the next screen is NP Statistics Displays the previous data screen Refreshes the display Quits ax_perfmon Pauses the currently displayed screen useful when replaying saved data Continues displaying updated screens used only after a pause command Does one screen update in pause mode Speeds up the display by reducing the interval between screen updates by 1 10 of one second Slows down the display by increasing the interval between screen updates by 1 10 of one second Speeds up the display by reducing the interval between screen updates by one second Slows down the display by increasing the interval between screen updates by one second On an NP with more than eight net interfaces the NP screen initially displays information for the first group of interfaces This command steps through information for the remaini
339. ows adding physical partitions ad8a and ad9a to the virtual partition vp1 ax_expand vpl ad8a ad9a The ax_expand command automatically edits the etc vpartab file to reflect the changes to vp1 However if you have comment lines in etc fstab that describe the member partitions of the virtual partition being expanded remember to update those comments The ax_expand command generates the following messages The file system being expanded is mounted on tmnt in this example about to grow vpl This may take some time dev rvpl1 683984 sectors in 668 cylinders of 16 tracks 64 sectors Initializing cg 41 Isolating tmnt for superblock adjustments tmnt released The message Initializing cg number indicates the cylinder group being initialized each cylinder group takes about one second to initialize The server prompt appears again when the expansion process completes If you interrupt the ax_expand command you might end up with additional space initialized for the virtual partition but not used The additional space is wasted Expanding a Striped Partition A striped partition cannot be enlarged directly instead it is expanded by replacement After the expansion the file system on the striped partition is moved to a new virtual partition that is bigger than the original one Expanding a striped partition involves two major steps First create a new bigger striped or concatenated partition Second run the ax_expand s command
340. p command with the T option dump OTdsbfu di 141000 11500 126 dev rast4c dev ad0c where d1 is the timestamp 8 2 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table 8 1 Arguments for the dump command Continued Argument Meaning m Name of the partition to be entered in etc dumpdates By default dump writes to etc dumpdates the name of the physical partition being backed up Give a name only if you are dumping a clone partition The name should be that of the source partition in ax_clonefs so future incremental dumps can obtain the appropriate information about the partition in etc dumpaates The following is an example dump Odsbmfu 65000 6700 126 dev rvp10 dev rast4 dev rvp11 In this example the partition name specified with the m option is dev rvp10 It is recorded in etc dumpaates The name of the clone partition dev rvp11 is not recorded v Verifies the contents of the tape tape_device_name The device name of the tape device to be used for the dump This must be specified if the f option is used dev nrastn The no rewind on close option must be used if you copy more than one file dev nrastnlo system onto the tape during the dump procedure dev nrastnc The n in the device name is the slot number in the range of available slots dev nrastnloc Append lo c or loc to the device name to indicate whether compression is used Refer to Table 4 2 and Table 4 3 on page 4 3 for more inform
341. pansion process is completely transparent to users of the file system on vpl The physical partitions used by vp1 before ax_expand now belong to a new virtual partition In the previous example ax_expand chooses the number 3 to name the new virtual partition simply because the number was not used by other virtual partitions Because the space in this virtual partition is free you can use this virtual partition for other purposes For example you can make a new file system on it The following entries are for vp1 vp2 and vp3 after the expansion process dev vp1 mirrored vp2 dev vp2 striped size 128 adl0a adlla adl2a adl3a dev vp3 striped size 128 ad7va ad8a ad9a Note The ax_expand command uses the ax_mconvert command to convert D the striped partition you want to enlarge to a mirrored partition If errors occur during the conversion ax_mconvert displays error messages For more information on ax_mconvert refer to the following section Expanding a One Membered Mirrored Partition Expanding a One Membered Mirrored Partition This section describes how to expand a one membered mirrored partition The procedure for enlarging a one membered mirrored partition depends on the type of its member partition If its member is a concatenated partition you can add one or more physical partitions to the member If the member is a striped partition you can expand it by replacing it with a new member 5 18 a Version 1 10 System Manage
342. param 1 pkgproto 1 pkgrm 1 pkgtrans 1 rm 1 removef 1 rsh 1C s2p 1 stty 1v tar 1 Awk to perl translator Display the architecture of the current host Auspex debugger Allows source level debugging and execution of programs downloaded to Auspex SPARC based processors Tape drive statistics Create tape archives and add or extract files Copy file archives in and out Convert and copy files with various data formats Eject media device from the drive Adds a file to the software installation database Display the processor type of the current host Calculate a message digest checksum for a file Magnetic tape control Run a command immune to hangups and quits Portable archive exchange The practical extraction and report language Installs an optional software package from CD ROM Store answers for an interactive package installation Verifies the installation of an optional software package Provides information on optional software packages Produce an installable package for optional software installation Lists parameters for pkg commands Generates prototype file entries for pkgmk command Removes an installed optional software package Translates an installable software package between formats Remove files Remove files from the software database Remote shell Sed to perl translator Set or alter the options for a terminal Create tape archives and add or extract files A 2 a Version 1 10 System Manager s Guide AU
343. pboot with no options a If you want to get crash dumps from all processors then all the processors must be rebooted Use the lt Break gt key to get to the PROM monitor prompt then give the following command which gives crash dumps and reboots all processors HP gt go 0 3 8 a Version 1 10 System Manager s Guide W AUSPEX Commands for Shutting Down the NetServer This section summarizes the characteristics of the commands and scripts that shut down or reboot a system The Auspex versions of these commands work in the same way as the ones in SunOS unless noted otherwise in this section For detailed information about these commands refer to the man pages a shutdown 8 a halt 8 a reboot 8 a fasthalt 8 a fastboot 8 All of these commands execute sync by default to write out information to the disk before shutting down or rebooting the system shutdown Use the shutdown command to shut down the system if it is currently in multiuser mode The command allows you to specify an amount of time before shutdown occurs and broadcast a message to all system users When executed without the h option shutdown brings the system down to single user mode with the h option the system enters monitor mode after the shutdown is complete The syntax of the shutdown command is as follows usr etc shutdown dfhklnr p processor time warning message The options are as follows d Dump system core f Shut down the s
344. pending on the type of partition you may need to do one or more of the following Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades a Fora mounted physical disk or CD ROM unmount the disk or CD ROM a For striped or concatenated virtual partitions that are not part of a mirrored virtual partition edit etc vpartab to comment out the virtual partition and then run ax_loadvpar so the system reads the revised table see Virtual Partition File on page 5 9 a For a mirrored virtual partition edit etc vpartab to comment out the entry for the member that is on the drive being replaced and redefine the mirror so it has only one member Finally run ax_loadvpar so the system reads the revised table see Virtual Partition File on page 5 9 Removing a Drive a 4 31 AUSPEX W Caution When you hot plug a drive make sure that no active tape drives A exist on the same SP Trying to remove a drive while a tape drive on the same SP is open for read or write access generates an error message To determine if a tape drive is busy use the ax_tapestats command If you try to run ax_tap
345. pendix A Appendix B Appendix C Appendix D Appendix E xiv Example ax_diskconf display n sa duis heise hee 5 24 Example ax_diskconf second table with p option 5 5 25 Recovering From Disk or File System Failures The NSinstall fortis is eia a a dake fii ee eee Ada es 6 31 Write Acceleration Write cache changing states after server power ON 0 00000 7 7 Backing Up the NetServer Backup scheme for 42 drives iac Maite seeone teak Sieh reste eee 8 10 Example of a clone of a mirrored partition 00000 eee ee 8 14 deal tape encodings iiid tetara deeded gos BEE RE ROG ou ae 8 18 1 9 tape encoding Measuring the NetServer s Performance Sample System Summary screen for an NS 7000 200 Series NetServer 9 7 Sample System Summary screen for an NS 7000 600 Series NetServer 9 7 Sample of an NP Statistics screen inh one ke KE Veena W el oe bed gas 9 9 NP Protocol Statistics Screen i405 hits one e Sees eee es ee ones 9 10 Sample FP Statistics screen sxe caesar shoe p ears Aa rrer 9 12 Sample SP Statistics screen sy van gol vise ey ene cot aren aat ee Ae lattes 9 13 Example of a Virtual Partition Statistics screen 0004 9 15 Sample output file created by ax_perfmon with a filter file 9 17 Data Display Wind Ow n 0 hy at tae ee iw eae hel Mah one 9 20 Buttons in the Data Display Window 000 c cece ee eee 9 20 File Load Window s aidia iG 8 haa ieee ioe es
346. pendix B in Sun s System and Network Administration Also you can refer to the Intro 2 man page for an introduction to SunOS system calls and error numbers Messages a E 1 Standard Boot Messages This section provides an example of boot messages that display as part of a normal boot process The boot sequence is annotated with circled numbers that refer to the boot event explanations in Table E 1 The example is for a NetServer with an HP VII NetServers with an HP VIII have a similar boot sequence IDPROM checksum test SPARC Processor 10 0070 Boot PROM Selftest C POST RAM Clear Setup MMU table Copying PROM Data to RAM EPROM Checksum Test 2 Sys Control Register Test SW Interrupt 1 level 1 Test SW Interrupt 2 level 6 Test PC Register Test ECC Valid Access Test Timer 0 Interrupt level 10 Test Timer 1 Interrupt level 14 Test Command Ready Interrupt Test U PTE Reference Bit Test U PTE Modified Bit Test U PTE Valid Bit Test U PTE Access Bit Test U TLB Test U Data Access Trap Test U Probe Test Cache Data RAM Test Cache Tag RAM Test Cache Write Miss Test Cache Write Hit Test NVRAM Battery Test GOOD BATTERY done NVRAM Memory Test saving restoring restoring checksum done NVRAM Test checksum 0x77 TOD Test127 119 95 127 123 127 95 127 POST complete Pass 1 Errors 0 Auspex NetServer SPARC G
347. ple shows the virtual partition vp4 commented out vi etc vpartab Virtual partition 3 Mirrored dev vp3 mirrored vp2 dev vp4 concat ad6e dev vp2 concat ad2e Run ax_loadvpar to install the new partition table Use ax_hot_plug to replace the problem drive with a spare drive Verify that the replacement drive has the correct label If necessary repartition the drive using format Edit etc vpartab to add the member partition from the replacement drive and to comment in any additional references to the added member partition or replacement drive The following example shows the member partition vp4 from the replacement drive added to the mirrored partition vp3 It also shows that the reference to vp4 is no longer commented out Caution To avoid erasing data the virtual partition you are adding must be A listed last This allows the data from the existing member partition vp2 in the example to be copied to the new member partition vp4 vi etc vpartab Virtual partition 3 Mirrored dev vp3 mirrored vp2 vp4 dev vp4 concat ad6e dev vp2 concat ad2e Run ax_loadvpar to install the new partition table After ax_loadvpar runs ax_mrestore automatically runs to update the mirrors Verify the completion of the mirror restoration process with ax_vpstat Recovering from Intermittent Disk Errors Without File System Isolation It is common for a drive to return to normal operation after being temporarily inaccessib
348. ps its system software on CD ROM however NetServer software is installed at the factory and it is usually unnecessary to reinstall the software when you take delivery of the system The CD ROM provides a convenient permanent medium for repairing a damaged root drive For instructions on installing the system software refer to Using NSinstall to Install Software From CD ROM on page 6 29 To repair a damaged root drive first boot the NetServer from the CD ROM in single user mode refer to Booting the Server at the Monitor Prompt on page 3 4 and then refer to Recovering a Damaged Root Disk on page 6 26 Slot 0 is always reserved for the root disk drive The device name for CD ROM drives installed in the NetServer is acdn where n is a number in the range of slot numbers available in your server and is the slot number of the drive The Auspex device driver acd 4 conforms to Sony Philips red book and yellow book specifications The NetServer does not support audio CDs Note If you want to boot the NetServer from the distribution CD by Auspex i convention you must have the CD ROM drive installed in slot 1 The rest of this section discusses procedures for mounting and unmounting file systems and ejecting the CD from the drive Mounting or Unmounting File Systems on CD ROM CDs are read only media and have fixed file systems and partitioning schemes The following describes the CD ROM file system formats supported by th
349. ption is equivalent to the Display Type field in the Histogram Properties window Display the statistics in bars This option is equivalent to the Display Type field in the Histogram Properties window Remove the current histogram from the Data Display Window It is equivalent to the Delete Hist button in the Load Window It does not remove the statistics file from which the histogram is derived Summing Parameters in Histograms Sometimes you might find it useful to sum two or more histograms to get an overall picture of server usage For example you can combine two histograms to display the total number of packets processed by Ethernet interfaces 0 and 1 on a server To combine histograms follow this procedure 1 Load the statistics as you normally would specifying the type of statistics you want to display 2 In the Load Window click on the Sum Parameters box A check mark appears in the box 3 To add another histogram to the one being displayed follow these steps a If the other histogram is from another statistics file select the file in the Load Window b Click on the Parameters button and specify the type of statistics to add to the current one Displaying Performance Data in Histograms a 9 35 The histogram now contains statistics that you loaded in steps 1 and 3 In the Load File window and Data Display Window the histogram title changes according to the summed parameters The following example shows a
350. r s Guide AUSPEX QD W AUSPEX Expanding a One Membered Concatenated Partition If the member of the mirrored partition is a concatenated partition add one or more physical partitions to the member by executing ax_expand For example ax_expand vp2 adl4c Remember to use the name of the mirrored partition vp2 in the above example not the name of its member in the command line Expanding a One Membered Striped Partition If the member of the mirrored partition is a striped partition use ax_expand with the s option to expand it by replacement For example 1 Create a concatenated or striped partition for example vp3 that is larger than the current member of the mirrored partition for example vp2 2 Enter the following command to enlarge the mirrored partition vp2 by replacing its current member with the new member vp3 ax_expand s vp2 vp3 Expanding a Two Membered Mirrored Partition Although the ax_expand command does not support the expansion of two membered mirrored partitions you can expand a file system on a two member mirrored partition with the following procedure 1 Use ax_mdetach to detach one member from the mirrored partition The change is automatically recorded in etc vpartab 2 Use ax_expand to enlarge the mirrored partition see the description provided in Expanding a One Membered Mirrored Partition 3 Use ax_mattach f to add a member to the mirrored partition The f flag c
351. r s hardware manual This section only describes how different models of the 4 mm tape drives define DDS format The NetServer supports two models of 4 mm tape drives WangDAT 2600 and WangDAT 2000 The term low density has different meanings for these models The WangDAT 2000 uses industry standard DDS when switch 2 is set to OFF and when you use rastnlo to name the drive On the WangDAT 2600 you can also set switch 2 to OFF and use rastnlo to name the drive but the drive operates in a low density mode that is different than the standard DDS A tape with data recorded in standard DDS format cannot be read by a WangDAT 2600 drive operating in low density mode Similarly data recorded in low density format on a WangDAT 2600 drive cannot be read by another tape drive that uses standard DDS To determine which model of 4 mm tape drive is installed in a particular slot use one of the following methods 4 Ifa sticker on the drive carrier shows that the drive accepts both 60 m or 90 m tapes the model is WangDAT 2000 a Ifnosticker is attached to your drive carrier remove the drive and check the bar code number on the back of the drive The number under the bar code is 3105 for WangDAT 2000 and 3103 for WangDAT 2600 Tape Drive Capacities Table 8 2 lists the approximate capacity of each type of tape drive The device name indicates the mode of operation For example rastnc indicates that data compression in Tape Drive Capacit
352. r which starts the NIS daemons including ypbind and ypserv Configuring the NetServer to Be an NIS Slave If you use the server as an NIS slave follow these steps 1 For more information about NIS refer to Sun s System and Network Administration manual Edit the etc ethers etc hosts etc bootparams and var yp lt domain gt ypservers files on the NIS master The etc ethers and etc bootparams files on the master must contain the slave server client information The etc hosts file on the master must contain the slave NetServer and slave client host name and IP address The var yp lt domain gt ypservers file which is a list of slave servers in the domain needs the slave NetServer name Verify the domain name bin domainname Enter the ypinit command on the master server as follows usr etc yp ypinit m Answer all the questions displayed by ypinit and add the slave NetServer name to the list of servers For more information on ypinit refer to the man page If you want to make sure you have set the type of NIS service and domain name properly in the NSconfig form on the slave server enter NSconfig to display the form again Verify the settings in the NSconfig form a Slave for Type of NIS service a NIS master s domain for Domain Name On the slave NetServer run the ypinit command usr etc yp ypinit s mastername Reboot the slave NetServer which starts the NIS daemons including ypbind and ypser
353. r Host Processor type the text uses the specific model number or Host Processor type Typographical Conventions In this guide different typefaces indicate different kinds of information The following table explains these typographical conventions Font Meaning Typewriter Indicates a literal screen message Bold In a command line indicates information to be entered exactly as shown In text indicates a command name or device name Italics In a command line indicates a nonliteral item or variable for which you substitute an appropriate value In text indicates a path name or a special term Hexadecimal values in the text are preceded with Ox and leading zeros are not always shown For example the notation 0x68 is used to indicate the hexadecimal address 00000068 xviii a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Special Messages The following special messages are used in this guide N Warning Warnings alert you to the danger of personal injury and call attention to instructions you must follow for your personal safety Caution Cautions call attention to instructions you must follow to prevent A damage to system hardware or software or loss of system data Note Notes call attention to important information you should be aware of as D you follow the procedures described in this guide Recommendation Recommendations call attention to an item or procedure per that is not required but
354. r aliases recognized by sendmail 8 for the local host See aliases 5 etc bootparams Contains new clients this is configured automatically by SetupClient See bootparams 5 etc contiguration Contains the server s primary host name interface flags each of which determines whether a network interface is enabled or disabled NIS type domain name and route daemon flag This is configured automatically by NScontfig etc dumpdates Updates dump records for individual file systems See dump 8 etc ethers Defines Ethernet address to hostname mapping for network hosts which is required for booting diskless SunOS clients This file is automatically updated by SetupClient See ethers 5 etc exports Defines all exported file systems for example root and swap partitions to specific clients File systems for diskless clients are configured automatically by SetupClient and SetupExec Each diskless or dataless client must have an entry in this file See exports 5 etc format dat Holds data for the format 8 program This file defines the known disk types and partition information etc fstab Defines local and network file system mount points See fstab 5 etc gateways Defines IP router information See routed 8C etc group Defines groups recognized by the system See group 5 etc hosts Defines hosts and IP addresses for network devices See hosts 5 New Installations a 2 5 Table 2 1 Key files that contain site
355. r is the HP The system uses fsck to check the unclean file systems when rebooting after a halt command reboot The reboot command runs etc rc shutdown to unmount all mounted file systems except usr and shuts down the running processes and reboots the server to multiuser mode Use reboot only if the server is currently in single user mode If not use shutdown first to warn users of the impending loss of service and then reboot the server from single user mode 3 10 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX At reboot fsck only checks the file systems that are neither clean nor stable A file system is marked clean if it has been unmounted before the reboot it is marked stable if its file system structure was inactive since the last sync before the reboot The following is the reboot command syntax usr etc reboot dnqs p processor boot arguments The options are as follows d Dump system core before booting n Avoid running etc rc shutdown and executing sync rc shutdown is an Auspex script that attempts to unmount all the mounted LFS and UFS file systems except usr and Because this option avoids unmounting the file systems if a file system was active before the reboot it is not marked clean and is checked by fsck at reboot This option is used during emergencies q Reboot the system quickly and ungracefully without first shutting down the running process
356. r kvm by the diskless SunOS clients of the architecture specified by Architecture type Path name of the directory in which to install the architecture independent executables When SetupExec is complete this directory is ready to mount as usr share by the diskless SunOS clients of the architecture specified by Architecture type Type of distribution medium either tape or CD ROM Location of the tape or CD ROM drive Local for a drive on the server or Remvhote for a drive on a remote host Tape or CD ROM device name and slot number for example acd1 for the Auspex CD ROM drive in slot 1 or rast9 for the Auspex tape drive in slot 9 Host name of the machine with the tape or CD ROM drive This field appears only if Drive location is Remote IP address of media host This field appears only if Drive location is Remote Path name of the directory on the host on which the CD ROM drive is mounted This field appears only if distribution medium is CD ROM The default mount point is cdrom If drive location is Local SetupExec automatically executes this mount command mount rt hsfs dev MEDIA_NAME cdrom If drive location is Remote execute mount on the remote host as follows before running SetupExec mount rt hsfs dev sr0 cdrom if the remote host is a non Auspex machine or mount rt hsfs dev acdn cdrom if the remote host is an Auspex server where nis the drive number NetServer Configuration C
357. r the file system check which allows you to release the file system successfully However the file system may be isolated again immediately with the same error In this scenario fix the disk problem before putting the file system online again Restoring a File System that Cannot Be Repaired If a permanent disk error causes a file system to be isolated follow these steps to make the file system available again 1 Enter ax_kill to terminate all the processes sleeping on the isolated file system Note Sometimes ax_kill fails to terminate all processes In this case a message i appears to ask you to kill the processes manually 2 Unexport the file system and unmount the isolated file system 3 Replace the failing disk Refer to Adding or Replacing a Drive on page 4 25 for information on how to replace a drive Run newfs and fsck to create and check a new file system on the disk and then restore the file system from the backup tape 4 Run fsck against the restored file system Mount the file system and re export it to make it available again Examples of Recovery Procedures After File System Isolation This section provides three scenarios in which file systems are isolated for different reasons For each scenario the symptoms are described first followed by the solution The purpose of each of these recovery procedures is to bring the file system back online as fast as possible while avoiding system reboots The media errors
358. ration in detail For Auspex hardware information or SunOS administration information refer to the appropriate manual in the following list a NS 7000 Model 150 250 Series Hardware Manual Auspex Systems Inc a NS 7000 Model 650 Series Hardware Manual Auspex Systems Inc a NS 7000 Model 700 Hardware Manual Auspex Systems Inc a NS 7000 Model 800 Hardware Manual Auspex Systems Inc a Storage Peripherals Manager s Guide Auspex Systems Inc a Version 1 10 Hardware Release Note Auspex Systems Inc a Version 1 10 Software Release Note Auspex Systems Inc a System and Network Administration Sun Microsystems Inc Terminology Several models of Auspex NetServers run NetServer version 1 10 software the 200 series 500 series 600 series 700 series and 800 series The NS 6000 NetServer is also supported but must be upgraded Version 1 10 software does not support a Host Processor IV HP IV or earlier boards a Ethernet Processor EP boards a HP V based FDDI Processor FDDIP boards a Auspex Primary Memory APM A xvii a File Processor FP boards a Storage Processor I SP I Storage Processor II SP II or Storage Processor III SP II boards SP III boards can be upgraded to supported SP III E boards In this guide where information applies to all Auspex network servers regardless of model or Host Processor type the text uses the generic term NetServer Where information applies to a particular model o
359. ready slot d drive d 1205 3 Recovered LUN communication error slot d drive d 1206 3 Recovered drive ID CRC error slot d drive d sector d 1207 3 Recovered drive DATA ECC error slot d drive d sector d Auspex Processor Error Messages a _ E 13 Table E 2 Auspex Processor Error Messages Continued Error Priority code Message text 1208 3 Recovered sector ID not found slot d drive d sector d 1209 3 Recovered drive seek failure slot d drive d 1210 3 Recovered retryable media error slot d drive d sector d 1211 3 Recovered correctable ECC error slot d drive d sector d 1212 3 Recovered drive sync transfer error slot d drive d 1213 3 Recovered drive power up or SCSI reset slot d drive d 1214 3 Recovered drive media corrupted slot d drive d sector d 1215 3 Recovered SCSI controller RAM error slot d drive d 1216 3 Recovered drive data path error slot d drive d 1217 3 Recovered drive power up diagnostic error slot d drive d 1218 3 Recovered drive message reject error slot d drive d 1219 3 Recovered SCSI drive internal error slot d drive d 1220 3 Recovered SCSI interface parity error slot d drive d 1221 3 Recovered SP detected SCSI parity error slot d drive d 1222 3 Recovered illegal SCSI message slot d drive d 1223 3 Recovered SCSI selection error slot d drive d 1224 3 Recovered SCSI disconnection error slot d dr
360. reen is updated at each time interval one second by default it may be difficult to determine for example the peak of CPU utilization for a particular process board 4 ax_perfhist allows you to compare statistics on different processor boards For example you can display two histograms simultaneously for the CPU load one for SPO and one for SP1 You can easily tell which board is busier during a given period of time ax_perfhist can display up to eight histograms at once You can also combine statistics gathered from different processor boards a You can customize the ax_perfhist windows to display the statistics that interest you most and filter out statistics that are irrelevant Also you can display only those statistics that exceed a user defined threshold zoom in on a few samples gathered in a short period of time change the scales used on the horizontal and vertical axes and So on a ax_perfhist displays statistics omitted in the ax_perfmon screens because of screen limitations a You can print statistical reports from ax_perfhist with a mouse click Note Because of changes in hardware components ax_perfhist may not be i able to display some data collected with pre 1 9 software 9 18 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Before Invoking ax_perfhist The ax_perfhist command is useful only after you have used ax_perfmon to save server statistics to a file Use the following co
361. rface 2 11 2 14 Primary Memory 9 47 Print Command output 9 40 Print command 9 38 Print Options Window 9 38 to 9 40 Printer serial 2 19 Printing online documentation xxi Printing histograms 9 38 Processor errors E 8 FMP 1 5 Host 3 3 Network 1 6 9 47 Storage 9 47 utilization 9 2 PROM download utilities PROM monitor booting from 3 4 functions of 3 3 prompt 3 4 10 11 10 12 Prompt PROM monitor 3 4 10 11 10 12 Properties of a histogram 9 25 Q Qty Hists button 9 23 Question mark key 2 10 Quiescent file system 8 4 10 7 R Raw disk device 4 3 rccommand 1 15 Reassigning disk label to another location 6 24 disk sector to a new location 6 25 superblock to another location 6 24 Reboot after a system crash 5 13 server fails to check file systems with a disk error 6 3 reboot command 1 15 3 2 3 8 3 10 Rebuilding the system kernel 10 9 Reconfiguring a virtual partition 5 14 Reconfiguring the kernel after NSinstall 6 34 Recoverable disk errors 6 8 Recovered errors E 8 Recovering from a damaged root disk 6 26 from a drive configuration change 7 11 from disk hardware errors from errors on a mirrored partition 5 13 from root disk failures 6 26 from write cache errors 7 9 Reduction of disk I Os 7 3 Reinstalling the operating system from CD ROM 6 26 Relabeling a disk 6 24 Releasing an isolated file system 6 14 Remapping a bad sector 6 23 Remote device location 2 25 Remote diagnostics 10 12 Remove Hist button 9 22
362. rface or changing an IP address on a NetServer you must restart the daemon in routed See the in routed 8C man page for further information Using Configuration Command Forms When invoked each configuration command displays a form This form contains a set of fields that provide information to the program Many fields contain a default which you can either accept or override by entering other data into the field To accept a field s default value simply leave the field unedited and the default is used when the form is executed Each field defaults either to the last value you used for that field or to the current value in etc configuration if the configuration file has been modified since the last time the configuration form was executed After you fill in all the fields in the form and execute the form the configuration program automatically executes a set of installation and configuration tasks Each form contains three regions see Figure 2 2 on page 2 13 for an example a The form s title date and time are at the top of the screen a The main body of the form in the center of the screen contains two types of data fields Toggle fields are preceded by a greater than symbol gt These fields may contain any value from a predetermined set of values To select a value type the first character of the value or use the space bar to scroll through the available values one by one You may also accept the default by leaving the
363. rite cache hit 2 overrides sect coalesce sec NY bufs 8kK 512B 1 2 33 43 53 6 re 83 10 11 12 13 Scocco cocococcseo wizard Thu Nov 21 14 26 26 1996 Resp time 0 0 Service time 0 0 ms Queue time 0 0 ms NVRAM size MB 14 15 163 Tzs 18 19 20 283 293 303 i 323 333 343 21 223 233 243 253 26 27 353 363 373 383 39 40 413 Scocco coocococceo Scocco eoccoccoccse Figure 9 6 Sample SP Statistics screen For all NetServers except the NS 7000 200 type s for information on subsequent SPs The NS 7000 200 supports only one SP The screen contains the following information A CPU utilization 4 Total response time which is the combined times of the Service time and the Queue time Capturing and Displaying Real Time Performance Data a 9 13 AUSPEX W a Number of operations per second In the bar following the number R represents a read operation and W represents a write operation a Write cache hit percentage is the sum of overrides and coalesced writes as a percentage of the total number of writes involving the write cache It measures how well the write cache is used in avoiding write operations Service time is the time taken in writing to disk Overrides per second indicates how well the write cache eliminates redundant writes of inode and indirect blocks An override is a write that is not performed because a later write to
364. rite the data Purging removes the data permanently from the cache For more information on write cache states refer to Write Cache States and Commands on page 7 6 How the Write Accelerator Improves Throughput a 7 3 How the SP Maintains Drive Information The SP function ensures that the cached data if any is written to disk when the system is booted The SP maintains a table of disk drive serial numbers and other cache control information that is checked at power up to determine if the disk drive configuration has changed If it has not changed the SP uses checksums to verify the integrity of the cache control data and then writes the data to disk When a new disk is added to the NetServer the ax_hot_plug or ax_add_device command automatically updates the SP with the drive serial number to support the write cache function Every disk drive shipped with a NetServer has a unique serial number This information is used by the write cache to verify that information is written to the correct location Before a disk drive is removed from the NetServer the ax_hot_plug or ax_remove_device command instructs the SP to flush the write cache 7 4 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Write Accelerator Restrictions The following restrictions apply to write acceleration a Only writes with 8 KB or less of data are supported by the Write Accelerator Writes with more than 8 KB of data to transfer are trea
365. rive d 1245 3 Recovered DMA VME bus error slot d drive d 1246 3 Recovered DMA VME bus parity error slot d drive d 1247 3 Too many recovered errors replace the disk ASAP slot d drive d 1248 3 Recovered drive servo lost slot d drive d 3001 1 readdirplus can t get inode d 3002 1 readdirplus got unallocated inode d 3003 1 getdirattrs can t get inode d 3004 1 getdirattrs got unallocated inode d 3005 1 isolating filesystem mounted on s 3100 1 bread error in ufs_addblocks Auspex Processor Error Messages a _ E 15 W AUSPEX Index Symbols thosts file 2 22 etc 2 4 etc aliases 2 5 etc bootparams 2 4 2 5 2 28 etc configuration 2 5 2 17 etc dumpdates functions of 2 5 specifying adate 8 2 specifying a partition name 8 3 8 16 writing to using dump 8 2 etc ethers 2 4 2 5 2 28 etc exports 2 5 2 21 2 28 4 30 etc format dat 2 5 4 7 4 9 etc fstab 5 15 home mount point 2 30 adding mountentry 4 29 default 4 19 defining file type for root 4 13 disk partition entries 4 19 file system mount points 2 5 modifying for write caching 7 8 nosuid option 4 20 read only ro option 4 5 type of file systems mounted on the HP 4 18 etc gateways 2 5 etc group 2 5 etc hosts 2 4 2 5 2 17 2 21 2 28 etc hosts equiv 2 6 etc inetd conf 2 6 etc netgroup 2 6 etc networks 2 6 etc passwd 2 6 etc printcap 2 6 etc protocols 2 6 etc rc boot 2 6 2 17 etc re local 2 6 etc re shutdown
366. rmat verification time Time to format and verify format Drive type in minutes 9 0 GB 310 4 0 GB 120 3 0 GB 130 2 0 GB 130 1 76 GB 130 1 35 GB 82 1 0 GB 90 663 0 MB 64 After formatting the disk partition the drive using the partition option as described in the next section Alternatively you can use ax_label to set up partitions and write the label to disk as described in Using ax_label to Partition and Label a Disk on page 4 41 Changing the Partitioning on a Disk The following procedure describes how to partition a disk using the partition option in the Format menu 1 Unmount all file partitions on the disk Make sure no virtual partitions residing on the disk remain mounted If you have not already done so follow the steps outlined in Accessing the Format Menu on page 4 36 Do not use the format command for DriveGuard arrays See the DriveGuard Manager s Guide for using ax_label 8 with DriveGuard arrays Select the partition option on the format menu The partition menu appears with a set of options relating to disk partitioning format gt partition PARTITION MENU a change a partition b change b partition c change c partition d change d partition e change e partition f change f partition g change g partition h change h partition select select a predefined table nam name the current table print display the current table label writ
367. rompt 6 32 a Version 1 10 System Manager s Guide the current root drive Use the default if you install to a slot other than slot 0 Yes when you do normal installations No when you use NSinstall to repair a partition other than root Default is a Yes if you want to initialize the file system in case it is damaged No if you don t want to initialize the file system Yes when you do normal installations No when you use NSinstall to repair a partition other than usr Default is g Yes if you want to initialize the file system in case it is damaged No if you do not want to initialize the file system Yes when you do normal installations No when you use NSinstall to repair a partition other than var Default is f Yes if you want to initialize the file system in case it is damaged No if you do not want to initialize the file system Yes when you do normal installations No when you use NSinstall to repair a partition other than usr openwin Default is d Yes if you want to initialize the file system in case it is damaged No if you do not want to initialize the file system AUSPEX W W AUSPEX Table 6 1 The NSinstall form fields Continued Field Description Possible values Set up export Whether to set up the export file system Yes when you install software for the first time No when you reinstall Remove the in Whether to remove the single in the file
368. rs that Cause File System Isolation 6 10 What Does the FP Do When an Attempted I O Fails 6 12 Repairing an Isolated File System with fsck 00004 6 13 Releasing a File System After Repairing It by fsck 6 14 Restoring a File System that Cannot Be Repaired 6 15 Examples of Recovery Procedures After File System Isolation 6 15 File System Isolation Caused by Media Errors That Cannot Be Fixed 6 15 File System Isolation After Media Errors That Can Be Fixed 6 17 File System Isolation Caused by Corruption 00005 6 18 Recovering from Permanent Disk Errors Without File System Isolation 6 19 Replacing a Disk Containing Unmirrored Partitions 6 19 Replacing a Disk Containing Mirrored Partitions 6 20 Recovering from Intermittent Disk Errors Without File System Isolation 6 21 Replacing a Drive that Is a One Member Mirrored Partition 6 22 Replacing a Drive that Has Multiple Partitions 6 22 Repairing Disk Sectors After an Unrecovered Media Error 6 23 If Sector 0 Disk Label Is Unreadable 0 0 0 e eee eee 6 24 If superblock Is Unreadable iis nstcuee bee age wue yeas ee abe a 6 24 If Other Disk Blocks Are Unreadable 00 000004 6 25 Recovering a Damaged Root Disk sia i hick ia mag wee eae eer es eas 6 26 Booting From a Backup
369. rtition The ax_vpstat display indicates how much of the restore is complete expressed as a percentage The RESTORING condition occurs when the partition is being updated by the ax_mrestore command For more information refer to the man pages for ax_diskconf ax_expand ax_loadvpar ax_mattach ax_mconvert ax_mdetach ax_mrestore ax_vold and ax_vpstat 5 12 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Automatic Error Recovery on Mirrored Partitions After a server crashes and reboots it automatically recovers errors on a mirrored partition to ensure that both members of the partition contain the same data However if disk errors occur on a mirrored partition during normal operations of the server you need to correct the errors manually using the procedures and commands described in Chapter 6 Recovering From Disk or File System Failures Figure 5 7 illustrates the sequence of events and changes in a mirrored partition s status before and after a system crash In this example vp1 and vp2 are the two members of a mirrored partition Status of both vp1 and vp2 is Synced at boot time Status changes to Active after the server is initialized and the file systems are mounted read write System crashes and reboots Each virtual partition remains active During normal shutdown status is set to Synced ax_loadvpar detects the Active status and marks vp2 Dirty ax_loadvpar selects the second
370. s FP statistics Displays CPU utilization for an FP statistics on file system operations per second and statistics on cache utilization Cache utilization helps determine when to add more primary memory SP statistics Displays the CPU utilization for each of the SPs statistics for disk operations per second and statistics for cache utilization Virtual partition Displays the number of I O operations for all the virtual partitions on each SP For statistics virtual partitions composed of multiple members the statistics indicate which member performed the operation To see a graphical representation of the data gathered by ax_perfmon use ax_perfhist which is another Performance Monitor command For more information on displaying statistics in histograms refer to Displaying Performance Data in Histograms on page 9 18 The following provides an outline of the procedure for displaying performance data in histograms 1 Store the ax_perfmon statistics in a file as shown ax_perfmon o outfile 2 Execute the ax_perfhist command which uses the file saved in step 1 as input The following is an example ax_perfhist df outfile The next subsections describe the following a How to start and control the Performance Monitor a Various Performance Monitor screens each section includes a sample screen captured from a NetServer and a description of the screen a How to analyze Performance Monitor data Starting the Performance Mon
371. s The following utility is available 4 ax_load_flash for downloading new Flash PROM firmware to the SP V and NP IV boards as well as to the FDDI and MLT 3 SBus adapters installed on the NP board As a result PROM version changes to these hardware components no longer require returning the board to the factory for a PROM upgrade Instead the upgrade is accomplished entirely through the download utility and firmware code files distributed either on the CD ROM for a major release or as part of a patch release The ax_load_flash download utilitys provides the following features a Verifies whether a specified processor board or SBus adapter s current PROM code matches a specified file containing PROM code 4 Reads a specified processor board or SBus adapter s current PROM code to a designated file for backup purposes 4 Replaces a specified processor board or SBus adapter s PROM code by downloading a specified file and automatically verifies that the PROM code now matches the file Caution The ax_load_flash command erases the original contents of the Flash A PROM Failure or interruption of the download process could leave the network interface or processor board inoperative Downloading incompatible firmware to the interface or board will also leave it inoperative It is highly recommended that only approved firmware revisions be used For FDDI and MLT 3 we also highly recommend that the interface be taken offline and physica
372. s the system boots to the default multiuser mode This option is not required when booting from the CD ROM because the kernel on the CD ROM automatically boots in single user mode c Performs a cold boot If you do not specify c the system performs a WarmStart in which several POST tests are skipped speeding up the time required to boot up to the console prompt This option does not modify WarmStart s default enabling Subsequent restarts skip POST tests unless the c flag is included in the command To invoke the default values for the b command enter the following HP gt b The system boots vmunix from device ad0 in multi user mode The command is equivalent to the following HP gt b ad 0 0 0 vmunix To boot in single user mode use the following command HP gt b s Table 3 1 shows some other examples Table 3 1 Examples of the b command Command Explanation b ad 0 2 0 a Boot vmunix from the backup root disk in slot 2 in multiuser mode with the ask me option which allows you to specify the root and swap partitions b ad 0 0 0 c Boot vmunix from the root disk in slot 0 in multiuser mode with the cold boot option which specifies that all POST tests be performed upon booting b vmunix backup Boot from an alternate kernel file on ad0 in multiuser mode Exiting Single User Mode to Multiuser Mode After you boot the server in single user mode the system provides you with a single user shell After you complet
373. s displayed by ax_perfhist 9 40 FP 9 43 HP 9 41 NP 9 41 SP 9 45 Statistics parameters in ax_perfhist 9 40 to 9 45 Statistics screens displayed by ax_perfmon 9 6 to 9 15 Status virtual partition 5 11 stopnfsd daemon 1 12 Storage Processor error messages features of 1 6 how it maintains drive information 7 4 number needed 9 47 number of virtual partitions 5 7 reading from a mirrored partition 5 5 support for virtual partition clones 8 12 warning message 6 10 Write Accelerator option 1 6 writing to a mirrored partition 5 5 Storage subsystems 1 4 Stripe size of a striped partition 5 8 5 10 Striped partition 5 4 5 9 Striped virtual partition moving 5 21 Subnet mask 2 15 Summing parameters in histograms Sun system architecture 2 22 SUNBIN CD or tape 2 22 SunOS enhancements 1 12 error numbers E 1 pointers to documentation 1 16 system calls E 1 version 2 30 SunOS kernel reconfiguring 2 5 Superblock 6 24 Super user privileges 3 2 Surface analysis 4 38 Swap file on client machine 2 30 Swap space on client machine 2 28 6 3 6 8 6 23 6 24 9 35 on root disk 3 7 sync after reboot 3 2 sync command 3 8 3 9 10 10 Synced status virtual partition 5 12 syslog conf 10 11 syslogd 10 11 System crashes 5 13 10 10 panics 10 10 E 8 System cache I O memory 3 11 System configuration data displaying 10 5 System kernel rebuilding 10 9 System Summary screen 9 6 9 7 System Summary statistics 9 3 T Tab key 2
374. s value causes a larger amount of space between the frame and the left edge of the window Y Pos The spacing between the frame containing the histogram and the upper edge of the window A greater Y Pos value causes a larger amount of space between the frame and the upper edge of the window Width The width of the box containing the histogram measured in pixels Height The height of the box containing the histogram measured in pixels X Title The title below the X axis By default it is the parameter name Y Title The title at the top of the histogram By default it is the date and the server name For example in Figure 9 15 the Y Title is Fri Feb 3 14 48 52 1995 Server acdelco Plot Title The string following Y Title By default it is the filename of the input file For Scaling Factor Threshold1 and Threshold2 Display Type X Axis Display example in Figure 9 15 the Plot Title is acdelco dat A multiplier for the labels on the Y axis For example if Scaling Factor is 2 all the values on the Y axis are multiplied by 2 The shape of the histogram remains unchanged By default Scaling Factor is 1 The value on the Y axis at which ax_perfhist draws a dotted line across the histogram The form of the histogram The default is Discrete which means that statistics are presented in bars if Display Type is set to Continuous a graph is plotted instead as shown in Figure 9 16 The units for the values on the X
375. sables the write acceleration function when the cache state changes from UNINITIALIZED or ON to BADCHECKSUM or DIRTY Refer to Managing the Write Accelerator on page 7 9 for error recovery procedures Managing the Write Accelerator The ax_write_cache command is used to manage the write cache in a number of ways This section describes the following procedures a Recovering from a write cache BADCHECKSUM or DIRTY error a Recovering from a disk media error a Recovering from a disk hardware error a Recovering from a system configuration change Each of these procedures requires that you log in as root Recovering From Write Cache BADCHECKSUM or DIRTY Errors If the SP indicates that the write cache is ina DIRTY or BADCHECKSUM state after initialization follow this procedure 1 Enter the ax_write_cache command to verify the cache state For example if the write cache is on SPO enter ax_write_cache s 0 state 2 Follow either of these steps depending on the write cache state Write Cache States and Commands a 7 9 a If the state is BADCHECKSUM enter the following command and then go to step 3 ax_write_cache s 0 purge b Ifthe state is DIRTY flush the unwritten data to disk ax_write_cache s 0 flush This command tries to write all data currently in the write cache to the file systems that use write caching If all data is successfully written to disk the write cache switches to OFF and you can
376. sh to disk operations NFS read link operations NFS null operations NFS get attributes operations Displaying Performance Data in Histograms a 9 43 FP lfs setattr NFS set attributes operations FP 1fs lookup NFS lookup operations FP lfs create NFS create operations FP lfs remove NFS remove operations FP lfs rename NFS rename operations FP lfs link NFS link operations FP lfs symlink NFS symlink operations FP lfs rmdir NFS remove directory operations FP lfs statfs NFS statfs operations FP lfs fsync NFS fsync operations FP lfs access NFS access operations FP lfs syncfs NFS sync file system operations FP lfs quota NFS quota operations FP lfs total Total number of NFS operations FP lfs max Maximum number of NFS operations FP fp data cache LRU Eject Age Average age BQ_LRU data bufs dropped from cache FP fp data cache LRU Eject Count s Count of BQ_LRU data bufs dropped from cache FP fp data cache AGE Eject Age Average age BQ_AGE data bufs dropped from cache FP fp data cache AGE Eject Count s Count of BQ_AGE data bufs dropped from cache FP fp meta cache LRU Eject Age FP fp meta cache LRU Eject Count s Count of BO_LRU ctl bufs dropped from cache FP fp meta cache AGE Eject Age FP fp meta cache AGE Eject Count s Count of BO_AGE ctl bufs dropped from cache FP fp inode cache Eject Age FP fp inode cache Eject Count s Average age of inodes dropped from cache Count of inodes dropped from cache FP fp ino
377. sk space You can pool more chunks of small disk space that are otherwise unusable into one large virtual partition You do not need to worry about exceeding the file system size limit while adding small physical partitions to a large virtual partition Note If you enter the df command on a client to display the amount of free i disk space on the server you might see negative numbers in the df output for virtual partitions that are larger than 2 GB The following example shows the df output containing negative numbers proto disks proto 1046719 1950330 169422 97 disks proto sp2vp4 However if you enter the df command at the server the number of kilobytes are displayed correctly For example dev vp150 7341889 6437142 170558 97 disks proto sp2vp4 Maximum Size of a Virtual Partition a 5 3 Different Types of Virtual Partitions Auspex supports three types of virtual partitions concatenated striped and mirrored Several commands are available for you to define virtual partitions and recover from disk or media failures that affect virtual partitions These commands are described in Virtual Partition Commands on page 5 10 You can set up virtual partitions on disks only you cannot use virtual partitions on CD ROM or tape drives Concatenated Partition A concatenated partition is a concatenation of 1 to 16 physical disk partitions from one or more disk drives allowing you to accommodate a file or file system l
378. spex on reboot gt NO Customer Site yourcompany Soe ee eo See eS Sa nage ane oe BOTTOM Use TAB CR or arrow keys to move among fields when done hit CTRL F to execute HELP F EXECUTE C ABORT L REDRAW SCRN R INITIAL VAL lt sp gt SELECT VAL a Figure 2 2 Sample NSconfig form NetServer Configuration Commands a 2 13 Table 2 3 The NSconfig form fields Field Description Possible values Server Hostname Ethernet Intf Fast Ethernet Intf FDDI Int ATM Intf Primary hostname Internet Address REFLECT Mode Host name of this server The hostname command uses this name to set the host name for this server Normally the name should be the primary host name of one of the interfaces below Host name is set during the execution of NSconfig and updated in the etc configuration file and the X console configuration file Numbered interfaces Names and states of the interfaces reflect the current configuration of the system For each interface use the space bar to toggle between UP and Dwn See the NSconfig 8 man page for ATM specific information NOTE ATM interfaces are available as optional products Refer to the appropriate documentation on the Auspex Premier Software Series CD ROM for more information Host name of the interface Only one interface needs to have a name that is the same as the server Other interfaces can have names that are
379. splay Window using the configuration information in 6histogram cfg The histograms display statistics loaded from host1 dat ax_perfhist cf 6histogram cfg df host1 dat pf hostlhistogram qt This command prints the histograms to the file named host1histogram ax_perfhist quits after it saves the screen image to hostIhistogram ax_perfhist cf 6histogram cfg df host1 dat lp qt This command prints out the histograms and quits when it is finished printing 9 46 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Analyzing Performance Monitor Data By observing the performance data over an extended period of time you can determine when to distribute file systems over multiple drives whether to change the distribution of client workstations on the attached networks and whether additional memory or processor boards in the NetServer will improve system performance a DolIneed more I O cache memory Observe the data cache age bar on the System Summary screen If the bar shows all A s or a s then all buffers in the cache were accessed within the last minute This is a strong indicator that your NP needs more memory All the I O cache memory is on NP boards A NetServer with enough memory has a few A s at the left representing the most active buffers but has higher letters towards the right As long as a few buffers that are 5 to 10 minutes old J or greater then the system does not need more memor
380. sr local bin is included in the PATH of login you cannot log in as root once usr local is isolated To ensure that you can log in to a server as root after a file system is isolated do not include commands that try to access an LFS file system in the root s Jogin file or any shell startup file such as cshrc and profile For more information on file system isolation refer to File System Isolation on page 6 10 For more information on LFS file systems refer to Mounting File Systems on the FP and HP on page 4 18 Root Login and File System Isolation a 2 31 100Base T Ethernet Full Duplex Auspex 100Base T Ethernet full duplex supports both full duplex and half duplex modes with the HME SBus adapter card The older BMAC SBus adapter card supports only half duplex mode 100Base T Ethernet full duplex is bundled into the Auspex base system software and requires no separate installation Configuring 100Base T Ethernet connections is the same as configuring 10Base T Ethernet connections Use NSconfig 8 to assign each connection an IP address subnet mask and so on HME interface performance may be hindered when the interface is set up with auto negotiation mode Use the NSconfig 8 or ifconfig 8 command to manually set the HME parameters of duplex mode and speed A 100Base T Ethernet interface connects to either a full duplex interface a half duplex interface or an Ethernet switch port Be sure that the Ethernet switch port
381. st7 DLT4000 format without compression rast7c DLT4000 format with compression rast7lo DLT4000 format without compression rast7loc DLT4000 format with compression For more information on how data compression affects tape capacity refer to Chapter 8 Raw Disk Partition Where a command refers to a raw disk partition which gives an application direct access to the partition the device name is preceded by an r as in rad7c Device Names a 4 3 AUSPEX D Device Special Files The Auspex operating system software is shipped with all necessary device special files If you need to create more Auspex device special files use the MAKEDEV auspex command Refer to the MAKEDEV auspex 8 man page for details If you want to create device special files for the SCSI port located on the HP refer to Attaching Drives to the Host Processor SCSI Port on page 4 35 To ensure that devices exist boot the system and watch for the kernel to recognize each device immediately after loading vmunix The kernel displays a single line for each device which contains the device name and hardware information For disk devices this information is displayed if the disk drive is formatted and labeled an unlabeled drive is listed as such This information is also saved to the file var adm messages An example of the boot sequence is provided in Appendix E 4 4 a Version 1 10 System Manager s Guide W AUSPEX Using CD ROM Drives Auspex shi
382. stration and enhanced security features Clients using Windows NT Server Windows NT Workstation and Windows 95 get e the ability to access UNIX based file systems on Auspex NetServers An Auspex NetServer using NeTservices can act as a Primary Domain Controller PDC or Backup Domain Controller BDC in a Windows NT domain Optional Products a 1 7 EtherChannel EtherChannel combines multiple 100Base T Ethernet interfaces into a single channel called a fat pipe All the interfaces in the EtherChannel load balance the traffic among themselves giving the concept of one fat pipe The Auspex EtherChannel implementation takes care of traffic directed to a router through the same port since EtherChannel does the load balancing based on IP addresses MAC addresses or with round robin algorithm The Auspex EtherChannel is a software only product When the packet is ready to go out on the net EtherChannel makes a decision at the driver level about which port to use ATM NP ATM Release 2 provides asynchronous communication from the NP board with data transfer rates up to 155 MB per second NP ATM Release 2 includes LAN Emulation LANE for Ethernet clients and allows you to configure three types of network interfaces FORE IP Classical IP and LEC FTP on the Network Processor FTP on the Network Processor improves File Transfer Protocol FIP transfer rates between NetServer disk storage and a network Data can flow directly from the
383. sts of one or two members Each member must be of the same size as the mirrored partition Each member must be a virtual partition and if there are two members they should be located on different physical drives to achieve greater reliability The drives however must be supported by the same SP The data written to the mirrored partition is queued and written to both members as soon as possible As a result using a mirrored partition ensures that critical file systems remain available even if a member of the mirror fails Note Using mirrored partitions is not a valid substitute for performing i backups It is still important to back up all data to tape When data is read from the mirrored partition the SP chooses the member from which the read can be more quickly performed using a short queue short seek algorithm Thus although a mirrored write takes slightly longer than an unmirrored write a mirrored read is faster than an unmirrored read Caution Normally if one storage component such as a member disk of a A mirrored partition fails during a write operation the data can still be written to the partition and no data is lost However if write caching is used for the file system on a mirrored partition a failing Write Accelerator might lead to a loss of data because data in the Write Accelerator s NVRAM is not mirrored For example after data has been successfully written to the write cache a write completion is returned to t
384. t Auspex makes to SunOS These changes are described in SunOS Enhancements on page 1 12 Booting Up and Shutting Down the NetServer a 3 1 Booting the NetServer Auspex boot shutdown and halt procedures are similar to SunOS procedures with these differences a The permissible boot devices are limited to ad for Auspex disk and CD ROM drives The NetServer does not support booting from tape drives Note The device type ad applies to both disk and CD ROM drives only when D booting the NetServer At all other times the CD ROM device type is acd a The Auspex boot procedure automatically implements the WarmStart feature in which boot or reboot time is improved by bypassing selected Power On Self Testing POST diagnostics If you wish to perform a cold boot which does perform all diagnostics the boot reboot and fastboot commands include a c flag which causes the reboot to perform the standard POST tests a You cannot boot the NetServer remotely from another device on the network a The commands fasthalt and fastboot have the same effect as halt and reboot on file system checking upon reboot That is all these commands cause the server not to check the file systems that unmounted cleanly before the operating system was halted The fasthalt and fastboot commands however also skip system cache I O testing which is performed when the system reboots after a halt or reboot command a The reboot and halt commands include
385. t X Label 0 00 Last X Label 12 00 X inec 0 00 Lower Y Label 0 00 Upper Y Label 18 64 Y ine 0 00 X Pos 9 Y Pos 9 Width 639 Height 182 X Title FPO cpu busy Title Fri Feb 3 14 48 52 1995 Serveriacdelco Plot Title acdelco dat Scaling Factor 1 0000000 Threshold 0 00 Threshold2 0 00 Display Type X axis Display Figure 9 14 Histogram Properties Window Table 9 3 Fields in the Histogram Properties Window Field Function Hist Input Filename Color Index Total Samples First X Label Last X Label X inc Lower Y Label Upper Y Label The number of the histogram you want to modify For example if you select 2 the fields in the current Histogram Properties Window affect only the appearance of the second histogram The name of the statistics file that contains the statistics displayed in the histogram This field is empty if you have not loaded the histogram The color used for the selected histogram The available colors are light gray red green blue yellow dark gray white and black The default color is black The total number of samples gathered and displayed in the histogram The first sample displayed in the histogram By default the histogram starts with the first sample and displays the time corresponding to the first sample as the first value on the X axis However you can for example specify that the histogram starts with the tenth sample by setting First X Label to
386. t command Once halted the system automatically transfers control to the PROM monitor program as indicated by the HP gt prompt The following example shows a typical halt command and the system s response halt Syncing disks done HP gt a In multiuser mode Halt the operating system using the etc shutdown command with the lh options This command brings down the system gracefully issuing warning messages to users before transferring control to the PROM monitor program For example the following command halts the system five minutes after the command is executed etc shutdown lh 5 system going down in 5 minutes System shutting down at 14 58 PST The system shutdown message is repeated every 60 seconds HP gt The l option sends the warning message only to users who are logged in shortening the time required for the shutdown if there are PC or Macintosh users on the network who mount file systems from the server The monitor prompt indicates that the system is ready to accept PROM monitor commands For more information about commands related to shutting down the system refer to Commands for Shutting Down the NetServer on page 3 9 Booting the Server at the Monitor Prompt This section describes how to boot the server after you halt the operating system The procedure in this section boots the server from the default boot device or the device you specify This section also explains the various
387. t half the capacity of a 112 meter tape in the same drive Table 8 4 lists dump and restore command examples for some tape drives These examples assume that you install a 4 mm tape drive as rast1 an 8 mm tape drive as rast2 a DLT4000 tape drive as rast7 a 1 4 inch tape drive as rast3 and a 1 2 inch tape drive as rast19 Table 8 4 Dump and restore command examples Drive type Tape size Command dump 4mm 60 meter dump Ofusdb dev rast1 6300 110400 126 dev rad0a 8mm 60 meter Exabyte 8200 dump Ofusdb dev rast2 6700 65000 126 dev rad0a DLT4000 1800 feet dump Ofusdb dev rast7 115000 81633 126 dev rad0a 1 4 inch 600 feet dump Ofusdb dev rast3 600 1000 126 dev rad0g 1 2 inch 2400 feet dump Ofudb dev rast19 6250 126 dev radOh restore 4mm 60 meter restore dev rast1 8mm 90 meter restore dev rast2 DLT4000 1800 feet restore dev rast7 1 4 inch 600 feet restore dev rast3 1 2 inch 2400 feet restore dev rast19 If you omit the b option in the restore command restore attempts to figure out the block size for the tape Note When using dump to back up a partition larger than 2 GB remember to D specify the raw partition name not the special block device name This applies to both physical and virtual partitions For example use dev rad2c and dev rvp2 to dump ad2c and vp2 respectively If you specify dev ad2c or dev vp2 in the dump command this error message appears bread lseek fails Tape Drive Capacity a 8 9 Back
388. ta in Histograms A 9 33 Auspex Systems Data Display Tool File 7 Properties 7 Print 7 Qty Hists 1 Hist Fri Feb 3 15325357 1995 Servertacdelco File acdelco dat 6 58 Fri 15 27 06 Fri 15327314 Fri 15327325 Fri 15327332 NPO cpu busy Ave 9 9 24 34 23 54 22 84 22 14 21 34 20 64 19 94 19 14 al gi er nN Fri 153 Figure 9 21 Histogram showing samples after zooming Menu for Zooming In and Zooming Out When you point anywhere within a histogram and hold down the right mouse button a menu consisting of 10 options appears The functions of these options can be summarized as follows Zooming in or zooming out samples you selected Resetting markers in a histogram Changing the scale on the Y axis Toggling between a continuous plot and a discrete plot gt gt gt gt p gt Removing a selected histogram Table 9 5 describes the function of each option in the pull down menu Some of the options affect only one histogram while others affect all histograms currently in display In the table the histogram in which you click to invoke the pull down menu is referred to as the current histogram 9 34 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table 9 5 Options in pull down menu in Data Display Window Option Function Zoom Zoom in on the samples delimited by the markers in the current histogram See Examining Samples in Histograms on page
389. tall var Partition newfs first Install usr openwin Partition newfs first Whether to install the root file system The physical partition on which to build the root This field appears only if you respond Yes to the Install root prompt Whether to initialize the root file system before installing the software This field appears only if you respond Yes to the Install root prompt Whether to install the usr files The physical partition on which to build usr usually the g partition This field appears only if you respond Yes to the Install usr prompt Whether to initialize the usr file system before installing the software This field appears only if you respond Yes to the Install usr prompt Whether to install the var file system The physical partition on which to build var This field appears only if you respond Yes to the Install var prompt Specifies whether to initialize the var file system before installing the software This field appears only if you respond Yes to the Install var prompt Whether to install the usr openwin files The physical partition on which to build usr openwin This field appears only if you respond Yes to the Install usr openwin prompt Whether to initialize the usr openwin file system before installing the software This field appears only if you respond Yes to the Install usr openwin p
390. tatd 8 ax_statd2 8 ax_keyenvoyd ax_keyenvoyd 8 ax_lfsd 8 ax_nfsd 8 and ax_lfsd ax_vold 8 ax_nfsd ax_vold Networking The nfsstat 8C command is modified to obtain Man pages for commands and statistics for the NP and UNIX Also an added option nfsstat daemons I interface displays NFS and RPC statistics about ax_netstat a network interface stopnfsd aXx_arp A new command ax_netstat 8C displays NP ax_fddistat network statistics A new command stopnfsd 8 shuts down the NFS daemons on the NPs A new command ax_arp 8C displays or flushes the Address Resolution Protocol ARP table on an NP A new command ax_fddistat 8C displays status of FDDI and MLT 3 network interfaces Performance The Auspex Performance Monitor command Chapter 9 monitor commands ax_perfmon 8 gives the system manager the ability to gather and display NetServer performance data The ax_perfhist 8 command displays server performance data in histograms 1 12 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Table 1 2 Changes to SunOS Continued Types of changes Description For further information see Commands for virtual partitions A disk management tool ax_diskconf 8 displays NetServer disk configuration data including information about virtual partitions vp 4 is the virtual partition device driver vpartab 5 is the virtual partition table ax_loadvpar 8 loads the partition table ax_vold 8
391. tdown 3 8 3 9 10 12 software configuration 2 9 syne 3 8 3 9 10 10 tunefs 4 22 umount 4 22 xwd 9 38 9 39 ypinit 2 4 Concatenated virtual partition changing to a mirrored partition 5 19 definition 5 4 moving 5 21 Configuration client specific 2 1 commands 2 9 displaying data about 10 5 form 2 9 HELP function 2 10 management 10 2 server specific 2 1 software 2 1 worksheets D 1 Configuration Window 9 30 Console port 2 18 AUSPEX QD W AUSPEX turning on and off 10 12 Conventions naming disks 4 10 naming partition tables 4 10 representation of variables E 8 Converting a striped or concatenated partition to a mirrored partition 5 11 5 19 Copying ax_config information to a file 10 6 backup root disk 6 27 data from one disk to another 6 22 6 23 the root partition using dd 6 26 Core dumps from an NP 10 10 CPU utilization monitoring 9 2 crash command 10 10 Crashes system 5 13 10 10 Creating a virtual partition 5 14 cron command 8 10 Ctrl C 2 10 Ctrl D 3 6 6 4 6 6 6 7 Ctrl F 2 10 Ctrl H 2 10 Ctrl L 2 10 Ctrl R 2 10 Ctrl U 2 10 Ctrl W 2 10 Customer support International customers xx North America xx D Daemons ax_errd 1 12 ax_isolated 6 12 ax_keyenvoyd 1 12 ax_lfsd 1 12 ax_nfsd 1 12 ax_startup 1 12 ax_statd 1 12 ax_statd2 1 12 ax_timed 1 12 ax_vold 1 13 5 11 nfsd 1 12 routed 2 11 2 16 stopnfsd 1 12 syslogd 10 11 Damaged root disk 6 26 Damaged root partition 6 26 Dam
392. te EEE 9 21 Displaying the Data Display and File Load windows simultaneously 9 22 Data Display Window showing two histograms simultaneously 9 23 Histogram Properties Window sre ties an ek occas ick sses AeA ee ARE 9 25 Effects of Histogram Properties parameters on a histogram 9 27 Server statistics plotted on a continuous graph 5 0055 9 28 Overall Properties Window lt 5 ie ao i RGA oe ow bem dinar aes 9 28 Configuration WIG Wakes os i cost aut fe Sos a cit te Cee ele Be a 9 30 Displaying six histograms as specified in a configuration file 9 31 Pull down menu for zooming samples between the markers 9 33 Histogram showing samples after zooming 000 0000 ee eee 9 34 Summing two histograms gi coe data de dade det aa MoO Wad had Ras 9 36 Print Options Window icspaniad s Jaton seagate MAG e a eae each 9 38 Special Maintenance How ax_config obtains configuration information 10 3 UNIX Manual Pages Online Documentation Time Zones Configuration Worksheets Messages A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Tables Chapter 1 Table 1 1 Table 1 2 Table 1 3 Chapter 2 Table 2 1 Table 2 2 Table 2 3 Table 2 4 Table 2 5 Table 2 6 Table 2 7 Table 2 8 Table 2 9 Chapier 3 Table 3 1 Chapter 4 Table 4 1 Table 4 2 Table 4 3 Table 4 4 Table 4 5 Table 4 6 Table 4 7 Table 4 8 Table 4 9 Table 4 10 Chap
393. te ete eee eee ees 10 9 System Panics and Crashes 0 ccc cee eee eens 10 10 Message Lopen oS Tur rie aa aA e Teg Sa dence te eae eats E Nene 10 11 Using the Continue Command 40 2 gays Sepak eo oer yew 10 12 REM OtEIIASTIOSULCS rs factor ee aah oie acs tea a OR aE edn soe tae BQ a aA 10 12 Managing Network Interfaces see kas eae a Lae ea ORE SS 10 14 Displaying and Flushing ARP Tables Used by an NP Board 10 14 Displaying Network Statistics for a Network Processor 10 15 Displaying NFS Statistics ds hia eye eee ie eae es 10 19 Network Changes i nia satan nie eb eee ined Beas 10 19 x 4 Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Appendix A UNIX Manual Pages About This Appendix ai sR a RPE LRN SING GRO Sonat A 1 User Commands sya Neh tani Bae ote ta lee SS iets A 2 System Calls omana ea te gaa yoy ooh A eae eal ean eae a A 3 Devices and Networking jst Ae ite tate ties Seni ge hk Rta A 4 File Formats meenen at am aaie ght Rianne E tana hemes gS A 6 Maintenance Commands cr isia wag eer ie aad SAS eet wae CEA ah A 7 Appendix B Online Documentation About This Appendix Caeenien ety tar thee Seas Orie rare ar rannin renr rreren ERE etree are ean B 1 Online Documentation s wd Soe ea uaAhin Ge aR Ge nak ee a Ble amare B 2 Installing Online Documentation ts gaecagraa da ini as Bad BAe abe B 2 Starting the Online Documentation 0 cece cee eee B 5 OV
394. ted as conventional writes a The backup battery has a minimum shelf life of two years at 70 F When the battery power is low the NetServer displays an error message to the console and attempts to flush data to disk before the write cache becomes unusable a The Write Accelerator III provides approximately 3 months of actual backup a Inthe NetServer each SP that uses write caching must have a Write Accelerator board installed on it The Write Accelerator cache memory is not shared among SP a Physical partitions must start on an 8 K boundary a AnSP V must have the latest PROM Flash code to access the full 8 MB Write Accelerator Look at the Write Accelerator notes in your system s hardware manual or check with your Auspex field representative to determine the PROM Flash code level of your SP V Write Accelerator Restrictions a 7 5 Write Cache States and Commands Understanding the state of the write cache is important for managing it The SP keeps track of seven cache states UNINITIALIZI OFF BADCHECKSUM DIRTY ON BATTERY_LOW NONE py ED If the write cache is present on the SP it powers up in the UNINITIALIZED state The write cache is initialized automatically at boot time from an entry in etc rc boot During initialization the write cache changes states several times as described in the following paragraphs and Figure 7 1 The state is set to OFF in either of these situations If
395. ted by an SP NP Statistics Screen The NP Statistics screen displays network statistics based on data collected from an NP Figure 9 3 shows a sample NP Statistics screen 9 8 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX wizard Total Bytes sec 4 ri 1 0 NP Stats for NPO CPU OZI MBUF stats max cur err Or Of Of 0 1 Of Of 0 3 0 fr B Of 0 E Of Of 90 Of Tg 0 at 64 32 at 1287 44 0 of Ht 0 Thu Nov 21 13352304 1996 Nfs work que busy 0 ITEMS a FO H Frames 0 Bytes 41 IBytes 22 OBytes 19 InErr 0 OutErr 0 Collis OutDisd 0 InBig InUndrsz InInval InAbort OFF MTU Figure 9 3 Sample of an NP Statistics screen The upper portion of the screen contains statistics about the NP and buffering information for packets going in and out of network interfaces See the MBUF stats row in Table 9 1 for an explanation of the MBUF stats interface definitions The NP processes frames from the networks The lower portion of the screen is a table listing statistics for each network interface Some statistics apply to more than one interface type and some apply to only one interface type Interface definitions for the ITEMS row are the same as for the MBUF stats For information on the next group of SBus interfaces type c For information on subsequent NPs type N Table 9 1 explains the network statistics Table 9 1 NP statistics Statistics Explanation CPU Util Util
396. tems Mounted on the HP You can mount file systems on the HP if any of the following conditions apply a The file system being mounted is not supported by the FP For example if the file system is the High Sierra File System or a third party file system mount it on the HP a If the file system is primarily used by the HP and rarely used by clients of the NetServer mount the file system on the HP This enables the applications running on the HP to access the file system faster than they would if they had to access the file system through the FP a If you want the HP to be able to use the file system even after the FP has failed mount the file system on the HP For example in order for the HP to log an FP failure to var adm messages the var file system must be mounted on the HP instead of the FP Note Operations on file systems mounted on the HP require HP CPU cycles i and the entire SunOS file system and virtual memory system As a result NFS access to these file systems is slower than access to LFS mounted file systems Mounting several file systems on the HP runs a greater risk of exhausting HP resources such as CPU cycles and virtual or physical memory Use the criteria described previously to determine whether it is necessary for the HP to manage the file systems In the etc fstab file the file type for file systems mounted on the HP is 4 2 4 18 a Version 1 10 System Manager s Guide W AUSPEX Auspex s Modifications
397. tems Mounted on the HP cis cc0i aad eikie oth atat ute oe ee ees 4 18 vi a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Auspex s Modifications to etc fstab ye Sonia evita eal ates 4 19 Specifying the File System Type soi 3 05055 ear eax oe gS 4 19 Mapping LFS File Systems to File Processors 00000 4 20 Enabling Write Acceleration in etc fstab n n nananana nananana 4 22 Other Commands Affected by LFS ii iis ce der even eek eens v8 4 4 22 File Processor Constraints 2 0 0 cece cece eens 4 22 File System Formats supported by UP Sissies 4a Sa 4 22 NFS Version 3 and UFS Limitations as ite hawks oe Maa aad 4 23 Specifying a Host Name in etc fstab 0 0 nunnana nnan 4 24 Adding or Replacing a Drive s o 02 igadad noth et bane Rais aways 4 25 Installing or Replacing a Drive After Booting 00000 00 4 26 Before You Start the Drive Installation Procedure 4 26 Using ax_hot_plug to Add a Drive to the Server 4 27 Installing or Replacing a Drive When the Operating System is Shut Down4 28 Configuring a New Disk DiivGss ive Sos cis Pet ne ed BGS 4 28 RemOviniga Drive vyse paeen Neigh aed Stina Pea dh Aig RG elle ihe SAPUE O REA 4 31 Removing a Drive When the Operating System is Booted 4 31 Before You Remove a Drive lt c 44 chek Mad eri OM A a Oe 4 31 Using ax_hot_plug to Remove a Drive 0 0 0 00 eee 4 32 R
398. ter 5 Table 5 1 Chapter 6 Table 6 1 Chapter 7 Table 7 1 Chapter 8 Table 8 1 Overview of NetServer Features NetServer features mirnesa Ricca n Marlyn ees Mie a ind eh BE ae adel 1 3 Changes to SunOS saavawiahoaaiees er eeeeev eee heat pe ESENTE OE 1 12 Pointers to Sun documentation 00 0000 ccc cee eee ees 1 16 Software Configuration Key files that contain site specific information 004 2 5 Keys for editing configuration forms 6 0000 c cece eee eee 2 10 The NSconfig form fields 2c a gaseit aya ach MA tele 2 14 Date and time formats yikes oe ets halve ence areal tee hens ee ea 2 17 Files modified by NSconfig goivretnete sah Cee nee A 2 17 The Setup ity s0rmnelossint cod sate ananunua nunnan rererere 2 19 Sun system architectures Gai au aaah atic ee a se Oa Ca 2 22 The Set pExec fori fields rercrsrnikhentecyiiiie ta sibni piini etay 2 25 The SetupClient form fields lt lt 246 suikordl piste Wie miweek ee 2 30 Booting Up and Shutting Down the NetServer Examples of the b command sj 0 n enw er we ea hr aaa eS 3 6 NetServer Storage Devices and File Systems CD ROM naming conventions 3 0 i iuti ewes SI ee en asta 4 2 Naming a 4 mm tape drive in slot 5 sc c005 oes gd cave oan tae eae 4 3 Naming an 8 mm tape drive inslot5 annaas nuanse 4 3 Naming a DLT4000 tape drive in slot7 1 6 cece eee 4 3 Default disk drive partition tables 0 000 cece ee 4 8 Stripe dr
399. ter describes how to configure your server for your environment It assumes that the Version 1 10 root drive is already installed If you are upgrading your system refer to the Version 1 10 Software Release Note for information on updating your root drive to Version 1 10 before proceeding This chapter includes the following major topics a Booting the server setting up the console and changing some system files on a newly installed system a Overview of configuration commands and description of cursor control within configuration forms a Configuring the NetServer software for your network environment The configuration procedures are required after you install a new NetServer or when you change the server configuration Software configuration described in this chapter falls into two categories a Server specific customization Specifying local time zone defining server and interface names and IP addresses setting up services such as NIS routing and configuring TTY ports The Auspex commands for server specific customization are NSconfig and SetupTty a Client specific customization Installing architecture dependent SunOS executables for client workstations that boot from the NetServer and notifying the NetServer which client workstations can boot from the NetServer The commands for client specific customization are SetupExec and SetupClient Although any NFS client can access data on a NetServer only diskless SunOS clients c
400. ter n in the table corresponds to the drive slot and is in the range of slot numbers available in your server the letter p is the name of the partition Table 4 1 CD ROM naming conventions Command Naming convention Examples boot adn where n defaults to 1 b ad 0 1 1 because the CD ROM must be in slot 1 to boot the server eject dumpfs acdn eject acd1 dumpfs dev acd1 dkinfo adn dkinfo ad1 mount the first partition acdn mount rt Ifs o fs 4 2 dev acd1 mnt mount rt hsfs dev acd1 mnt mount any partition adnp mount rt Ifs o fs 4 2 dev adia mnt_adia acdnp adnp mount r hsfs dev acd1a mnt_adia Tape Drives Auspex does not currently ship all the drives in the following tables We include information for previously shipped drives for those sites upgrading system software Follow these guidelines when naming tape drives a Tape drive devices are named rastn where n corresponds to the drive slot and is in the range 1 209 Slot 0 is always occupied by the root disk A tape drive with this name uses the rewind option That is when a program finishes using a tape drive the drive rewinds the tape a Tape drive devices with the no rewind option are named nrastn With this option when a program finishes using a tape drive the drive does not rewind the tape a To indicate tape density append lo loc or c to the device name Refer to Table 4 2 and Table 4 3 for the differences among lo loc and c in 4 mm and 8 mm t
401. ters in the Overall Properties Window 3 Click on the Apply button to apply the change The values apply to all histograms in the Data Display Window Note The new Display Average parameter value does not take effect G immediately after you click on the Apply button It takes effect only in the histogram that is redrawn For example if three histograms are currently displayed changing Display Average does not affect any of the histograms However when you load new statistics to the third histogram the change takes effect in the third histogram Similarly if you redraw all the histograms for example by changing the Hist Placement parameter the new Display Average value takes effect in all of the redrawn histograms Table 9 4 Fields in the Overall Properties Window Field Possible values Functions Auto Draw Hists Auto Draw default Manual Draw Display Display Value default Average Hide Value Automatic Y Max Value default scale Preset Value Hist Placement Columns default Rows Number The histogram is drawn automatically you cannot adjust the size of a histogram by dragging it To resize a frame you can drag the frame that contains a histogram similar to the way you resize a window on the screen The average value of the statistics is displayed below the X axis The average value of the statistics is not displayed The highest value on the Y axis is set to the maximum value of the statisti
402. th virtual partition numbering in subsequent upgrades 2 At the root prompt create a directory for the file system mkdir library 3 At the root prompt mount the file system mount library As usual if the CD ROM does not need to be mounted each time the system boots do not edit fstab skip step 1 in the previous example Simply mount the CD ROM with the mount command as in the following examples the first for a UFS CD ROM and the second for an HSFS CD ROM mount rt lfs o fs 4 2 dev acdl1 library mount rt hsfs dev acdl library Unmount the file system as you would file systems on a disk drive for example umount library Removing a CD ROM From the NetServer You can eject a CD ROM from the drive in one of two ways a Enter the eject acdn command at the shell prompt where n is the slot number for the drive For example eject acdl ejects the CD ROM from the drive installed in slot 1 of the NetServer Depending on the type of CD ROM drive the CD is either in a CD ROM caddy or directly in the CD ROM drive tray If file systems are mounted on the CD ROM drive the NetServer first unmounts the file systems and then ejects the CD caddy a Press the eject button on the CD ROM drive If file systems are mounted on the drive the CD does not eject If the button does not work and there are no mounted file systems on the CD check to see if the jumper setting on the drive is set correctly The drive must be removed fr
403. the average age of all items ejected from the cache since the last screen update Eject count which is the number of items ejected per second from the cache 9 12 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Hit access rate shows the percentage of cache accesses for items already in the cache Hit access count is the number of times per second the cache was accessed The bar following the hit access count shows the total number of cache accesses un indicates a hit and a miss For the data and meta caches the first line shows LRU buffer statistics and the second line shows AGE buffer statistics The LRU buffers are the buffers that are more likely to be used soon They are only ejected from the cache when no AGE buffers are left AGE buffers contain data considered less likely to be referenced again than the data in LRU buffers 4 Age distribution of items in the caches An A represents items that have been accessed within the past minute B represents items that have been accessed within the last two minutes and so on Z represents items that have not been accessed for 26 minutes or more Uppercase letters represent LRU buffers lowercase letters represent AGE buffers SP Statistics Screen The SP Statistics screen displays statistics for one SP Figure 9 6 shows a sample SP Statistics screen 1 0 SP Stats for SPO CPU Utilization OPS sec O to 2000 W
404. the ax_perfmon disk statistics for drives that consistently experience a high volume of write operations If your SP has a Write Accelerator installed and the disk I O activity continues to appear high consider adding a Write Accelerator to another SP board if that board does not already have one Then you can distribute the busy file systems on different SPs Refer to Chapter 7 Write Acceleration for more information 9 48 a Version 1 10 System Manager s Guide W AUSPEX 10 Special Maintenance About This Chapter This chapter describes Auspex specific features and commands that may not be essential for your NetServer s day to day operation but are valuable for system management and reporting server problems to Auspex This chapter discusses the following types of commands or procedures a A NetServer configuration tool the ax_config command that organizes configuration data such as part numbers for field replaceable units and revision numbers for software modules Caution Only Auspex authorized personnel should modify the configuration A file var adm config report Use ax_config only to display configuration information or print out the information This chapter describes only the contents of the report and how to display it a A NetServer utility ax_load_flash to download Flash PROM firmware to the SP and NP boards as well as to 10Base T Ethernet FDDI and MLT 3 SBus adapters installed on the NP
405. the same location hit a request currently in the write cache Queue time is the time spent waiting in the queue Coalesce per second measures the number of write operations that the write cache coalesces Coalescing data improves efficiency because data destined for adjacent disk areas are written with one write operation NV bufs is zero unless a Write Cache board is present NVRAM size shows the size of the write cache Disk activity For each disk controlled by the current SP the number of I O operations is displayed In the bar following the number R represents a read operation and W represents a write operation On this screen the disks are numbered relative to the current SP Disks controlled by an SP V are numbered 0 to 41 Note The NS 7000 200 which supports the SP IV or SP V supports a i maximum of 35 drives For a detailed description of write caching refer to Chapter 7 9 14 a Version 1 10 System Manager s Guide W AUSPEX Virtual Partition Statistics Screen The Virtual Partition Statistics screen displays statistics for virtual partitions controlled by the current SP Figure 9 7 shows a sample Virtual Partition Statistics screen 21 16303321 1996 oO lt 1 0 Virtual Partition Stats wizard on SP1 64 of 256 YP s 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 eoocooccocococococpocococojcococlclmc
406. tics since boot time repeat every 24 lines in the output If no interval is specified only one line of statistics displays to show the statistics since boot time With the interval option the command displays statistics continuously until you interrupt by pressing Ctrl C Also the statistics are presented in different formats depending on whether an interval is specified Managing Network Interfaces a 10 15 The following example shows the output from the ax_netstat i command witha specified interval host0 gt ax_netstat i I ael 2 input ael output input Total output packets errs packets errs colls packets errs packets errs colls 1265620 0 1941468 0 14404 1265620 0 1941468 0 14404 8 0 27 0 0 8 0 27 0 0 33 0 56 0 0 33 0 56 0 0 18 0 14 0 0 18 0 14 0 0 19 0 14 0 0 19 0 14 0 0 21 0 12 0 0 21 0 12 0 0 43 0 40 0 0 43 0 40 0 0 22 0 T3 0 0 22 0 13 0 0 2 0 3 0 0 2 0 3 0 0 1 0 8 0 0 1 0 8 0 0 4 0 15 0 0 4 0 15 0 0 9 0 9 0 0 9 0 9 0 0 2 0 11 0 0 2 0 11 0 0 5 0 34 0 0 5 0 34 0 0 5 0 10 0 0 5 0 10 0 0 2 0 2 0 0 2 0 2 0 0 4 0 23 0 0 4 0 23 0 0 17 0 T5 0 0 17 0 15 0 0 7 0 8 0 0 7 0 8 0 0 4 0 2 0 0 4 0 2 0 0 8 0 9 0 0 8 0 9 0 0 input ael output input Total output packets errs packets errs colls packets errs packets errs colls 1265857 0 1941800 0 14404 1265857 0 1941800 O 14404 6 0 5 0 0 6 0 5 0 0 T 0 2 0 0 1 0 2 0 0 The following example shows the output from the ax_netstat i command without a specified
407. tion in the pull down menu Table 9 6 Pull down menu options displayed in the input file list Option Function Load File Load All Histograms Sum File with Existing File Load File to Threshold1 Delete File Load statistics from a selected file to a selected histogram For example if stats dat and histogram 4 are selected in the File Load Window data from stats dat are displayed in histogram 4 The most recently used parameter is used for the selected histogram Suppose histogram 2 currently displays data from stats dat and uses the HP hp usr_time parameter If you want to use histogram 2 to display data from stats2 dat use Load File from the pull down menu Statistics from stats2 dat replace the statistics from stats dat but the histogram continues to use the HP hp usr_time parameter Similar to Load File except all the histograms defined in the Hist list are affected For example if histograms 1 through 8 are displaying statistics from Stats dat you can select stats2 dat and invoke Load All Histograms to load statistics from stats2 dat The histograms continue to use the parameters used before you loaded statistics from the new file Statistics of the same type but from different input files are added For example if histogram 1 currently contains statistics about HP hp usr_time collected in input file 1 you can add statistics about HP hp usr_time collected in input file 2 The resulting histogram shows the
408. tition Caution Use ax_label on Auspex disk drives only Do not use ax_label for disk drives attached to the SCSI port on the HP Labeling a disk means writing the partitioning table and various disk geometry constants to the label sector sector 0 of the disk and informing the ad device driver of these changes Caution As with format make sure no file systems are mounted on the drive A on which you are going to partition using the ax_label command Also labeling a drive can erase all data on the drive back up the files on the drive before using ax_label if you want to preserve them The syntax for ax_label is as follows ax_label abcdefgh size disk The option preceded by a dash is the partition name Specify the size in MB immediately after each partition name a size of all means that all remaining sectors in the disk volume are assigned to the specified partition You can define up to seven partitions Specify the disk number for example ad2 at the end of the command Note In the ax_label command the partition names are usually entered in i alphabetical order However if you assign a size of all to a partition specify it as the last partition regardless of its name The following is an example for partitioning a 4 GB disk ax_label a 400 b 1580 d 300 e 3200 f 2000 g 500 h all ad5 This example divides ad5 into seven partitions After all sectors are assigned to partitions a b d e f and g the remaining spac
409. tition retains the name of the virtual partition being converted The following example shows how to convert a striped partition vp1 to a one member mirrored partition ax_mconvert vpl The ax_mconvert command modifies etc vpartab to create a new virtual partition for example vp2 to be used as the only member of the newly formed mirrored partition vp1 In this example vp2 is mapped to the physical partitions that vp1 used before the conversion After modifying etc vpartab ax_mconvert invokes ax_loadvpar to load the virtual partition table If errors occur during the loading of the table messages generated by ax_loadvpar appear For more information on ax_loadvpar errors refer to the ax_loadvpar man page The following is the vpartab entry for vp1 before the conversion dev vpl striped size 128k ad3a ad4a ad5a The following are the vpartab entries for vp1 and its member after the conversion dev vpl mirrored vp2 dev vp2 striped size 128k ad3a ad4a ad5a Figure 5 9 illustrates the conversion process Before ax_mconvert After ax_mconvert vp1 partition to be converted vp1 one member mirrored partition KOTA Oooo E 1 4 2 5 3 EEI vp2 newly created partition that uses the same ad7a ad8a ad9a physical partitions as vp1 before the conversion A Ho ad7a ad8a ad9a el e ol a Figure 5 9 Using ax_mconvert to change a partition from striped to mirrored 5 20 a Version 1 10 System Manager s Guide AUSPEX
410. tm outgoing bytes sec NP atm incoming errors NP atm outgoing errors NP atm outgoing discards NP atm outgoing error percentage NP atm Max transmit frame size mtu NP atm Offnet transmit frame size offmtu FP FP cpu busy FP lfs mount FP lfs unmount FP lfs read FP lfs write FP lfs readdir FP 1lfs readdirplus FP lfs access FP lfs commit FP lfs readlink FP lfs null FP lfs getattr Number of errors received by the Ethernet interface Number of errors sent by the Ethernet interface Number of collisions on the Ethernet interface Number of outgoing packets discarded by the Ethernet interface Percentage of errors sent by the Ethernet interface Number of packets processed by the ATM interface Number of bytes processed by the ATM interface Number of bytes received by the ATM interface Number of bytes sent by the ATM interface Number of errors received by the ATM interface Number of errors sent by the ATM interface Number of outgoing packets discarded by the ATM interface Percentage of errors sent by the ATM interface The MTU value on the ATM interface The MTU value on the ATM interface for networks not directly attached to the server CPU utilization on the FP LFS mount operations LFS unmount operations NFS read operations NFS write operations NFS read directory operations NFS V3 read directory plus operations NFS V3 file access check NFS V3 commit flu
411. to etc fstab This section discusses how etc fstab on the NetServer is different from the one shipping with standard SunOS he Recommendation Print out a copy of etc fstab each time you modify the file If a NetServer drive fails having a record of the file systems that reside on each drive makes it easier to recover from the failure Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades Specifying the File System Type The implementation of the LFS file system type requires a modification to entries in the etc fstab file In general file systems used only by the server are configured as the 4 2 file system type for example and usr while file systems exported to clients are configured as the LFS file system type for example home Figure 4 10 shows the default etc fstab file shipped with the NetServer For further details see the fstab 5 man page Mounting File Systems on the FP and HP a 4 19 File systems managed by the FP and exported to clients are mounted as type LFS File systems managed by the HP are of type 4 2 The set UID and set GID bits on program
412. to the drive The drive engages and automatically loads the caddy Close the drive door If the CD is installed directly into the CD ROM drive 1 2 Open the CD ROM drive Place the CD in the CD ROM drive with the CD artwork facing out Make sure that the CD is secured by the three plastic latches at the corners of the CD ROM tray Close the CD ROM drive door Booting the server from the CD 1 Access the PROM monitor The PROM monitor prompt appears HP gt Enter the boot command as follows HP gt b ad 0 1 1 The NetServer displays the following messages Remounting and replenishing etc Remounting and replenishing var Remounting and replenishing tmp AUSPEX CD ROM MAINTENANCE MENU OPTION 1 Format label the root drive 2 NSinstall 3 Maintenance shell Select an option 1 2 3 2 Booting the NetServer a 3 7 AUSPEX W If drive 0 is not labeled as a root drive or the drive is not present error messages appear prompting you with the appropriate action 3 Select an option from the Maintenance Menu Selection 1 displays the Format menu Selection 2 displays a message about adding swap space and then runs NSinstall For example Adding swap on dev ad0b The message appears if you have swap mounted on the b partition of drive 0 See Chapter 6 of this manual for instructions on using NSintall 8 Selection 3 puts you into a maintenance single user mode shell from which
413. triped size 128K AC ad37c ad38c ad39c vp77 Mirrored vp75 Concatenated SYNCED ad30c vp76 Concatenated SYNCED ad3lc vp271 Striped size 128K ad57c ad58c Figure 5 6 Example from using the ax_vpstat command The meaning of each mirrored status is as follows ACTIVE SYNCED DIRTY DAMAGED RESTORING This partition is identical to the other member of the mirror if the mirror has more than one member and no unrecovered write error occurred on the partition The mirror is currently mounted read write or open for writing This status is similar to ACTIVE status but the partition is not currently mounted read write or open for writing This partition is not identical to the other member of the mirror because the system crashed while the mirrored partition was ACTIVE or RESTORING Writes to the mirror are not being performed on this member partition An unrecovered write error occurred on this partition and was detected when the SP tried to access the partition Because of the error this partition is not identical to the other member of the mirror Writes to the mirror are not being performed on this member This status is reset to ACTIVE after you repair or replace the disk and run ax_mrestore to update the data on the partition This partition is not identical to the other member of the mirrored partition but writes to the mirrored partition are being performed on this pa
414. ture 1 3 Hot plugging a drive 4 26 HSFS High Sierra File System 4 5 I O cache memory 9 47 I O monitoring 9 2 I Os reduction of 7 3 Inconsistent file systems 6 3 6 13 Incremental backup 6 26 8 2 init command 1 12 2 19 Inode cache 9 12 Inodes for restored files 6 26 freeing 6 13 number per cylinder group 4 22 Input fields in configuration forms 2 10 Input file list 9 37 installboot command 6 26 6 28 Installing software to a spare root Index 8 A Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX drive 6 31 Interfaces Ethernet 2 14 FDDI 2 14 Intermittent disk errors 6 21 International customer support xx Internet address 2 11 2 14 2 30 Invisible disk on a running system 6 19 Isolating file systems effects on network clients 6 13 error messages in var adm messages 6 15 killing sleeping processes on 6 15 sample procedures for recovering from 6 15 to 6 19 stale file handles seen by clients 6 16 K Kernel alternate file 3 6 executables 2 30 panic messages rebuilding 10 9 reconfiguring 2 5 setting parameters 2 6 Keys question mark 2 10 Backspace 2 10 10 11 Break 10 12 Ctrl C 2 10 Ctrl F 2 10 Ctrl H 2 10 Ctrl L 2 10 Ctrl R 2 10 Ctrl U 2 10 Ctrl W 2 10 Delete 2 10 Down Arrow 2 10 Home 2 10 Left Arrow 2 10 Page Down 2 10 Page Up 2 10 Return 2 10 Right Arrow 2 10 Space Bar 2 10 Space bar 2 10 Tab 2 10 Up Arrow 2 10 Killing sleeping processes on an isolated filesystem
415. u on page 4 36 Do not use the format command for DriveGuard arrays See the DriveGuard Manager s Guide for using ax_label 8 with DriveGuard arrays 2 From the Format menu enter format The format option applies to the disk that you specified earlier The following example shows screen output generated by format format gt format Ready to format Formatting cannot be interrupted and takes xxx minutes estimated Continue y Beginning format The current time is Tue Feb 15 16 58 05 1995 Formatting done Verifying media pass 0 pattern Oxc6dec de pass 1 pattern Ox 6db db6 d Total of 0 defective blocks repaired format gt xxx in minutes varies with the type of drive The process of formatting and verifying can take up to several hours depending on the size of the drive To ensure the integrity of the disk you should allow this process to complete Caution Interrupting the format of a drive causes the drive to enter an A irrecoverable state The drive must be reformatted When the process is finished the disk is formatted By default the entire drive is formatted and two passes of surface analysis are run on the disk If any defects are found in the surface analysis the system repairs them automatically Table 4 10 on page 4 39 lists the approximate amount of time it takes to verify the format of each drive type 4 38 a Version 1 10 System Manager s Guide W AUSPEX Table 4 10 Disk drive fo
416. u su su su su n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 n4 GTa TOGA T O TA TE OARA T Aa TO A Ob Oh Cie S E a T root usr CD Kvm CD Install Networki System_V USys CD SunView_ SunView_ Text Demo OpenWi OpenWi OpenWi User_D Manual TEI C RES Q G ng Us De ROM file ROM file D ROM file CD ROM file CD ROM file ROM file ers CD ROM file mo CD ROM file CD ROM file CD ROM file CD ndows_Users CD ROM file ndows_Demo CD ROM file ndows_Fonts CD ROM file iag CD ROM file ROM file D ROM file D ROM file Debugging CD ROM file SunView_Programmers CD ROM Shlib_Custom CD ROM file Graphics CD ROM file uucp Games CD ROM file CD ROM file Security CD ROM file OpenWindows_Programmers CD ROM This final message indicates that SetupExec is finished 2 26 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX SetupClient Figure 2 10 is a flowchart describing the procedure for configuring client information on the server using SetupClient Do you want server to boot diskless clients Do nothing Yes No Run SetupExec yet Run SetupExec No Is server an NIS slave Yes Change etc ethers etc hosts and bootparams on the NIS master server
417. u risk copying corrupted data or configuration errors to the backup drive which may not be able to boot the system if the current root drive fails Backing Up With a Large Number of Disks a 8 11 Online Backup The backup procedures in the previous sections require that the file system be inactive when being copied to tape This section describes how to back up a file system while it is still active The online backup procedure involves these two major steps 4 Create a virtual partition as a clone of an active file system to be backed up 4 Back up the clone partition to tape Details about dump are not repeated in the following sections See Using the Dump and Restore Commands on page 8 2 for information Cloning a Virtual or Physical Partition A clone is a physical or virtual striped or concatenated partition that duplicates the data in another partition which can also be physical or virtual The partition being duplicated is called the source partition Think of a clone as a snapshot of the source partition which can be resident on another SP Because the clone and the source partition contain the same data immediately after the snapshot is taken the partitions look like the members of a mirrored partition However there are significant differences between cloning and mirroring as described in Table 8 5 Table 8 5 Differences between cloning and mirroring a partition Cloning Mirroring As data is copied fr
418. uide AUSPEX W W AUSPEX y 1 0 System Summary cell Mon Nov 25 13 16 54 1996 CPU Utilization SPO OPs sec O 80 in bar NPO Bytes sec HP 3 usr Rack Rack2 Rack3 Rack4 Rack5 8 sys UAN E Pi edol 197 89 idle df kol emf 0 lf ll fool 0 J E J foil 0 oE gi Desf 1 Iel ll 1 NFS Bytes sec NPO 0 TI 197 NP1 O 0 197 Total NFS Bytes sec l FP LFS OPs sec DataCacheAge FPO 0 ol J NNNNNNNNNN2 FP1 O of 22222222222 0 Total LFS OPs s SP DISK OPS s Hit Disk time SPO OxI ol 0 0 0 ms o ae Figure 9 1 Sample System Summary screen for an NS 7000 200 Series NetServer M 1 0 System Summary wizard Thu Nov 21 13 21 58 1996 CPU Util NFS Bytes sec HP CPU Util NPO Bytes sec SPO 0Ps sec NPO O I 1C 985 14 usr adoL 87 lt 0 80 bar NP1 O I 1C 268 92 sys FOOL 8981 R1 NP2 12C J E 125 762 idle HOOL I NP3 ORT 1 E 346 a oes NP4 OZI l I 0 1s 1724 Total Bytes sec J I LFS OPs sec DataCacheAge FP OZI OL 222222222221 R2 FP1 OZI ol 1 CARARAAAARARAI TE FP2 OZI ol 1 CAAAAAAARARAI 1 FPS O I ol Czzzzzzzzzzz FP4 OZI ol 1 CAAAAAAARARAI S i 0l Total LFS OPs s P oL J JT SP DISK OPS s Hit Disk Time R3E SP 424 L 0 0 0 ms SP1 OZC ol 0 0 0 SP2 OZI
419. uming process and can result in losing changes to many site specific configuration files Before using NSinstall to install the operating system refer to the procedures outlined as follows In most cases you can use one of the less drastic methods and avoid running NSinstall Booting From a Backup Root Disk Use this procedure if you have a backup root disk from which you can boot the server 1 Power off the server 2 Remove the backup root disk from its slot if it is installed in the server 3 Remove the damaged root disk from slot 0 and replace it with the backup root disk 4 Select an unused slot for the damaged root disk 5 Power up and boot the NetServer in single user mode refer to Booting the Server at the Monitor Prompt on page 3 4 D Determine whether the damaged root disk can be repaired using fsck or format 7 Ifthe entire disk is damaged go to step 8 If only the root partition was damaged not the entire disk follow these steps a Re create the root partition using one of the following methods a Run newfs on the root partition and restore the root partition from the most recent level 0 backup Note After you restore from tape the restored files may be on different inodes If you attempt to run an incremental backup on the restored files the backup may not select the files you expect Thus the first backup you run on the restored root disk should be a full level 0 backup a Use the dd 1 command
420. untry or territory Telephone number Country or territory Telephone number Australia 1 800 121 194 Luxembourg 0800 2721 Austria 0 660 7912 Malaysia 800 4509 Belgium 0800 1 0180 Netherlands 06 0222158 Denmark 80 01 03 60 Norway 800 1 1294 France 0800 91 21 32 Philippines 1 800 116 0005 0800 26 38 22 Germany 0130 81 8306 Poland 00 800 441 1220 Hong Kong 800 4803 Spain 900 95 4445 Ireland 1800 55 3343 Sweden 020 797326 Israel 177 440 9285 Switzerland 0800 55 1339 Italy 1678 73541 Taiwan 0080 14 9580 0044 22 12 2414 0038 13 0374 Japan United Kingdom 0800 28 7739 Korea France has two toll free numbers The number in is directed to the Paris office during the working day International customers who purchase Auspex products from an authorized distributor or reseller should contact that distributor or reseller for support To comment on the content of this online help facility send email to Auspex Technical Publications techpubs auspex com xx a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX World Wide Web For more information about Auspex Systems Inc and its products use the following resource location on the World Wide Web http www auspex com Printing Online Documents You can print individual pages or the entire contents of the online manual To print individual pages use the ax_docs command and choose Premier Software Series Menu in the Main Menu window Choose the lt opf product gt d
421. up another menu with choices for all of the optional products offered with this release Clicking on one of the optional product choices brings up a window with the document for that product a System Hardware Menu This entry brings up another menu with entries for all the hardware manuals and the Hardware Release Note Click on the entry for your system such as the NS 7000 Model 700 Series to see the manual for that system a Field Service Menu This entry brings up a menu of field service documents including FRU documents and the Field Service Guide a Auspex Homepage This entry connects to the Auspex home page over the World Wide Web provided you have an Internet browser running on your system a Help button The help button a large question mark explains how to use the Acrobat Reader window menus and icons Using the System Manager s Guide as an example note that when the window containing that document appears a table of contents sidebar entry appears in the window The entries are called bookmarks A bookmark with more levels of table of content entries can be expanded to other levels by clicking on the arrow for that entry Clicking on the entry B 6 a Version 1 10 System Manager s Guide itself takes you to that part of the document Figure B 2 shows a portion of the System Manager s Guide window with the bookmarks on the left The Main Menu choice takes you back to the documentation Main Menu D Main Menu D
422. use permanent damage to the drive If you commented out entries in etc vpartab in step 1 remove the signs for those entries Run ax_loadvpar if you edited etc vpartab Use dkinfo to verify that the disk has the correct disk label If it does not use format to repartition the disk Use newfs and fsck to create and check the new file systems to be restored Restore data to the new drive from the most recent backup tapes Remove the signs you added in etc fstab in step 1 so file systems can be mounted on the new drive Mount and export the file systems existing on the new drive so network users can access them again 6 6 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Using the System Without the Failing Drive The following procedure describes how to make the system available to network users as soon as possible without replacing the failing drive Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades 1 Comment out the entries in etc fstab and etc vpartab that reference the bad drive 2 Use ax_hot_plug or ax_remove_device to remove the drive If the
423. usually modifies the configuration information to facilitate future troubleshooting efforts Non interactive Mode Use ax_config in non interactive mode if you want to display configuration information or write information to a file so you can transmit or print it Displaying Configuration Information To display configuration information enter the ax_config command as follows ax_config d ax_config displays configuration data obtained from the software and hardware probe To include data from var adm config report enter the following ax_config da If the configuration file does not exist the following message appears ax_config usr adm config report not found These commands however do not write data to the config report file Starting ax_config a 10 5 Copying Configuration Information to a File To save the configuration information to a file enter the ax_config command as follows ax_config o filename da This command saves the hardware data from config report hardware probe data and software probe data to a file which can then be transmitted or printed You can also manually edit this file to add information about hardware components that ax_config cannot probe 10 6 A Version 1 10 System Manager s Guide AUSPEX W W AUSPEX Flash PROM Download Utility This section describes the flash PROM download utility used to install new versions of PROM code on NetServer processor boards and SBus card
424. utables from your Sun supplier a Each architecture dependent executable uses a variable amount of disk space depending on the release The export file system is installed by default on a partition that allows enough space for about two executables If you need room for more executables move export to a larger partition before running SetupExec If SetupExec runs out of space trying to install the executables it generates an error message and terminates oz Recommendation Installing export on a nonroot disk allows you to replace the root disk if necessary without having to restore export a If the CD ROM containing the client executables is on a remote host execute the mount command on the remote host before running SetupExec on the NetServer mount rt hsfs dev sr0 cdrom if the remote host is a non Auspex server or mount rt hsfs dev acdn cdrom if the remote host is an Auspex server where n is the slot number for the CD ROM a When using either a remote tape or a remote CD ROM SetupExec accesses the remote host using rsh The hosts file rhosts on the remote machine must include the name of your NetServer and have the correct root permissions a Run SetupExec before you run SetupClient If you add a client with SetupClient and the appropriate architecture dependent executables are not found SetupClient generates an error message and terminates The Sun architectures used on various Sun workstations are listed in Tab
425. ution media from Sun or other vendors The purposes of usr openwin are as follows a Users with X terminals or workstations on the network can log in to the NetServer to run X applications For example the xdm files in usr openwin provide a login form for X terminal users on the network to log in to a NetServer After they log in they can run X programs such as xterm or view the online NetServer information Note If you want to read the man pages for the OpenWindows commands D and files shipped with the Auspex software set MANPATH to usr openwin man using the following syntax setenv MANPATH usr openwin man a usr openwin contains font files and remote configuration files for an X terminal that is used as the system console Because usr openwin is NFS exported an X terminal can fetch the font files and download the configuration file from a NetServer Figure 4 7 shows the structure of the usr openwin file system usr openwin bin demo etc gt share etc include gt share include lib lost found man gt share man modules share Figure 4 7 The Auspex usr openwin file system 4 16 a Version 1 10 System Manager s Guide W AUSPEX export file system Figure 4 8 shows a sample export file system which contains directories that are exported to client systems including the architecture dependent executables in export exec The directory structure allows you to support multiple versions of each
426. v 2 4 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX NetServers and UDP Checksumming On the NetServer no flag needs setting for UDP checksummed packets from clients Auspex NetServers check the UDP flag in a packet and deal with the packet accordingly Clients can send packets checksummed or not the NetServer correctly deals with the packets and then sends replies in kind Modifying Site Specific Files In addition to the preceding procedures you may also need to customize two types of site specific files a The SunOS kernel Reconfigure the kernel in two situations if you need to add devices that are currently not supported by the generic kernel usr sys aushp conf AUSPEX1 or if you change the site specific parameters in the kernel Refer to Sun s Network and System Administration manual for more information on reconfiguring the kernel Refer to Kernel Parameters on page 2 6 for information about site specific kernel parameters See the section Rebuilding the System Kernel on page 10 9 for reconfiguration instructions a Various files in the etc directory Table 2 1 lists the files that contain site specific information such as mail aliases and trusted hosts The table also describes the purpose of each file and lists man pages containing more information Table 2 1 Key files that contain site specific information File name Purpose of the file etc aliases Defines mail addresses o
427. vailable file system and the error recovery procedure depend on these factors a Cause of the problem a When the problem occurs during system reboot or when the system is up and running in multiuser mode a Whether the file system is isolated as a result of the problem that is whether the FP takes the file system offline until the problem is corrected 6 2 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Disk Problems at System Reboot When the system is booting it detects all the disk drives that have been installed The power up messages list disks that are visible to the system For example server_name vmunix ad20 lt Seagate 4GB cyl 3605 alt 1 hd 21 sec 110 gt server_name vmunix ad22 lt Seagate 4GB cyl 3605 alt 1 hd 21 sec 110 gt server_name vmunix ad23 lt Seagate 4GB cyl 3605 alt 1 hd 21 sec 110 gt In this example a disk installed in slot 21 is not visible The following describes how the system reacts to a disk that is invisible at boot time a Ifno file system is mounted on the disk the system can successfully reboot a If the disk is a member of a virtual partition reboot stops and the system displays the following error messages Virtual partition download failed help WARNING file systems have NOT been checked After fixing the virtual partition download problem run fsck p to check the file systems befor going multi user Then the system enters single user m
428. ver for guidelines on handling drives 2 Edit etc vpartab and etc fstab to comment out entries in the table that refer to the drive to be removed Use the etc shutdown command to shut down the operating system 4 Remove the drive 5 Reboot your system Using ax_hot_plug to Add and Remove Multiple Drives You can use ax_hot_plug to add and remove multiple drives with a single command Note For information on adding removing or replacing HDDA drives refer i to your hardware manual Run ax_hot_plug only on the local console If you try to run it from a remote login session the following error message appears ax_hot_plug must be on dev console to use this command Caution Never run ax_hot_plug on an SP that is formatting a drive or A running ax_mrestore or ax_clonefs Doing so may hang the system 1 Enter the ax_hot_plug command specifying the slot number s of the drives to be added or removed For example if you wish to add devices in slots 15 through 17 and remove a device in slot 9 enter the following ax_hot_plug add 15 17 remove 9 The system suspends the activity on the SP so no I O operations are in progress on the SCSI bus spins down the drives prompts you to add or remove the first specified Removing a Drive a 4 33 AUSPEX D drive in this example adding a drive to slot 15 and tells you to type a carriage return when you finish adding or removing the drive 2 When prompted at the system c
429. ves a disk drives a 4mm 8 mm or 1 4 inch tape drives a CD ROM drives a DLT4000 tape drives supported in a DLT4000 drive rack only If you do not have enough vacant slots to accommodate the number of drives you want to add you may have an additional drive rack or HDDA drive drawer installed provided your server can accommodate more racks or an expansion cabinet This upgrade can be arranged through your Auspex representative For a detailed description of the available drive configuration options refer to the hardware manual for your NetServer The instructions in this section assume that your NetServer has enough vacant drive slots to accommodate the drive s to be installed Two ways exist to install a new drive depending on whether the NetServer s operating system is booted The following procedures provide instructions for using both installation methods Adding or Replacing a Drive a 4 25 AUSPEX D Installing or Replacing a Drive After Booting This section explains how to install a new drive or replace an existing drive while the NetServer is powered on and the operating system is booted This procedure does not apply to the root drive To replace the root drive follow the procedure for replacing a disk drive when the operating system is shut down Note If you are installing or replacing a CD ROM drive by convention i Auspex recommends installing the first CD ROM in slot 1 Install additional CD ROM drives in a
430. ware and software information in Adobe PDF format and an Acrobat reader utility to read the files For a more detailed list of Auspex specific commands and utilities refer to SunOS Enhancements on page 1 12 Compatibility with Compatible with UNIX SunOS ONC NFS TCP IP Ethernet VME SCSI X11 industry standards OpenWindows and FDDI standards Compliant with POSIX 1003 1 FIPS 151 a SVID Issue 2 and X Open Portability Guide Version 2 Distribution medium CD ROM For hardware features such as the maximum number of network interfaces supported amount of I O cache memory and storage capacity refer to the hardware manual appropriate to your NetServer model 1 4 a Version 1 10 System Manager s Guide AUSPEX W W AUSPEX NetServer Hardware Architecture The combination of processor boards and drive subsystems within a NetServer varies from one configuration to another However on all NetServer models the Auspex FMP architecture distributes performance limiting I O functions to multiple dedicated processors bypassing the UNIX operating system to significantly reduce software overhead As a result Auspex s FMP architecture delivers a dramatic improvement in network I O performance An example of a processor board is the Network Processor NP An NP board contains one CPU for protocol processing and another for file system processing When this guide uses a processor name for example Network Processor
431. wearing out or the head may need cleaning Analysis Write operations are retried because the drive needs to skip the bad spots on the tape Solution If the number of errors per megabyte of data transferred exceeds 20 clean the drive heads or switch tape brands Read Error Command tar xvf dev rst2 Error messages vmunix ast2 14 recoverable errors per MB transferred vmunix ast2 the tape may be wearing out or the head may need cleaning Analysis Read operations detect bad spots on the tape Solution Ignore the error message unless because of the error the drive cannot read data from the tape Write Failure Error Command dump Oucbsdf 100 6700 65000 dev rast2 dev ad2c Error messages DUMP Tape write error 5098 feet into tape 1 vmunix ast2 file mark write failed vmunix ast2 error sense key 0x3 media error vmunix sense f 00 4300011200000000 0011000 23 1 21 Analysis The tape has a very large bad spot that it fails to skip Sometimes this occurs because the tape drive heads are dirty Solution Clean the heads and try again using the same tape If the error happens again use a new tape Tape Jam Error Scenario Tape cannot be ejected from the drive after you press the release button on the drive Analysis Tape tension is incorrect confusing the tape drive logic Solution Remove the tape drive and replace it in the drive slot to reset the tape drive logic When the drive executes the self test repeatedly press
432. which changes the existing virtual partition to a single member mirrored partition with the new partition as its member Contents of the virtual partition being enlarged are copied to the new partition Note ax_mconvert used by ax_expand must create a new virtual partition to be the member of the mirrored partition If all virtual partition numbers have already been used ax_mconvert generates this error message ax_mconvert all VP devices in use Figure 5 8 illustrates what happens to a striped partition when you enlarge it The procedure for enlarging a striped partition follows the figure 5 16 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Before ax_expand After ax_expand vpi partition to be enlarged vp1 one member mirrored partition vp2 new virtual partition gt EEEE BRIERE TE Le Ee unused physical partitions Figure 5 8 Expanding a striped virtual partition using ax_expand s To enlarge a striped partition follow these steps Caution The following line appears at the top of the etc vpartab and etc fstab A files on systems running Version 1 10 VP256 ENABLED do not delete Do not delete this line Also it must always be the first line of the file If using RCS control add any text below the first line Deleting or moving the first line causes problems with virtual partition numbering in subsequent upgrades Create a new striped or concatenated partitio
433. wing list of drives AVAILABLE DRIVE TYPES 0 drives 0 10 are not supported in the NetServer 11 HP 663MB 12 HP 1GB 13 HP 1 35GB 14 ICROP 1 76GB 15 ICROP 1 76GB NON ALIGNED 16 HP 2 0GB LT ICROP 2 0GB 18 HP 2 1GB T9 ICROP 3GB 20 Seagate 4GB 21 Seagate 4GB NON ALIGNED 22 ICROP 4GB 23 ICROP 9GB Using the Format Command a 4 37 AUSPEX W 24 Seagate 9GB 25 Seagate 9GB NON ALIGNED 26 other Specify disk type enter its number 20 Auspex uses the formatted capacity to define the disk type Normally you only need to accept the default drive type because format automatically recognizes the type for the selected drive However if you plan to use the dd command to copy contents between two drives of different capacity specify the size of the source drive when you use the type option on the target drive to avoid confusing the operating system For example if you plan to copy from a 1 GB drive to the target drive which is 1 35 GB select HP 1GB from the menu for the target drive so it is formatted as a 1 GB drive Formatting a Disk Using the Format Option Low level formatting of a disk is normally not necessary Drives are preformatted at the factory before shipment Caution This procedure erases all existing data on the disk Be sure to back up A the data before formatting the drive 1 If you have not already done so follow the steps outlined in Accessing the Format Men
434. x Environment a 1 11 SunOS Enhancements Each NetServer runs a modified SunOS kernel that communicates with the intelligent Auspex processors to enhance NetServer performance Most UNIX services come directly from SunOS for complete compatibility However Auspex makes some changes to the Sun operating system to integrate the Auspex processors into the standard operating system o Note The majority of the SunOS standard utilities and files remain unchanged This publication documents the changed portions of SunOS only it does not document the unchanged portions Consequently it is very important that you have the SunOS documentation To obtain the latest documentation contact Sun Express at 1 800 USE SUNX or Auspex The most significant changes to SunOS are listed in Table 1 2 Not all of the changes are described in the rest of this guide because some changes for example the ax_statd 8 daemon are transparent to the system administrator For information about the more visible changes see the appropriate man page Table 1 2 Changes to SunOS Types of changes For further Description information see Added or modified Three added daemons coordinate the processors Man pages for daemons ax_startup 8 ax_errd 8 and ax_timed 8 These ax_startup daemons like init 8 must run even in single user ax_errd mode ax_timed ax_statd Additional daemons supporting the NetServer in ax_statd2 multiuser mode are ax_s
435. y Note The difference between uppercase and lowercase letters in the data D cache age bar is insignificant as long as some buffers are greater than J or 4j the NetServer has enough memory Refer to the ax_perfmon 8 man page for details A DoI need more NPs A NetServer might need another NP for two reasons first individual networks are too heavily loaded and another interface is required to attach additional networks to the NetServer second an NP is CPU bound The top section of the System Summary screen shows CPU utilization for all processors so you can easily identify a CPU bound NP If an NP s CPU utilization is consistently above 80 percent the processor may be a bottleneck You may avoid adding an extra processor by rearranging the way networks are attached One NP may have two heavily loaded networks while another has two lightly loaded networks Putting one lightly loaded network and one heavily loaded network on each processor may eliminate the bottleneck To identify heavily loaded networks check the bytes per second bar on the NP Statistics screen A Do lI need more SPs Add more SPs if you are adding more devices The SP IV and SP V support 42 drives each Adding SPs may also be indicated if SP processor utilization is consistently above 80 percent a Are my file systems and client workstations properly distributed across the available disk drives The disks in a NetServer can handle up to 70 to 8
436. y a 8 5 Exabyte 8500 format is used For more information on the relationship between device names and data compression refer to Table 4 2 and Table 4 3 on page 4 3 Note When using data compression you may or may not be able to store to D the tape the amount of data indicated in Table 8 2 depending on how compressible your data is The table lists the capacity of the 112 meter tape in an Exabyte 8200 drive if the device name is rastnlo Because this drive does not use data compression do not include lo in the device name for an Exabyte 8200 tape drive Table 8 2 Tape drive data capacities Tape and drive type rastnlo rastnloc rastn rastnc 4 mm 60 meter 1 3 GB Not applicable 2 6 GB WangDAT 2 6 GB WangDAT 2600 WangDAT group compression or DDS 1 3 GB WangDAT DDS 4 mm 90 meter 1 95 GB Not applicable 3 9 GB WangDAT 3 9 GB WangDAT 2000 Standard group compression or DDS 1 95 GB Standard DDS 8 mm 112 meter 2 3 GB Not applicable 2 3 GB Not applicable Exabyte 8200 8 mm 112 meter 2 3 GB Not applicable 4 7 GB Not applicable Exabyte 8500 8 mm 112 meter 2 3 GB 4 7 GB 5 GB 10 GB Exabyte 8505 8 mm 112 or 2 3 GB 4 7 GB 5 GB 10 GB 160 meter 3 5 GB 7 GB 7 GB 14 GB Exabyte 8505XL Quantum 10 GB 20 GB 20 GB 40 GB CompacTape IV 1800 feet DLT4000 1 4 inch 600 feet 150 MB Not applicable 150 MB Not applicable QIC 150 1 2 inch 2400 feet 40 MB Not applicable 150 MB 6 250 BPI Not applicable 1 600 BPI
437. y have lost some of the server configuration information To reconfigure your NetServer run SetupTty SetupExec and SetupClient Note If your server has been using a customized kernel remember to reconfigure the kernel after running NSinstall This ensures that the kernel contains the devices and software options appropriate for your environment 6 34 a Version 1 10 System Manager s Guide W AUSPEX Write Acceleration A A A About This Chapter This chapter covers the following topics Overview of write acceleration Restrictions that apply to the write cache A description of the write cache states and the commands used to manage those states Instructions for enabling the write cache Instructions for disabling the write cache Instructions for managing the write cache including error recovery procedures Write Acceleration a 7 1 Write Acceleration Overview Auspex NetServers offer an optional Write Accelerator to accelerate client NFS write operations The Write Accelerator is a daughter board with nonvolatile cache memory which fits onto the SP board The terms write cache and Write Accelerator are used interchangeably in the NetServer s screen messages and in this guide The Write Accelerator offers these advantages that make it easy to manage a Write acceleration is available on an individual file system basis a The Auspex Performance Monitor statistics show write cache activities a V
438. y not be a member of more than one virtual partition Each SP can manage a maximum of 256 virtual partitions NetServer software can support up to five SPs and 1280 virtual partitions All members of a virtual partition must be managed by the same SP If a disk partition is a member of a virtual partition no disk partition that overlaps that member can be part of another virtual partition Operations that open a drive or a partition with write permission cannot be run on a virtual partition or on a disk partition that is a member of a virtual partition For example you cannot run the format command on a virtual partition nor can you format a disk partition that is a member of a virtual partition You must format and label the disk drive before creating any virtual partitions on it Similarly you cannot run newfs ax_clonefs or NSinstall on a disk partition that underlies a virtual partition For example if you try to run newfs on ad8c while ad8a is a member of a virtual partition an error message similar to the following appears dev rad8c cannot create Device busy When a member of a virtual partition starts at the beginning of a disk that is if the member is partition a c d or e the first 8 KB of disk space is not used as part of the virtual partition in order to preserve the disk label The physical partition c the complete disk is one cylinder larger than any combination of other physical partitions whose size theor
439. you can run commands with root privilege For example select this option if you need to repair a damaged root disk refer to Recovering a Damaged Root Disk on page 6 26 To return to the Maintenance Menu type exit on the command line This completes the procedure for booting the server from the CD Rebooting the Server to Multiuser Mode If the server is currently in single user mode use the reboot command to reboot the server without first going to monitor mode The reboot command executes the sync command to complete all the disk writes loads the kernel into memory and uses fsck to check unclean file systems before bringing up the server in multiuser mode If the server is currently in multiuser mode do not use the reboot command Instead use shutdown to notify users of the reboot and bring the system down to single user mode Then use reboot to boot the system to multiuser mode DataGuard and Rebooting A reboot after a crash on a system running DataGuard can either interrupt or maintain functionality on processors other than the HP a If you used the lt Break gt key to get to the PROM monitor prompt after a crash use the hpboot d command to reboot the HP only Without the d option all the processors will reboot losing the DataGuard protection of uninterrupted functionality a If you use hphalt to get to the PROM monitor prompt after a crash be sure to use the d option with hphalt To reboot the HP only use h
440. your NetServer s hardware manual for more information 3 Set up the console terminal serial parameters Refer to your NetServer s hardware manual for more information Booting the NetServer This section gives you enough information about booting the server so you can get started with the server configuration commands For more information on booting refer to Booting the NetServer on page 3 2 1 After you power on the server it displays the boot messages on the console screen and boots in multiuser mode by default Refer to the NetServer s hardware manual for information on how to power on the server 2 When the server displays the login prompt log in as root Press Return when it prompts you for the password 3 After you have logged in enter the shutdown command without arguments to go to single user mode 2 2 a Version 1 10 System Manager s Guide AUSPEX QD W AUSPEX Run NSconfig to Customize Your Server for the First Time In single user mode run NSconfig to configure your server For overall information on server configuration refer to NetServer Configuration Commands on page 2 9 for information on the NSconfig command refer to NSconfig on page 2 10 After you are finished with NSconfig continue with the procedure in Changing the Root Password on page 2 3 Figure 2 1 outlines the procedures after you run NSconfig Run NSconfig See NSconfig on page 2 10 Bo
441. ystem in the manner of fasthalt h Execute halt k Simulate shutdown of the system Do not actually shut down the system l Send the shutdown message only to users logged in to the system not to users who have file systems mounted from the system This option causes the shutdown command to complete in a shorter time pe Recommendation When using the l1 option be sure to notify the users who have mounted file systems from the server before you shut it down n Avoid running etc rc shutdown and executing sync rc shutdown is an Auspex script that attempts to unmount all the mounted LFS and UFS file systems except usr and Because this option avoids unmounting the file systems if an LFS file system was active before the reboot it is not marked clean and is checked by fsck at reboot This option is used during emergencies r Execute reboot Commands for Shutting Down the NetServer a 3 9 p processor Shut down only the named processor if processor is the HP time The time that the system is shut down warning message The message you want to broadcast to users who have files mounted on this server explaining why the server will be shut down when it will be back up and so on If you specified Yes in the Notify Auspex on reboot field in the NSconfig form your NetServer automatically sends electronic mail to Auspex at reboot The mail messages allow Auspex to be informed of NetServer reliability at customer sit
442. ystem so it does not affect the operation of the server When you are finished repairing the file system use ax_fsutil 8 to put it back on line The ax_expand 8 command enlarges a file system on a concatenated or striped virtual partition fsck 8 allows you to repair a file system even when it is mounted provided the File Processor has taken the file system offline ax_kill 8 kills all processes sleeping on an isolated file system It is useful when the isolated file system can not be repaired by fsck and added back online Once the processes are killed you can unmount and replace the bad disk that caused the file system isolation 1 14 a Version 1 10 System Manager s Guide Mounting File Systems on the FP and HP on page 4 18 Maximum Size of a Virtual Partition on page 5 3 Online Backup on page 8 12 File System Isolation on page 6 10 Repairing an Isolated File System with fsck on page 6 13 Restoring a File System that Cannot Be Repaired on page 6 15 Man pages for mount exports ax_expand AUSPEX W W AUSPEX Table 1 2 Changes to SunOS Continued Types of changes Description For further information see System administration tools Upon a system crash the getcores sh 8 command compresses and fars all core files for delivery to Auspex When enabled the reporter sh 8 command prepares and mails site reports to Auspex Customer Service

Version 1.10 System Manager`s Guide

Contents

Download Pdf Manuals

Related Search

Related Contents