Home

Power HawkTM Series 900 Diskless Systems

1. he ge Flash Boot cete ec Rees Remote File Sharing our ER Res SWAP Space nice ia RE RE PE HUP ER Configuring Diskless Loosely Coupled System Hardware Prerequisites Disk Space Software Prerequisites Licensing Chapter 2 Netboot System Administration Configuration Overview Installing a Loosely Coupled System Installing Additional Boards SBC Client Board Modifying an Existing Automatic Netboot Configuration Client Configuration eee The Client Profile Required Parameters Required NFS Related Parameters Hosts Tables oce EE ds Configuring Clients Using netbootconfig Creating and Removing a Client Configuration Subsystem Customizing the Basic Client Configuration Modifying the Kernel Configuration 1 mknetbstrapz cule i A sr HER pU es Contents c lil 1 1 1 2 Power Hawk 900 Series Diskless Systems Administrat
2. 1 3 SBC Client Board Configuration 1 3 Modifying an Existing Automatic Netboot Configuration 1 12 Client Configuration Rp ee RA 1 13 244 1 Tbe Client Profile File tee mem Det 1 13 2 4 1 1 Required 1 13 2 4 1 2 Required NFS Related Parameters 1 14 24 1 3 Hosts Tables ps 1 15 2 4 2 Configuring Clients Using netbootconfig 1 16 2 4 2 1 Creating and Removing a Client Configuration 1 16 2 4 2 2 Subsystem Support 1 17 Customizing the Basic Client Configuration 1 18 2 5 1 Modifying the Kernel 1 18 2 5 1 1 kernel modlist add eese esses 1 19 2 5 1 2 ciones Rene RR B e 1 19 2 5 53 config utility coelo taa 1 19 2 5 4 1dtuneob ieu i trem eee um ES 1 20 2 5 2 Custom Configuration 1 20 2 5 2 1 S25client and KOOclient rc Scripts 1 23 2 5 2 2 memfs inittab and inittab 1 23 2 5 2 3 Vfstab Table ss is sisssetisz e eee edes 1 24 2 5 24 kernel modlist add 1 25 2 5 2 5 memfs files add 1 25 2
3. The sleep 3 line above will provide time for you to abort out of the Flash autoboot sequence with a Ctrl c keyboard sequence should you need to make any configuration changes to the client To test the Flash boot configuration with autoboot configured issue the ASTRix command reboot cr The board should reset and enter STAR and STAR should then autoboot ASTRix ASTRix should execute the nvram startup script which should boot the client from Flash If there is a problem with the Flash autoboot configuration then you may abort the ASTRix startup script with a lt Ctrl gt c keyboard sequence and make any required changes If the board is configured to manually boot the system without using an ASTRIx startup script then test the Flash boot configuration with the fol lowing ASTRix command boot 5 cr ASTRix should load and boot the client boot image from Flash 3 4 The Raw Write Method To convert the netboot client into a Flash Raw Write booted client follow the procedure below This procedure assumes that the user has already setup and booted the client suc cessfully as a netboot client using procedures described in the Netboot System Adminis tration chapter 1 Reset or power cycle the client board and wait for STAR to autoboot into ASTRix If an automatic netboot startup script previously been setup to netboot the client abort the startup script using a Ctrl c keyboard sequence C
4. The idtuneobj 1M utility can be used interactively or it can process an ASCII command file that the user may create and specify Although the unix kernel need not be rebuilt the tunable should be modified in the client s kernel configuration see the config utility section above to avoid losing the update the next time a unix kernel is rebuilt Refer to the idtuneobj 1M online man page for additional information 2 5 2 Custom Configuration Files 2 20 The files installed under the usr etc diskless d cluster conf custom conf directory may be used to customize a diskless client system configuration In some cases a client s configuration on the File Server may need to be removed and re created This may be due to file corruption in the client s virtual root directory or because of changes needed to a client s configuration In such cases the client configuration described by these files may be saved and used again when the client configuration is re created The s option of netbootconfig 1M must be specified when the client configuration is being removed to prevent these files from being deleted The custom files listed below and described in depth later in this section are initially installed under the nfs and emb directories under the usr etc diskless d cluster conf custom conf directory Some of these files are installed as empty templates while others contain the entries needed to generate the basic diskles
5. dev etc tmp or var Any files in tmp and var tmp are destroyed when the client system is rebooted A dash jn this field may be used to indicate that the path name is the same as that specified for the Path on server field Options a The always option Update the file or directory each time this table is processed o The once option Install the file or directory only if it doesn t already exist Some example vroot files add entries are shown below Example 1 This example specifies that the files contained in the directory home me test dir on the File Server system should be copied into the client s virtual root directory client virtual root users me test dir whenever mknetbstrap 1M processes this file the a option home me test dir users me test dir a Example 2 This example specifies that the single file home me timer c located on the File Server system should be copied into client virtual root users me timer c whenever mknetbstrap 1M processes the vroot files add file the a option 2 27 Power Hawk Series 900 Diskless Systems Administrator s Guide home me timer c users me timer c a Example 3 This example specifies the single file etc app11 on the File Server system should be copied to the client virtual root etc appll if the target file does not already exist in the client s virtual root directory o option etc appll o The following are some addit
6. usr bin devefg 0755 usr bin cut 0755 sbin mknod 0755 usr bin mkdir 0755 sbin fsck 0755 etc fs ufs fsck 0755 etc fs xfs mount 0755 etc fs ufs mount 0755 sbin df 0755 Fh Fh Fh Fh Fh Fh Fh Fh Fh Fh Embedded client systems do not have access to the kernel configuration directory which is needed to generate the device node entries However the device node must be created on the client system because it s minor number carries information that is unique to the running system For this reason special steps must be taken during client boot up to create the device nodes The sample script below will do the necessary steps to add a local disk on an embedded client This script may be found on the File Server system under the path usr etc diskless d sys conf memfs d add disk sh Note that you must set the variables FSTYPE and PARTITIONS to the appropriate values etc clients lt client dir gt custom conf mkprivate S25client cat usr etc diskless d sys conf memfs d X add disk sh gt gt S25client vi S25 client Verify that the script illustrated on the next page after step 5 was appended to S25client and set the variables FSTYPE and PARTITIONS Generate a new netboot image mknetbstrap r all client Start up script to mount a local disk on a client configured as embedded The variables FSTYPE and PARTITIONS should be set to the appropriate values To be able to run this script
7. CPU 1 standing by This is the example NVRAM startup script Synergy Microsystems ASTRix Rev 1 01 06 Jan 29 2003 19 42 00 The above is the default ASTRix command prompt NOTE Some of the following steps discuss creating or modifying files that are located in the nvram directory when running under ASTRix The nvram directory is created by ASTRix when ever it is started up and the contents of the NVRAM filesystem are cop ied into this directory Users can save up to 24KB of files in the NVRAM filesystem Files placed in the nvram directory are auto matically copied to the filesystem whenever the boot reboot star or halt commands are issued Thus the files in nvram that are saved in the NVRAM filesystem are preserved across resets and power shut down allowing users to retain files from one boot to the next 8 If you do NOT wish to enable ECC checking in SDRAM memory and you have already disabled post script execution in step 6 above then skip ahead to step 9 The rest of the ECC configuration involves setting up two post scripts The nvram post script is executed by STAR after reset or power on and the nvram post05 script is executed at the end of the ASTRix boot 5 com mand sequence just before the client netboot image execution begins To create the nvram post script issue the following ASTRix commands cd nvram cr rm post lt cr gt ignore any error messages
8. This section describes the steps using netbootconfig 1M for creating the environment on the File Server that are necessary for supporting loosely coupled netboot diskless clients Major topics described are client profile files Client Configuration Using netbootconfig page 2 16 Information about each client is specified in a client profile file The system administrator creates and updates these profile files for each netboot client and then invokes netbootconfig 1M to create on the File Server system the file environment necessary to support a private virtual root directory and a private boot image for each client 2 4 1 The Client Profile File 2 4 1 1 For each client SBC installed in the cluster a client profile file must be created in the etc profiles directory This section explains the various parameters that are contained in a client profile file All of the parameters located in a client profile file are specific to that one client SBC You should use netbootconfig 1M to print out a starting template of a client profile with the P option which defaults to P client So for example to create a client profile for a client with a hostname of wilma do the following netbootconfig P gt etc profiles wilma vi etc profiles wilma You must then update the resulting client profile file modifying the parameter values in the file to fit the specific characteristics of that loosely coupled netboot cli
9. command language These scripts are available to both NFS and embedded type client configurations Since embedded configurations run in init level 1 and NFS configurations run in init level 3 the start up script is executed from a different rc level directory path depending on the client configuration Any changes to these scripts are processed the next time the mknetbstrap 1M utility is invoked on the File Server For embedded clients a new memfs cpio image and a new boot image is generated An embedded client must be rebooted using the new boot image in order for these changes to take effect For NFS clients the modified scripts will be copied into the client s virtual root and are accessed by the client during the boot process via NFS Therefore the boot image does not need to be rebuilt for an NFS client and the changes will take effect the next time the system is booted or shutdown These scripts may be updated in one of the two subdirectories nfs or emb under the usr etc diskless d cluster conf custom conf directory so that the changes apply globally to all clients If the customizing is to be applied to a specific client the customized rc file should be created in the etc clients client profile filename gt net custom conf directory If there is already an existing shared customization file and those customizations should also be applied to this client then a private copy of the shared rc file should be created with the m
10. error correction code ECC FDDI flash autobooting flash booting Glossary 4 A generic term applied to coding schemes that allow for the correction of errors in one or more bits of a word of data Fiber Distributed Data Interface The process of booting a target from an image in its Flash memory rather than from an image downloaded from a host Flash booting makes it possible to design targets that can be separated from their hosts when moved from a development to a production environ ment See definition for flash autobooting flash burning flash memory FTP ftp File Server function host host board host name host system Glossary The process of writing a boot or other image into a Flash memory device On client Power Hawk Series 900 boards this is usually accomplished with ASTRix ffsw or fw com mands A memory device capable of being occasionally rewritten in its entirety usually by a special programming sequence Like ROM Flash memories do not lose their contents upon power down The File Transfer Protocol is used for interactive file transfer The File Server has special significance in that it is the only system with a physically attached disk s that contain file systems and directories essential to running the Power MAX OS The File Server boots from a locally attached SCSI disk and provides disk stor age space for configuration and system files for all clients clients depend
11. hour field manually to adjust the hour to fit your own local timezone For example to set the date to March 14 2006 and the time to 1 15 00 P M in the EDT timezone enter the following STARO gt date 060314 181500 X 3 14 2006 TUE 18 15 00 GMT 0 Note You must add 4 hours to the above HH field in order to enter the correct GMT time that corresponds to the current EST timezone with daylight savings time in affect add 5 hours if daylight sav ings time is not in currently active 5 This step describes how to configure the IP address networking parameters that will be used by the onboard Ethernet controller when executing under STAR and ASTRix To setup the networking parameters use the STAR config command again STARO gt config Enter cr until the Target IP address prompt appears This parameter should be set to the IP address of this client SBC This IP address should be the onboard Ethernet address of this client SBC that is already in the File Server SBC s etc hosts file Enter this IP address fol lowed by a cr The next config parameter Host IP address should be set to the IP address of the File Server for this client SBC This IP address should be the same IP address of the File Server that is located in the File Server s etc hosts file Enter the appropriate IP address followed by a cr The address contained in the Host IP address parameter will be the HOST environment variable that is used lat
12. sbin myprog See Custom Configuration Files on page 2 20 for more information about the memfs files add table and the S25client re script 2 5 4 2 Launching an Application for NFS Clients 2 30 Clients configured with NFS support mav either add application programs to the memfs root file svstem or thev mav access applications that reside on the File Server across NFS The advantage to using the memfs root file svstem is that the file can be accessed locallv on the client system rather than across the network The disadvantage is that there is only limited space in the memfs file system Furthermore this file system generally uses up physical memory on the client system When the client system is booted from an image stored in flash using the raw write method with the ASTRix fw flash write command this is not the case since the memfs file system remains in flash until the pages are accessed and brought into memory To add files to the memfs root file system follow the procedures for an embedded client above When adding files to the client s virtual root so that they can be accessed on the client via NFS the easiest method is to place the file s in one of the directories listed below This is because the client already has permission to access these directories and these directories are automatically NFS mounted during the client s system initialization Storage Path on File Server Access Path on the Client usr usr sbin
13. server The disk server hostname is the one specified by the disk clients when they are mounting the disk client s NFS partition s and the various disk clients hostnames are used on the disk server for the share command line in the etc d stab file 1 Enable the nfssrv kernel module for the client that has the local disk attached disk server The following commands should be executed on the File Server system of the disk server client cd etc clients disk server dir custom conf mkprivate kernel modlist add vi kernel modlist add Add the following nfssrv 2 For each partition to be shared add an entry similar to the example entry shown below Note that disk client 1 disk client refers to a list of nodes that want to share this partition See the d stab 4 manpage for more information disk server vrootpath is the path to the virtual root directory of the node with the local disk attached 4 vi disk server vrootpath etc dfs dfstab Example entry share F nfs o rw root disk client 1 disk client n d diskoso disk0s0 disk0sO B 1 Power Hawk Series 900 Diskless Systems Administrator s Guide B 2 3 Generate a new netboot image for the disk server client by executing the following command mknetbstrap ralledisk server On each disk client node that wants to share the disk partitions we need to generate a mount point directory for each partition to be mounted across NFS These partiti
14. the following binaries must be A 3 Power Hawk Series 900 Diskless Systems Administrator s Guide A 4 added to the memfs file system via the memfs files add custom table Example entries follow Hf sbin expr 0755 Hf usr bin devcfg 0755 Hf usr bin cut 0755 Hf sbin mknod 0755 Hf usr bin mkdir 0755 Hf Sbin fsck 0755 Hf etc s ufs fsck0755 Hf etc fs xfs mount0755 Hf etc fs ufs mount 0755 Hf sbin df 0755 Set FSTYPE and PARTITIONS FSTYPE ufs file system type ufs xfs PARTITIONS 0 1 2 3 4 5 6 disk partitions to be mounted etc mnttab Initialize gt gt etc vfstab disk 0 Create the device directories usr bin mkdir p dev dsk usr bin mkdir p dev rdsk dk db In this loop the device nodes are created based on major minor device information gathered from the call to devcfg The fsck 1m utility is executed for each file system a mount point directory is created and the file system is mounted and then verified using df usr bin devcfg m disk usr bin cut f3 while read majmin do maj echo majmin usr bin cut f1 d min echo majmin usr bin cut f2 d H Creates fsck and mounts the partition for i in SPARTITIONS do create the device nodes minor sbin expr min i echo gt Creating nodes dev dsk disk s i c echo and dev rdsk disk s i sbin mknod dev dsk di
15. the loading of the boot image cannot be separated from the initiation of execution within that image 1 To boot from the ethernet network the board s firmware ASTRix must be configured to boot from the network The boot sequence is initiated by resetting the board by cycling the power on the board or by manually issu ing the ASTRix command to execute a TFTP boot The manual ASTRix method is only available when a console terminal is connected to the disk less system 2 To boot from flash ROM the board s firmware ASTRix must be config ured to boot from flash through the use of a ASTRix startup script The boot sequence will be initiated whenever the board is reset power is cycled on the board or by manually executing the ASTRix startup script or flash boot commands The manual ASTRix method is only available when a console terminal is connected to the diskless system 1 1 3 Net Boot Toolset The Net Boot toolset is provided for creating the diskless configuration environment on the File Server and for creating boot images The diskless configuration environment includes the generation of the virtual root as well as the creation and modification of relevant system configuration files The virtual root serves as the environment for config uring a client s kernel building the boot image and as one of the partitions which is NFS mounted by an NFS client The Net Boot tool is executed on the File Server 1 3 Power Hawk Series 900
16. 5 If the corresponding package software was added on the File Server after the client s virtual root was created you must first bring the client s virtual root directory up to date by using the v option of netbootconfig 1M before adding subsystem support Example 1 Create netboot client wilma s configuration and also add support for the RCFBS subsystem netbootconfig C a RCFBS wilma Example 2 Remove support for the RCFBS subsystem from netboot clients wilma and fred netbootconfig U r RCFBS wilma fred Customizing the Basic Client Configuration This section contains information on the following major topics Modifying the Kernel Configuration page 2 18 Custom Configuration Files page 2 20 Modifying the Client Profile Parameters page 2 28 Launching Applications page 2 29 Launching an Application Embedded Clients page 2 29 Launching an Application NFS clients page 2 30 2 5 1 Modifying the Kernel Configuration 2 18 A diskless client s kernel configuration directory is resident on the File Server and is a part of the client s virtual root partition Initially it is a copy of the File Server s etc conf directory The kernel object modules are symbolically linked to the File Server s kernel object modules to conserve disk space By default a client s kernel is configured with a minimum set of drivers to support the chosen client configuration The set of drivers configured by
17. Glossary 1 10base T Glossary 1 A ARP Glossary 1 ASTRix description 1 3 1 5 Glossary 10 Flash access 3 2 manual iv netboot 2 34 SBC client board configuration 2 3 banked flash 3 2 block device Glossary 1 driver Glossary 2 boot device Glossary 2 Boot Flash 3 1 3 2 boot image characteristics 1 2 1 5 1 7 creation 1 7 2 32 dependencies 2 32 description 2 31 file definition Glossary 2 file system 1 2 boot status 2 34 booting Index basics 1 2 Flash file system 3 3 Flash Raw Write 3 5 netboot 2 31 procedure 1 9 C client configuration 2 13 2 16 customize 2 18 kernel 2 18 types 1 1 definition 1 4 Glossary 3 embedded see embedded client NFS see NFS client profile 2 13 2 28 SBC 2 3 shutdown 2 35 config utility 2 19 configuration clients netbootconfig 2 16 custom files 2 20 diskless systems 1 17 2 1 kernel 2 18 crash 4 1 4 2 custom configuration files 2 20 customize basic client configuration 2 18 D debugging 4 1 definitions of terms 1 4 Glossary 1 dfstab 1 14 dfstab diskless 1 14 disk space requirements 1 18 disk add a local A 1 disk clients B 1 disk server B 1 diskless boot basics 1 2 Index 1 Power Hawk Series 900 Diskless System Administrator s Guide implementation 1 7 operation description 1 1 embedded client configuration 2 14 definition 1 4 launching an application 2 29 overview 1 1 file server 1 3 1 4 Glossary 5 Flash autobooting Glossary 4 boot see Fla
18. client s profile file may be created by invoking the mkswap command usr etc diskless d sys conf bin d mkswap vrootpath lt megabytes gt NOTE As previously mentioned the client should be first shutdown before issuing the mkswap command if the NFS client is cur rently up and running 2 5 4 Launching Applications 2 5 4 1 Following are descriptions on how to launch applications for Embedded Clients NFS Clients Launching an Application for Embedded Clients For diskless embedded clients all the application programs and files referenced must be added to the memfs root file system via the memfs files add file See section memfs files add Table on page 2 25 for more information on adding files via the memfs files add file As an example the command name myprog resides on the File Server under the path home myname bin myprog We wish to automatically have this command executed from the path sbin myprog when the client boots This commands reads from a data 2 29 Power Hawk Series 900 Diskless Systems Administrator s Guide file expected to be under myprog data This data file is stored on the File Server under home myname data myprog data The following entries are added to the mem s files add table f sbin myprog 0755 home myname bin myprog f myprog data 0444 home myname data myprog data The following entry is added to the client s start up script d Client s start up script
19. ethernet connection When the client is not on the same local Ethernet network the system administrator may setup a routing table entry that will provide access to a remote file server through a local gateway Whether manually booting or autobooting a client the client SBC must first be set up with the information it needs to do Net Booting This is accomplished with the STAR config command and possibly with an additional ASTRix route command or the appropriate commands in an ASTRix startup script Once this setup is done the client should not need to be reconfigured unless one or more of these parameters change since the STAR config command information and ASTRix startup script is saved in NVRAM and is pre served across reboots and power cycles The following information is required by the STAR config command in order to config ure a client that will be performing a net boot Target IP address The local client SBC s IP address which ASTRix will use as the return address for TFTP data transfer For NFS clients this is the Ethernet IP Address of this client SBC Host IP address The IP address for the File Server which should already be defined in the etc hosts file You may also Net Boot an Embedded Client Since this client kernel does not support net working no IP address has yet been defined for it In this situation select a unique IP address to use An address should be selected that decodes to the same local subnet tha
20. file system cannot be corrupted by unexpected system crashes which might leave a writable file system in an inconsistent state The boot image that is burned into flash is the same boot image that is downloaded via an Ethernet network connection when a developer is using the Net Boot method As such a developer may use the Net boot method when they are actively modifying the boot image during the development phase and then they may burn the final boot image into flash when they enter into the deployed phase of the application There are no tools specifically targeted towards creating boot images for Flash Booting Instead the standard loosely coupled configuration tools are used for building the image Since the File Server is only connected to the client system via an ethernet connection the flash must be burned via a process known as network loading Preparation for loading the boot image into flash is the same method used in the Net Boot method The ASTRix tftp command is used to load the boot image into memory without executing that image The boot image is then burned into flash from memory by using either the ASTRix ffsw flash filesystem write command or the ASTRix fw flash write command Once the image is burned into flash then all subsequent booting can be automatically ini tiated at reset or power cycle time by setting up an ASTRix startup script that will execute the appropriate boot command that will boot the bootable im
21. more important than overall system performance The method of transmitting data from a terminal to a user without processing This mode is defined in the line discipline modules Remote copy allows files to be copied from or to remote systems rcp is often compared to ftp The half of a STREAMS module or driver that passes messages upstream Remote login provides interactive access to remote hosts Its function is similar to telnet A set of instructions that perform a specific task for a program Driver code consists of entry point routines and subordinate routines Subordinate routines are called by driver entry point routines The entry point routines are accessed through system tables Remote shell passes a command to a remote host for execution Single Board Computer SCSI driver interface SDI sequential I O SLIP A collection of machine independent input output controls functions and data structures that provide a standard interface for writing Small Computer System Interface SCSI drivers T O operations to the same file descriptor that specify that the I O should begin at the cur rent file offset Serial Line IP The SLIP protocol defines a simple mechanism for framing datagrams for transmission across serial line Glossary 9 Power Hawk Series 900 Diskless Systems Administrator s Guide Server See definition for File Server and host STAR ASTRix A pair of board resident Flash monitor utilitie
22. need to modify just a few lines of the already existing template If you wish to use the startup script template file then you need to first download the template version of the startup script from the file server SBC On the file server SBC issue the following command as root netbootconfig P astrix 2 tftpboot astrix Then on the client SBC console terminal issue the following ASTRix com mands cd cr tftp g r astrix HOST cr cp astrix nvram startup cr cd nvram cr vi startup cr At this point simply modify the CLIENT line using vi so that CLIENT is set to the hostname of this client For example if wilma is the hostname of this client then change this line to CLIENT wilma The only other 2 lines that MAY need to be uncommented and modified are the routing entry lines Only uncomment and setup these 2 lines if the client accesses the file server SBC through a gateway For example if the file server s network is 129 148 42 0 and it is accessed locally by the client through the gateway 129 148 43 196 with the client residing on the 129 148 43 0 network then the routing lines in the startup script should be uncommented and modified to be Power Hawk Series 900 Diskless Systems Administrator s Guide 2 10 route add net 129 148 42 0 gw 129 148 43 196 netmask 255 255 255 0 dev ethO To instead create the startup file from scratch instead of using the template file above is
23. or deconfigured using options to mknetbstrap 1M 3 kernel configuration can be done using the config 1M utility and then rebuilding the unix kernel 4 The idtuneobj 1M utility may be used to directly modify certain kernel tunables in the specified unix kernel without having to rebuild the unix kernel kernel modlist add The kernel modlist add custom table is used by the boot image creating tool mknetbstrap 1M for adding user defined extensions to the standard kernel configuration of a client system When mknetbstrap 1M is run it compares the modification date of this file with that of the unix kernel If mknetbstrap 1M finds the file to be newer than the unix kernel it will automatically configure the modules listed in the file and rebuild a new kernel and boot image This file may be used to change the kernel configuration of one client or all the clients For more information about this table see Custom Configuration Files on page 2 20 2 5 1 2 mknetbstrap Kernel modules may be configured or deconfigured via the k option of mknetbstrap 1M A new kernel and boot image is then automatically created For more information about mknetbstrap 1M see the online manual page 2 5 1 3 config utility The config 1M tool may be used to modify a client s kernel environment It can be used to enable additional kernel modules configure adapter modules modify kernel tunables or build a kernel You must use the
24. r option to specify the root of the client s kernel configuration directory If you do not specify the r option you will modify the File Server s kernel configuration instead of the client s For example if the virtual root directory for client rosie was created under vroots rosie then invoke config 1M as follows Power Hawk Series 900 Diskless Systems Administrator s Guide 2 5 1 4 idtuneobj config r vroots rosie After making changes using config 1M a new kernel and boot image must be built There are two ways to build a new boot image 1 Use the Rebuild Static menu from within config 1M to build a new unix kernel and then invoke mknetbstrap 1M mknetbstrap 1M will find the boot image out of date compared to the newly built unix file and will automatically build a new boot image 2 Use mknetbstrap 1M and specify unix on the rebuild option r In situations where only kernel tunables need to be modified for an already built host and or client kernel s it is possible to directly modify certain kernel tunable values in a client and or host unix object files without the need for rebuilding the kernel The idtuneobj 1M utility may be used to directly modify certain kernel tunables in the specified unix or Dynamically Linked Module DLM object files The tunables that idtuneobj 1M supports are contained in the usr lib idtuneobj tune database file and can be listed using the 1 option of idtuneobj 1M
25. root and is accessed via NFS from the client system Therefore the boot image does not need to be rebuilt after modifying the initab table and the changes to this table will take effect the next time the system is booted or shutdown Like the other customization files these tables may be updated in one of the two subdirectories nfs or emb Changes made under the usr etc diskless d cluster conf custom conf directory apply globally to all nfs or embedded clients that share this File Server If the changes are specific to a particular client then a private copy of the shared file should first be created in that client s private customization directory by using the mkprivate tool and then edited in that client s custom conf directory 2 5 2 3 vfstab Table 2 24 The vfstab table defines attributes for each mounted file system The v stab table applies only to NFS client configurations The v stab 4 file is processed when the mountall 1M command is executed during system initialization to run level 3 See the v stab 4 online manual page for rules on modifying this table Configuring an NFS client configuration causes this table to be installed with entries needed for basic diskless system operation and these entries should not be removed or modified The v stab table is part of the client s virtual root and is accessed via NFS The boot image does not need to be rebuilt after modifying the v stab table the changes will take effe
26. sbin opt opt virtual root etc etc virtual root var var virtual root users users As an example the command name myprog was created under the path home myname bin myprog To have this command be accessible to all the diskless clients on the File Server we could mv 1 or 1 the command to the sbin directory mv home myname bin myprog sbin myprog Netboot System Administration If only one client needs access to the command it could be moved or copied to the etc directory in that client s virtual root directory mv home myname bin myprog virtual root etc myprog To access an application that resides in directories other than those mentioned above the File Server s directory must be made accessible to the client by adding it to the dfstab 4 table and then executing the share 1M or shareall 1M command on the File Server To automatically have the directories mounted during the client s system start up an entry must be added to the client s v stab file See Custom Configuration Files on page 2 20 for more information about editing the v stab file 2 6 Booting and Shutdown This section describes the following major topics The Boot Image page 2 31 Creating the Boot Image page 2 32 Net Booting page 2 33 Verifying Boot Status page 2 34 Shutting Down the Client page 2 35 2 6 1 The Boot Image The boot image is the file that is loaded from the File Server to a diskless clie
27. the user should not modify this setting 2 4 1 2 Required NFS Related Parameters 2 14 The following netboot client profile parameters are required only if the client is a NFS client S YS CONFIG nfs The values in these parameters are ignored if the client is an embedded client SYS CONFIG emb AUTOBOOT This parameter specifies whether this client should be shutdown whenever the File Server is shutdown The value for this parameter should be either y If set to y then the client will be shutdown by the File Server when ever the File Server is shutting itself down Example AUTOBOOT y NOTE When this parameter is set to y then a hidden file named autoboot will be created by netbootconfig 1M under this client s virtual root directory the VROOT parameter path This file will serve to indicate that the client SBC should be automatically shutdown by the File Server SBC whenever the File Server is shutdown This autoboot file may be manually removed or created in the client s virtual root directory as needed Netboot System Administration SWAP SIZE This parameter is the size in megabytes of remote swap space Swap space is implemented as a file named virtual root dev swap file which resides on the File Server in the client s virtual root directory and which is accessed over NFS The recommended value for this parameter is 1 5 times the size of the amount of physical memory located o
28. to user pages that cannot be swapped and freed if and when the number of free real memory pages becomes low If these daemons did not maintain an adequate number of free memory pages for page allocations then applications might become blocked forever in the kernel waiting for their page allocation requests or internal kernel memory page allocation requests to complete There are a number of possible swap space configurations on client systems 1 no swap space typically this would be a client configured in Embedded mode 2 remote swap space client would be configured as a NFS diskless system with the swap space accessed through the NFS subsystem 3 local disk swap space client configured in either NFS or Embedded mode client configured to use a local disk for swap space When there is no swap space or a small amount of swap space it may be necessary to modify the default values of certain system tunables in order to maximize system performance and user virtual space capacities The following are some of the system tunables that are relevant to system swap space management in a system with little or no secondary storage swap space 1 15 Power Hawk Series 900 Diskless Systems Administrator s Guide 1 16 1 Systems with no swap space should be tuned such that process swapping does not become aggressively active before process growth is limited by virtual swap reservations as this will impact system performance without provid
29. 0 Diskless Systems Administrator s Guide NFS client In an NFS client configuration the File Server provides UNIX file systems for the client system A client system operates as an NFS client of the File Server This configuration allows substantially more file system space to be available to the client sys tem for storing an application and application data than an embedded configuration The LCS configuration described may contain a mix of Power Hawk Series 700 and Power Hawk Series 900 systems The file server in this LCS configuration may be either a Series 700 or Series 900 system FILE SERVER HOST SBC FAST WIDE SCSI 2 SYSTEM DISK gt RS 232 PORT m svsren CONSOLE Sa z a m Css SBC SBC D gt ie Dunk N NETBOOT CLIENTS Figure 1 1 Loosely Coupled System Configuration 1 1 2 Diskless Boot Basics The first step in creating a diskless system is to create a boot image which contains both the operating system and a file system that contains at a minimum the executable needed to boot the PowerMAX OS This file system which is bundled into the boot image can Introduction also be used to store application programs and data UNIX commands and libraries or any other file that might live in a disk based partition The size of this file system is limited since it must either be copied into memory or must reside in flash ROM The File Server i
30. 5 2 6 vrootfiles add Table 1 26 2 5 3 Modifying the Client Profile Parameters 1 28 2 5 4 Launching 1 29 2 5 4 1 Launching an Application for Embedded Clients 1 29 2 5 4 2 Launching an Application for NFS Clients 1 30 Booting and Shutdown lesse eee 1 31 2 6 1 The Boot Image ene Rte eed pn 1 31 2 6 2 Creating the Boot 1 32 2 6 2 1 Examples on Creating the Boot Image 1 33 2 6 3 Net Booting NIE 1 33 2 6 3 1 NetbootUsing 1 34 2 64 Shutting Down the 1 35 Power Hawk Series 900 Diskless System Administrator s Guide 2 Netboot System Administration 2 1 Configuration Overview This is a overview of the steps that must be followed in configuring a loosely coupled configuration Some of these steps are described in more detail in the sections that follow A loosely coupled system consists of a File Server and one or more diskless clients which download their private boot image which resides on the File Server A loosely coupled system uses an ethernet network connection between each diskless client and the File Server for communication There is no sharing of a VME bus in this configuration The following instructions assume th
31. 6 3 Synergy Manual Name Document Number Raptor DX VMEbus Dual G4 Dual PMC amp StarFabric PowerPC Single Board Computer User Guide 02 0426 UG V YFD lt REV gt MantaQX3 VAFQ VMEbus Quad G4 Single PMC amp StarFabric PowerPC Single Board Computer User Manual 815138 Version VER Synergy Microsystem STAR User Guide 03 0072 UG STAR lt REV gt Synergy Microsystem ASTRix User s Guide 03 0076 UG ASTX lt REV gt Related Specifications Title Pubs No IEEE Common Mezzanine Card Specification CMC P1386 Draft 2 0 IEEE PCI Mezzanine Card Specification CMC P1386 1 Draft 2 0 Compact PCI Specification CPCI Rev 2 1 Dated 9 2 97 a lote ette rais deett alo a aps tebe ea tA Chapter 1 Introduction Overview Aceh gree tee Loosely Coupled Systems _ 5 Diskless Boot Basics Net Boot Toolset rss aea A D finitions cle eR a nESDS PETS ERE ets Hardware Model 920 Hardware Feature Summary Model 940 Hardware Feature Summary Diskless 1 Virtual Root secs pe Boot Image Creation and Characteristics MEMFS Root Booting i undue err e ps Net Boot
32. Clients Using netbootconfig on page 2 16 for more infor mation On the File Server system execute mknetbstrap 1M to create the boot images of each diskless client See the Booting and Shutdown section on page 2 31 section for more information Oneach client connect a console terminal and power up the netboot client Use STAR to set the hardware clock and networking configuration and optionally use ASTRix to configure an ASTRix netboot startup script Use ASTRix to set the hardware clock to setup the ASTRix networking con figuration and to optionally configure a ASTRix netboot startup script See SBC Client Board Configuration on page 2 3 for more details Oneach client either manually boot into ASTRix from STAR if STAR is not configured to automatically boot ASTRix and then manually execute ASTRix fftp and boot commands to boot the client or reset or power cycle the client board if the client is configured to auto matically start up ASTRix from STAR and to execute an ASTRix tftp boot startup script See Net Booting for more details on booting up the client On the File Server system customize the client s virtual root configuration as needed and then run mknetbstrap 1M to process the changes If a new boot image is created as a result of the changes shutdown the client and then reboot it See Booting and Shutdown on page 2 31 for more information Netboot System Administration 2 1 2
33. Diskless Systems Administrator s Guide The Net Boot toolset consists of the tools netbootconfig and mknetbstrap These tools handle loosely coupled clients that boot via an ethernet network or from flash ROM where no common I O bus based communication will be utilized on the client system The netbootconfig tool is used to create the diskless configuration environment for a diskless client The mknetbstrap tool is used for creating a diskless client s boot image More information is provided on the Net Boot tool in Chapter 2 Netboot System Administration 1 2 Definitions Loosely Coupled System A Loosely Coupled System LCS is a network of Single Board LCS Computers SBCs One of the SBCs must have a system disk and is referred to as the File Server and all other SBCs are generally referred to as clients An ethernet connection between the File Server and the client systems provides the means for inter board communication File Server The File Server has special significance in a Loosely Coupled system as it is the only system with physically attached disk s that contain file systems and directories essential to running the PowerMAX OS etc sbin usr var tmp and dev The File Server boots from a locally attached SCSI disk and provides disk storage space for configuration and system files for all clients There is only one File Server in a Loosely coupled system clients depend on the File Server for booting s
34. Examine the client system locally on the client system using crash When kdb is configured into a client s kernel the k sequence will cause the system to drop into kdb The sequences b i and h all cause an immediate reset of the board where control is subsequently returned to the STAR ASTRix firmware The kdb package is provided with the kernel base package A client kernel may be configured with kdb and its boot image may be configured to enter kdb early in the boot process A console terminal must be connected to the client board to interact with kdb On client boards the console debugger support is not present However if system level debugging on a client is desired then it is possible to use kdb To use kdb the kdb and kdb util kernel drivers must be configured into the client s unix kernel Except for the kdb consdebug and star or smon commands kdb operates without any differences from a normal system when executing on a client The consdebug command is not available on netboot clients and the star or smon command will reset the board and return control to the STAR ASTRix firmware Power Hawk Series 900 Diskless Systems Administrator s Guide 4 3 crash By default the client kernel configuration is not built with kdb support The kdb kernel modules may be configured into the client s kernel via the k option to the mknetbstrap command gt mknetbstrap kkdb kkdb util The client system may also be programmed to sto
35. Hawk Model 920 features l 6 Power Hawk Model 940 features 1 6 Power Hawk Networking Structure l 13 PPP Glossarv 8 processor limits l 19 profiles 2 13 2 28 R random I O Glossary 8 raw I O Glossary 9 Raw Write 3 2 3 5 rcp Glossary 9 referenced publications iv remote file sharing description 1 12 Glossary 8 networking structure 1 13 rlogin Glossary 9 root virtual 1 3 1 7 rsh Glossary 9 S S25client rc script 2 21 2 23 SBC client configuration 2 3 sequential I O Glossary 9 Serial Port A 1 17 shared files 1 12 shutdown 2 35 SMTP Glossary 10 SNMP Glossary 10 Source Code Control System SCCS Glossary 10 STAR iv 1 5 1 10 2 3 subsystem support 2 17 swap space 1 5 1 15 2 15 Glossary 11 Index Synergy Monitor STAR ASTRix 1 5 syntax notation iii system console 1 17 system disk 1 5 system run level 1 5 Glossary 11 tables dfstab 1 14 dfstab diskless 1 14 hosts 2 15 inittab 2 23 kernel modlist add 2 19 2 25 memfs files add 2 25 memfs inittab 2 23 vfstab 2 24 vroot files add 2 26 TCP Glossary 11 TELNET Glossary 11 Trivial File Transfer Protocol TFTP 1 5 Glossary 11 U UDP Glossary 12 unix kernel 1 7 1 8 2 18 unix bstrap 1 7 1 8 User Flash 3 1 3 2 user limits 1 19 V vfstab 2 21 2 24 virtual root 1 3 1 7 virtual swap 1 15 vroot files add 2 21 2 26 Index 3 Power Hawk Series 900 Diskless System Administrator s Guide Index 4 Spine for 1 Binder Product Nam
36. Installing Additional Boards To add additional boards after the initial configuration follow steps 2 through 8 described above 2 2 SBC Client Board Configuration This section describes the procedure for configuring a SBC board as a client SBC a loosely coupled system The user should also refer to the Synergy Microsystems ASTRix User Guide and the Synergy Microsystems S T A R User Guide for additional information regarding STAR and ASTRix commands and features The following steps should be followed in order to set up a board as a client SBC 1 STARO gt config Connect a terminal to Serial UART Port A Console if one is not already connected The next step is to update some of the STAR configuration parameters Power on or reset the board and watch for the resulting input prompt If the STARO gt prompt appears then skip the rest of this step and go to step 3 If STAR is automatically setup to boot ASTRix then following prompt will eventually appear on your screen without any user intervention To return back to STAR from ASTRix enter the following command at ASTRix prompt Star and then enter cr The system will reboot itself into STAR without start ing up ASTRix At this point you should be running STAR and sitting at the STARO gt prompt This step describes how to enable the automatic startup of ASTRix by STAR To enable the automatic startup of ASTRix use the STAR config comman
37. Power Hawk Series 900 Diskless Systems Administrator s Guide N 0891090 030 lt January 2006 Copyright 2006 by Concurrent Computer Corporation All rights reserved This publication or any part thereof is intended for use with Concurrent Computer Corporation products by Concurrent Computer Corporation personnel customers and end users It may not be reproduced in any form without the written permission of the publisher The information contained in this document is believed to be correct at the time of publication It is subject to change without notice Concurrent Computer Corporation makes no warranties expressed or implied concerning the information contained in this document To report an error or comment on a specific portion of the manual photocopy the page in question and mark the correction or comment on the copy Mail the copy and any additional comments to Concurrent Computer Corpora tion 2881 Gateway Drive Pompano Beach FL 33069 Mark the envelope Attention Publications Department This publication may not be reproduced for any other reason in any form without written permission of the publisher UNIX is a registered trademark of the Open Group Ethernet is a trademark of Xerox Corporation PowerMAX OS is a registered trademark of Concurrent Computer Corporation Power Hawk is a trademark of Concurrent Computer Corporation Other products mentioned in this document are trademarks registered trademarks or trade nam
38. STRix boot command configuration slot for boot ing the client netboot image This step is required for both manual and automatic booting To setup an ASTRix boot command configuration slot number that may always be used for booting the client netboot image boot 5 n lt client gt bstrap s c The command above will setup the booting of the client s netboot image where client should be replaced by the actual hostname of the client For Netboot System Administration example if this client s hostname is wilma then the following boot com mand line to setup configuration slot number 5 would be boot 5 n wilma bstrap s c The s option sets the configuration without actually executing the boot operation and the c option clears the command line which not used by PowerMAX OS client netboot images NOTE The hostname of this loosely coupled client MUST be equal to the name of the client profile file that is located on the File Server SBC in the etc pro files directory This step will configure the board for either manual or automatic down loading and booting of the client If you wish to always manually boot the client then go to b below a To have ASTRix automatically attempt to boot the client the nvram startup script must be installed To create a new netboot startup script you may either manually type in the entire startup script or you may start with a startup script file template so that you only
39. The server SBC must be accessible from all client SBCs via an Ethernet LAN For Power Hawk Model 920 systems the on board Galileo GT64260 Ethernet controller should be used for this connec tion for Power Hawk Model 940 systems the on board Marvell MV 64460 Ethernet controller should be used For Power Hawk Series 700 systems the on board Symbios Ethernet controller Sym bios SYM53C8585 should be used for this connection Power Hawk Series 900 Diskless Systems Administrator s Guide 1 5 2 Disk Space Requirements The table below details the amount of available disk space required per client single board computer for the virtual root partition These values are for the default shipped configuration Added applications may increase disk space requirements Values in this table do not include swap space for the diskless system The amount of swap space is con figurable but should be at least one and one half times the size of physical memory on the single board computer A client s virtual root directory can be generated in any disk partition on the File Server The root and var file systems are not recommended for use as client virtual partitions Client Configuration Disk Space NFS 25 Megabytes Embedded 15 Megabytes 1 5 3 Software Prerequisites 1 18 The following software packages must be installed on the host system prior to installing the diskless package prerequisite packages listed al
40. age that is stored in flash Refer to Flash Boot System Administration chapter for more details Power Hawk Series 900 Diskless Systems Administrator s Guide 1 4 5 Remote File Sharing 1 12 Clients configured as Embedded Clients must have in their memory based root file system all the files needed for booting and all the files needed to run applications This is because Embedded Clients do not have networking by definition and therefore will not have access to remote files on the File Server The memfs root file system of a client configured with NFS need only contain the files required for booting When the client system reaches init state 3 it is able to NFS mount and access the File Server s directories The NFS mounts are executed from a start up script in etc rc3 d Two different inittab files are used in booting an NFS configuration When the etc directory in a client s virtual root is NFS mounted the original inittab file is overlaid with the one in the virtual root The directory etc rc3 d is then re scanned to execute start up scripts in the virtual root The directories usr sbin and opt are completely shared with the server while etc and var are shared on a file by file basis Listed below is the NFS mount scheme in use Path on File Server Mount Point on Client usr usr sbin sbin opt opt virtual rootpath etc etc virtual rootpath var var virtual rootpath dev dev virtual root
41. ap 1M to limit the number of client boot images which are simultaneously processed Netboot System Administration 2 6 2 1 Examples on Creating the Boot Image Example 1 Update the boot image of all the clients configured with netboot client profile files in the etc profiles directory Limit the number of clients processed in parallel to 2 mknetbstrap p2 all Example 2 Update the boot image of clients wilma and fred Force the rebuild of the unix kernels and configure the boot images to stop in kdb early during system initialization mknetbstrap r unix b kdb wilma fred Example 3 Update the boot image of all the clients configured with netboot client profile files in the etc profiles directory Rebuild their unix kernel with the kdb module configured and the rtc kernel module deconfigured Limit the number of clients processed in parallel to 3 mknetbstrap p3 k kdb k rtc all 2 6 3 Net Booting Netboot diskless clients boot from an image downloaded via an ethernet network connection Net booting also referred to as Network booting is performed by the ASTRix ROM based firmware using the TFTP Trivial File Transfer Protocol RFC783 network protocol netboot diskless clients depend on the File Server for the creation and storage of the boot image Once booted netboot clients configured with NFS support continue to rely on the File Server for accessing their system files via NFS Clients configured as e
42. at all the prerequisite hardware has been installed and each netboot client s on board Ethernet controller is attached to a subnet on which the File Server system may be accessed either directly on the same subnet or through a gateway For details see Looselv Coupled System Hardware Prerequisites on page 1 17 2 1 1 Installing a Loosely Coupled System Follow these steps to configure a loosely coupled system 1 Install the File Server with the prerequisite software packages the diskless package and all patches Refer to the Software Prerequisites on page 1 18 and the applicable system release notes for more information On the File Server system configure and mount file system s other than root or var that can be used to store the virtual root directories for each client If not already present an entry for this file system must be added to etc v stab 4 An existing file system can be used but there must be sufficient file space to hold the client virtual root files The boot images for each netboot client are placed into the tftpboot directory by mknetbstrap 1M Since these images tend to be large in size the system administrator may want to mount a sufficiently large file system at the tftpboot mount point if the File Server s root partition is not able to accommodate the projected client boot image disk space requirements See the Disk Space Requirements section on page 1 18 for details of the amou
43. ator s Guide 2 4 2 Configuring Clients Using netbootconfig The netbootconfig 1M toolis used to create remove or update one or more diskless client configurations For more details on running this tool see the manual page available online Netbootconfig 1M gathers information from the client profile files and stores this information in a ksh loadable file named client profile under the client s virtual root directory The client profile is used by netbootconfig 1M by other configuration tools and by the client initialization process during system startup It is accessible on the client from the path client profile Netbootconfig 1M appends a process progress report and run time errors to the client private log file etc profiles client hostname 1log on the File Server or if invoked with the t option to stdout With each invocation of the tool an option stating the mode of execution must be specified The modes are create client C remove client R and update client U 2 4 2 1 Creating and Removing a Client Configuration 2 16 When creating new client configurations the client profile parameters must already be set up in the etc profiles client profile files see The Client Profile File on page 2 13 before using netbootconfig 1M to create the new client configurations The etc hosts file should also already contain the appropriate entries for the new netboot clients see Hosts Tables on
44. automatic rebuild of one or more of the boot image components In addition updates to user configurable files also affect the build of the boot image A list of the user configurable files and the boot image component that is affected when that file is modified are shown in Table 2 1 These files are explained in detail under section Customizing the Basic Client Configuration on page 2 18 Table 2 1 Boot Image Dependencies Boot Image Component User Configurable File unix kernel kernel modlist add memfs cpio memfs files add memfs inittab K00client embedded client configurations only KS25client embedded client configurations only The boot image components are created under etc con cf din the client s virtual root directory The boot image itself is installed into the tftpboot directory under the name of client profile filename gt bstrap For example a netboot client with a client profile filename of target1 would have a boot image installed with a pathname of tftpboot targetl bstrap 2 6 2 Creating the Boot Image 2 32 The mknetbstrap 1M tool is used to build the boot image This tool gathers information about the client s from each client s client profile file located in the etc profiles directory Some example uses follow Building a boot image is resource intensive When creating the boot image of multiple clients in the same call use the p option of mknetbstr
45. comment out or leave these lines out of the startup file Netboot System Administration The tftp while loop that downloads the client s netboot image serves two purposes The first purpose of the loop is to continually retry the tftp com mand should any errors occur during the transfer while the file server SBC is up and running The second purpose is to accommodate situations when the client and file server SBCs are in the same rack and it is desirable to have the netboot clients boot themselves up after a power cycle sequence without the need for any additional manual resets In order to ensure that a client can successfully download its boot image via tftp the file server sys tem MUST be completely booted into run level 3 before attempting to net boot any clients Thus the tftp retry loop gives the client the ability to actively wait for the file server to boot up and enter run level 3 The execution of the startup script may be manually aborted by entering lt Ctrl gt c at the console terminal should there be a need to do so Once aborted the startup script may be manually started by entering nvram startup cr The startup script file uses boot configuration number 5 to reset the board and begin execution of the client s netboot image therefore boot configu ration number 5 MUST have been previously setup in step 9 above At this point the ASTRix netboot client configuration setup has been com pleted Once th
46. ct the next time the system is booted or shutdown Like other NSF only customization files these tables may be updated in the client shared nfs subdirectory Changes made under the usr etc diskless d cluster conf custom conf nfs directory apply globally to all NFS clients that share this File Server If the changes are specific to a particular client then a private copy of the shared file should first be created in that client s private Netboot System Administration customization directory by using the mkprivate tool and then edited in that client s custom conf directory 2 5 2 4 kernel modlist add Table New kernel object modules may be added to the basic kernel configuration using the kernel modlist add file One module per line should be specified in this file The specified module name must have a corresponding system file installed under the virtual rootpath etc conf sdevice d directory For more information about changing the basic kernel Configuration see Modifving the Kernel Configuration on page 2 18 Changes to this file are processed the next time mknetbstrap 1M is invoked causing the kernel and the boot image to be rebuilt When modules are specified that are currently not configured into the kernel per the module s System 4 file those modules will be enabled and a new unix and boot image will be created If mknetbstrap 1M finds that the modules are already configured the request will be ignored A client mus
47. d An example of the entire output of the config command is shown below Enter a new value or a return to skip to back up or to exit BOOT Configs reset when defaults set Port A baudrate 9600 Port B baudrate 9600 Share console port Y Power Hawk Series 900 Diskless Systems Administrator s Guide 2 4 PCI Enum powerup delay milliseconds Enable quiet boot mode Enable ASTRix autoboot Enable post script Skip PCI Enumeration Skip PMC Configuration L1 Miss Queue Depth 1 3 6 L3 Size L3 1 2 MB Enable Built in self tests 1 at powerup only 22 at all hard resets at all resets 0 NMoUogzzruiao Il USER Configs unchanged when defaults are set Board serial number 0x1301892 Target IP address 129 134 32 80 Host IP address 129 134 32 81 Print Boot Banner y Init mem EDC N Legacy Mode Input N Enable ChangeLog Message Xu PCI 0 Start Addr ex 0x80000000 0 80000000 PCI 0 Size ex 0x40000000 1GB 0x30000000 PCI 1 Start Addr ex OxB0000000 OxB0000000 PCI 1 Size ex 0x10000000 256MB 0x40000000 Hit Enter to finish STARO gt 3 Hit the lt cr gt key until the Enable ASTRix autoboot prompt appears and then enter Y lt cr gt Then continue entering cr until the STARO gt prompt reappears NOTE Once the autoboot ASTRix parameter is enabled it is still possi ble to get back into STAR without immediately go
48. d is recommended for final deployed environments While actively testing and modifying the application and boot image the Net boot method may be used via an ethernet connec tion The Flash boot method can then be used to burn the final version of the boot image into flash when the application is deployed Power Hawk Series 900 Diskless Systems Administrator s Guide 1 4 4 1 Net Boot Net Boot is a method of loading a kernel image into a client s SDRAM over an Ethernet connection and then initiating execution of that image Since the file server and the client are only connected via Ethernet the file server cannot actively force a client to accept and boot an image rather the client must initiate the transfer from the fileserver and cooperate with the file server to complete the transfer This client initiation takes one of two forms manually entering and executing ASTRix commands at the client s attached console terminal or automatic execution of the ASTRix commands by the client SBC when ever it is powered up or reset through the use of an ASTRix startup script Net booting is performed by ASTRix using the TFTP Trivial File Transfer Protocol RFC783 This is a standard protocol that is supported by PowerMAX OS For additional information pertaining to TFTP refer to the Network Administration manual 0890432 Any client SBC can be Net Booted as long as the client SBC has access to the file server through the client s
49. d sets up its execution environment when the boot image is executed on the client a copy of the compressed cpio image from memfs cpio and a bootstrap record used to communicate information about the client to the kernel and its bootstrap At the time of booting boot files are created as needed based on dependencies established by the makefile bstrap makefile under the usr etc disk less d svs conf bin d directory see table below 1 7 Power Hawk Series 900 Diskless Systems Administrator s Guide Boot File Description Dependencies unix unix kernel kernel modlist add memfs cpio cpio image of all files to be loaded in the unix client s memory based root file system memfs files add system configuration files unix bstrap bootstrap image unix memfs cpio 1 4 3 MEMFS Root Filesystem 1 8 A memory based filesystem called the memfs filesystem becomes the root filesystem of a client as part of its booting process As the client completes its boot it may mount other file systems that are available to it perhaps those on local disks or from across the net work These other file systems do not replace the original memfs root filesystem but instead augment it with their extra directories and files Files needed by diskless applications can be located either in the memfs root filesystem of a client or on the File Server in the client s virtual root directory For embedded systems all us
50. d will read the boot image from the file server and download it into the client s SDRAM Step 2 This step applies only to the Flash boot method After downloading the boot image to SDRAM the image may be burned into flash This is accom plished manually at the client s attached console terminal using the ASTRix ffsw Flash File System Write command or the ASTRix fw flash write command Once burned into flash the boot image no longer needs to be downloaded across the network int SDRAM step 1 is no longer required Step 3 The final step of the boot operation is to execute the downloaded boot image This step is accomplished with the ASTRix boot command The ASTRix boot may be initiated automatically within an ASTRix startup script or manually at the client s attached console terminal The ASTRix boot command allows for booting images from various devices including booting images directly from memory for the Net boot method and boot ing images from flash the Flash boot method The entire Net boot method may be set up so that the sequence is automatically initiated after any reset or power up cycle by using the ASTRix startup script In addition once step 2 Flash burn has been accomplished for the Flash boot method an ASTRix startup script may be used to automatically load and boot the flash boot image after any reset or power up cycle Booting from flash is significantly faster and more reliable than the Net boot method an
51. de L3 cache per CPU 256 MB 1 GB onboard memory 64 MB Boot User Flash 128 MB of NVRAM battery backed Two mailboxes Hi amp Lo per CPU Real time clock calendar 4 digit year Two asynchronous RS 232 423 serial ports up to 115 Kbps Fast Ethernet 10Base T 100Base TX 3 ports Programmable 4 bit TTL compatible general purpose I O GPIO Dual 64 bit PMC slot Model 940 Hardware Feature Summary SMP compliant PCI to VME64 bridge rated at 50 MB sec Dual G4 PowerPC 7457 2 MB backside L3 cache per CPU 256 MB 1 GB onboard memory 64 MB Boot User Flash 128 MB of NVRAM battery backed Two mailboxes Hi amp Lo per CPU Real time clock calendar 4 digit year Four asynchronous RS 232 423 serial ports up to 115 Kbps Dual Gigabit Ethernet Ports Single 64 bit PMC slot Introduction 1 4 Diskless Implementation 1 4 1 Virtual Root The virtual root directory is created on the File Server for each client when the client is configured The virtual root directory is used to store the kernel build environment clus ter configuration and device files In addition for clients configured with NFS the client s etc var tmp and dev directories are created here and NFS mounted on the client during system initialization Each configured client has its own unique virtual root on the File Server which is used as the configuration environment for that client A client s virtual root direc
52. default for an NFS client and for an embedded configuration are listed in modlist nfs netboot and modlist emb netboot respectively under the directory path usr etc diskless d sys conf kernel d These template files should not be modified For diskless clients only one copy of the unix file the kernel object file is kept under the virtual root When a new kernel is built the current unix file is over written System diagnostic and debugging tools such as crash 1M and hwstat 1M require access to the unix file that matches the currently running system Therefore if the kernel is being 2 5 1 1 Netboot System Administration modified while the client system is running and the client is not going to be immediately rebooted with the new kernel it is recommended that the current unix file be saved Modifications to a client s kernel configuration can be accomplished in various ways the commands referenced below should be executed on the file server system 1 Additional kernel object modules can be automatically configured and a new kernel built by specifying the modules in thekernel modlist add custom file and then invoking mknetbstrap 1M The advantage of this method is that the client s kernel configuration is recorded in a file that is utilized by mknetbstrap 1M This allows the kernel to be easily re created if there is a need to remove and recreate the client configuration 2 Kernel modules may be manually configured
53. download and boot the client s netboot image then test the booting of the client by either resetting or power cycling the board If the client is not configured to autoboot then type the following commands at the client s console terminal STARO gt astrix cr tftp g r client profile filename bstrap SHOST cr boot 5 cr where client profile filename is equal to the client profile filename of the client on the File Server system Boot configuration slot number 5 should have already been setup as specified in the section SBC Client Board Configuration Verifying Boot Status If the client is configured with NFS support you can verify that the client was successfully booted using any one of the following methods e rlogin 1 or telnet 1 from the File Server system or attach a terminal to the console serial port and login You can also use the ping 1M command to verify that the network interface is running However this does not necessarily mean that the system successfully booted If the client does not boot verify that the NFS daemons are running by executing the nfsping 1M command on the File Server An example run of this command is shown below H nfsping a nfsping rpcbind is running nfsping nfsd is running nfsping biod is running nfsping nfsping nfsping nfsping nfsping nfsping Netboot System Administration mountd is running lockd is running statd is running bo
54. e 0 5 from top of spine Helvetica 36 pt Bold v gt X e o Volume Number if any Helvetica 24 pt Bold Volume Name if any Helvetica 18 pt Bold Manual Title s Helvetica 10 pt Bold centered vertically within space above bar double space between Power Hawk 900 each title Diskless Svstems Administrator s Guide Bar 1 x 1 8 beginning 1 4 in from either side Part Number Helvetica 6 pt centered 1 8 up 0891090
55. e file server has created this client s netboot image entry the following command to reboot the system and test the new client s netboot setup reboot b If you wish to always manually boot the client from ASTRix then take the following steps If this board was previously setup with a nvram startup script for automatic netbooting then remove this nvram startup file so that auto matic netbooting will no longer occur rm nvram startup cr The above command only needs to be issued once and only if the board was previously setup for automatic netbooting of the client If the client accesses the file server through a gateway then enter the fol lowing ASTRix command Otherwise skip this command route add net XXX XXX XXX XXX ZW XXX XXX XXX XXX V netmask xxx xxx xxx xxx dev ethO Replace the above net and gw and netmask IP addresses with the appro priate values The net IP address should be the remote network that the file server resides on and the gw IP address should be the local gateway sys tem that provides the client with access to the remote network netmask is typically set to a value of 255 255 255 0 for class 3 IP addressing 2 11 Power Hawk Series 900 Diskless Systems Administrator s Guide To download the client netboot image from the file server enter the follow ing command tftp g r client bstrap cr where client should be replaced by the actual hostname of the client I
56. each client would typically be booted using the netboot tftp method as previously described in the Netboot Sys tem Administration chapter When entering the deployed phase the client could be converted from network tftp booting to Flash booting where the already existing client boot sequence would need to be slightly modified from copying a netboot image from the file server to instead loading and booting the client boot 3 1 Power Hawk Series 900 Diskless Systems Administrator s Guide image directly from Flash AII the other steps of the client s boot sequence are performed as before Flash booting should not be thought of as a totally separate function from booting a client but as a minor and significant adjustment to the standard client boot sequence that must always occur Any client boot image which is netbootable can be stored into Flash and booted directly from Flash No special preparation or treatment of the client boot image itself is needed There are several advantages to storing the client boot image in Flash It gives a client some independence from a File Server SBC For embedded clients especially it is possible to develop a boot image that makes no ref erence to a File Server at all thus resulting in a client that can be placed into a standalone configuration during the deployed phase Bootup times are faster when booting from Flash as no boot image down load occurs This can be especially important d
57. eckIshPLoeBIP IGGNa A 4 1 Crash usd Ra I oe AGER Vas en re odia epa x da Aid 4 2 Appendix A Adding a Local Disk A 1 Appendix Make Client System Run in NFS File Server Mode B 1 Glossaty ote v REPE ein A an he Glossary 1 Index ee bebe eb be est Index 1 List of Illustrations Figure 1 1 Loosely Coupled System Configuration 1 2 Figure 1 2 Power Hawk Networking Structure 1 13 List of Tables Table 2 1 Boot Image 2 32 vi 1 1 1 2 1 3 1 4 1 5 1 6 1 Introduction OVGrVIEW i eie ewe esee veh eere stis 1 1 1 1 1 Looselv Coupled Systems 8 1 1 1 1 2 Diskless Boot Basics sese 1 2 1 1 3 Net Boot Toolset case d IA l 3 Definitions i NU me d EUIS 1 4 Hardware Overview os sii eu Rer pe pr RE EC RS 1 6 1 3 1 Model 920 Hardware Feature Summary 1 6 1 3 2 Model 940 Hardware Feature Summary 1 6 Diskless Implementation l 7 14 1 Virtual Root 555a A SEES 1 7 1 4 2 Boot Image Creation and 1 7 1 4 3 1 8 144 Bootin
58. ent SBC The client profile file is loaded by the loosely coupled tools and various start up scripts After the initial client is configured with netbootconfig 1M this file must not be modified Each required parameter must be assigned a value in the ksh loadable format lt parameter value gt No spaces are allowed on either side of the equal sign Parameters specified as optional may be left blank lt parameter gt An explanation of each of the parameters follows Required Parameters The following client profile parameters are required for all loosely coupled netboot clients Power Hawk Series 900 Diskless Systems Administrator s Guide VROOT This parameter is the directory path name under which the client s virtual root directory will be created The directory will be created if it doesn t already exist Example VROOT home vroots wilma SYS CONFIG This parameter specifies the client configuration to be either NFS or embedded An NFS client is configured with networking executes in multi user mode and has the ability to swap memory pages to a remote swap area on the File Server An embedded client does not have networking support cannot swap out memory pages and runs in single user mode This parameter should be set to either nfs or emb Example SYS CONFIG nfs BOOT IFACE net Specifies the networking interface used for loading a diskless client s boot image This parameter is set to net by default and
59. ent depend on the diskless system architecture Appendix provides instructions on how to add a local disk to a client Appendix B provides instructions on how to make a client system run in NFS File Server mode The Glossary explains the abbreviations acronyms and terms used throughout the manual The Index contains an alphabetical list of all paragraph formats character formats cross reference formats table formats and variables The following notation is used throughout this guide italic Books reference cards and items that the user must specify appear in italic type Special terms may also appear in italic list bold User input appears in list bold type and must be entered exactly as shown Names of directories files commands options and man page references also appear in list bold type iii Power Hawk Series 900 Diskless Systems Administrator s Guide list Operating system and program output such as prompts and mes sages and listings of files and programs appears in 1ist type Brackets enclose command options and arguments that are optional You do not type the brackets if you choose to specify such option or arguments Referenced Publications Concurrent Computer Corporation Manuals Vendor Manuals Title Pubs No System Administration Manual Volume 1 0890429 System Administration Manual Volume 2 0890430 Power Hawk Series 900 PowerMAX OS Version 6 3 Release Notes 0891089
60. er If the changes are specific to a particular netboot client a private copy of the shared memfs files add file should first be created in that client s private customization directory etc profiles client profile filename gt net custom conf by using the mkprivate tool and then editing it in that custom conf directory Guidelines for adding entries to this table are included as comments at the beginning of the table A file may need to be added to the memfs files add table if 2 25 Power Hawk Series 900 Diskless Systems Administrator s Guide 1 The client is configured as embedded Since an embedded client does not have access to any other file systems then all user files must be added via this table 2 Theclient is configured with NFS support and a the file needs to be accessed early during a diskless client s boot before run level 3 when the client is able to access the file on the File Server system via NFS b itis desired that the file is accessed locally rather than across NFS For NFS clients the system directories etc usr sbin dev var opt and tmp all serve as mount points under which remote file systems are mounted when the diskless client reaches run level 3 Files added via the memfs files add table should not be installed under any of these system directories if they need to be accessed in run level 3 as the NFS mounts will overlay the file and render it inaccessible Files added via the me
61. er applications and data must be placed into the memfs root filesystem since by definition no other file systems are available to such clients The tools used to build boot images provide a mechanism for adding user defined files to the memfs filesystem contents and wraps those contents into the boot image that will be later downloaded into the client When the boot image is downloaded into a diskless cli ent system it is resident in memory whether or not the files are being used This means that the number and size of files that can be placed into the memfs file system is thus lim ited This effect is minimized if the boot image is burned into flash using the raw write method via the ASTRix fw flash write command When the boot image resides in Flash using the raw write method then only those files actually in use will reside in i e be cop ied into physical memory at any time In this mode of operation the root filesystem behaves more like a normal filesystem pages are automatically fetched from the Flash as needed and if not modified by applications are automatically released when other needs for the space become more urgent It is possible for applications running on the client to write to memfs files however there not being a disk associated with these files the changes will be lost on the next reboot of that client Moreover such pages remain permanently in memory until the files contain ing them are deleted or truncated This t
62. er on when running under ASTRix NOTE If the client and server SBC are not connected to the same net work and the client accesses the server SBC through a gateway then this routing issue is handled under ASTRix which is dis cussed later on in step 10 below You may optionally setup the client SBC so that ECC error correction and detection is enabled in the SDRAM memory If you do NOT wish to enable ECC for the SDRAM memory then set the Enable post script parameter of the STAR config command to and proceed to step 8 The enabling of ECC checking in SDRAM is accomplished through the use of post script commands The execution of post script commands may Netboot System Administration be enabled when running STAR but the creation and setup of these post scripts is accomplished while running under ASTRix So at this point simply enable post script execution using the STAR config command Enter cr until the following parameter appears Enable post script and then set this parameter to Y The creation of the proper post scripts will be discussed in step 8 The remaining setup of the client is accomplished while running ASTRix Since STAR configuration parameters have changed reset the board so that the new STAR parameters will now take affect Use the STAR reboot command to reset the board STARO gt reboot cr The board should reset itself and STAR should automatically boot into ASTRix Y
63. ervice UDP A term used in UNIX derived systems indicating the level of services available in the system Those at init level 1 are single user systems which in turn is typical of embedded systems running on cli ent SBCs Those at init level 3 have full multi user networking and NFS features enabled and is typical of client SBCs that run as netboot clients See init 1M for complete details Swap reservation space referred to as virtual swap space is made up of the number of real memory pages that may be used for user space translations plus the amount of secondary storage disk swap space available Clients in the NFS configuration utilize a file accessed over NFS as their secondary swap space Embedded clients which are usually also Flashboot clients generally do not utilize a swap device but if a local disk is available then they too may be configured with a swap device Power Hawk Series 900 Diskless Systems Administrator s Guide 1 3 Hardware Overview 1 3 1 1 3 2 1 6 The Power Hawk Model 920 and 940 platforms are supported only as a loosely coupled client Motherboard Number Form Designation System Platform of CPUs Factor Netboot Flashboot VYFD Power Hawk Model 920 2 VME 6U yes yes VAFQ Power Hawk Model 940 4 VME 6U yes yes Model 920 Hardware Feature Summary SMP compliant PCI to VME64 bridge rated at 50 MB sec Dual G4 PowerPC 7455 7457 2 MB backsi
64. es of the manufactures or marketers of the product with which the marks or names are associated Printed in U S A Revision History Level Effective With Original issue January 2006 000 PowerMAX OS Release 6 0 beta Pevious Release Januarv 2004 010 PowerMAX OS Release 6 1 Previous Release November 2004 020 PowerMAX OS Release 6 2 Current Release January 2006 030 PowerMAX OS Release 6 3 Scope of Manual Preface Intended for system administrators responsible for configuring and administering diskless system configurations Structure of Manual Syntax Notation This manual consists of a title page this preface a master table of contents four chapters local tables of contents for the chapters two appendices glossary of terms and an index Chapter 1 Introduction contains an overview of Diskless Topography Diskless boot basics configuration toolsets definition of terms hardware overview diskless implementation configuring diskless systems and licensing details Chapter 2 Netboot System Administration provides an overview of the steps that must be followed in configuring a loosely coupled system LCS configuration Chapter 3 Flash Boot System Administration is a guide to configuring a diskless single board computer SBC to boot PowerMAX OS from flash memory Chapter 4 Debugging Tools covers the tools available for system debug ging on a diskless client The tools that are available to debug a diskless cli
65. f the tftp command fails then you may re enter the tftp command to try the download again Lastly to boot the client enter the following ASTRix commands sync cr boot 5 cr Note Boot configuration slot number 5 MUST have been previously setup as specified in step 9 The file server SBC should be at run level 3 before you attempt to boot the netboot client image 2 3 Modifying an Existing Automatic Netboot Configuration 2 12 If you have setup STAR ASTRix to automatically boot the client and at some point you wish to modify this configuration then shutdown the client if it is currently running PowerMAX OS or otherwise reset or power cycle the client board and then issue a lt Ctrl gt c from the console keyboard when the tftp g r lt client gt bstrap HOST message appears under ASTRix This will abort the file transfer from the file server At this point you may make modifications to the ASTRix post scripts and or the ASTRix startup script and then either issue reboot if you wish to reboot the client without any STAR configuration changes or issue star if you wish to enter STAR and modify one or more of the STAR config parameters Once the STAR config parameters have been modified you may enter STARO gt reboot cr to reboot the board and to have the STAR config parameters take affect and have STAR automatically boot ASTRix Netboot System Administration 2 4 Client Configuration
66. fying additional options For example one might update the files under the virtual root directory Some examples are shown below Example 1 Create the diskless client configurations for all clients that have netboot client pro file files in the etc profiles directory Process at most three clients at the same time netbootconfig C p3 all Example 2 Remove the client virtual root configuration of netboot client rosie Send the output to stdout instead of to the client s log file netbootconfig R t rosie Example 3 Update the virtual root directories of netboot clients fred and barney Process one client at a time netbootconfig U v p 1 fred barney 2 4 2 2 Subsystem Support A subsystem is a set of software functionality package that is optionally installed on the File Server during system installation or via the pkgadd 1M utility Additional installation steps are sometimes required to make the functionality of a package usable on a diskless client Subsystem support is added to a diskless client configuration via netbootconfig 1M options when invoked in either create or update mode Subsystem support is added to a client configuration via the a option and removed via the r option For a list of the current subsystems supported see the netbootconfig 1M manual page or invoke netbootconfig 1M with the help option h 2 17 Power Hawk Series 900 Diskless Systems Administrator s Guide 2
67. g i dij eg e pitas dee dhe dead eet ek kes 1 9 144 1 Net iR A a g a 1 10 144 2 Flash EI EE 1 11 1 4 5 Remote File Sharing 1 12 1 3 6 SWAP Space iere here NR ever Ve e bet 1 15 Configuring Diskless 1 17 1 5 1 Looselv Coupled System Hardware Prerequisites 1 17 1 5 2 Disk Space 1 18 1 5 3 Software Prerequisites 1 18 Licensing Information 20 eee l 19 Power Hawk Series 900 Diskless System Administrator s Guide 1 Introduction 1 1 Overview This manual is a guide to diskless operation of PowerMAX OS Diskless operation encompasses the ability to configure boot administer and debug systems that do not have attached system disks It should be noted that such a system might have attached non system disks Each diskless system runs its own copy of the PowerMAX operating svs tem This manual specifically discusses Power Hawk Series 900 diskless configuration Power Hawk Series 900 systems are currently supported in Loosely Coupled System LCS con figurations where a mix of Power Hawk Series 700 and Series 900 systems may be con figured within the same LCS configuration Only the configuration and use of Power Hawk Series 900 systems are discussed in this manual The user should refer to the Power Hawk Series 700 Diskless System Admini
68. guration do affect even embedded systems so this section should be read even for these users Normal systems have a disk partition reserved for swap space For diskless nfs clients swap space is implemented using a regular file created in the client s virtual root directory and accessed over NFS The size of the swap file is user configurable Swap reservation space referred to as virtual swap space is made up of the number of real memory pages that may be used for user space translations plus the amount of secondary storage disk swap space available If no secondary storage swap space is available then the amount of virtual swap space degenerates to the number of real mem ory pages available for user address space A virtual swap space reservation is made by decrementing the amount of available virtual swap space If no virtual swap space is available then the reservation will fail and subsequently either the page fault or segment create operation will not succeed Virtual swap reservations are made so that as real memory becomes low the pageout and process swap daemons can guarantee that there will be an appropriate number of user pages that can be swapped out to secondary storage and subsequently freed in order to maintain an adequate level of free memory pages for new page allocations Even when there is no swap space configured into the kernel the virtual swap reservations will prevent the kernel from over committing real memory
69. h the other embedded or nfs clients therefore a change to one of these files will affect all the clients in the loosely coupled system NOTE Please note that if the File Server is a Power Hawk Series 700 sys tem and it is also supporting Series 700 closely coupled clients then changes to the shared custom conf files also affect the File Server s closely coupled clients as well as the File Server s loosely coupled clients See the Power Hawk Series 700 Diskless Systems Administrator s Guide for more details on closely cou pled clients Under each client s private custom conf directory two commands mkprivate and mkshared are available to change the state of a custom file from being shared to being private Before creating a new version mkshared will save the current version to a file 2 21 Power Hawk Series 900 Diskless Systems Administrator s Guide named lt customfile gt old mkprivate will move the current version to a file named lt customfile gt linked To make a change that is private to a client l verify that the custom file is NOT symbolically linked d etc clients lt client gt net custom conf Is 1 lt customfile gt 2 if the file is currently symbolically linked first break the link mkprivate lt customfile gt 3 verify that the file is a regular file and edit the file 1s 1 customfile H vi lt customfile gt To make a change that will affect all the diskless clients configured to sha
70. ies up precious physical memory This can be alleviated only by the addition of a swap device to the system to which the system can write these dirty pages to as necessary or by careful consideration and minimization of how much file writing is done by embedded applications into the memfs root filesystem Memfs file systems stored in Flash will be in a compressed format in order to make maximum use of this relatively tiny device Please refer to the Flash Boot System Admin istration chapter for more details on Flash boot configuration 1 4 4 Booting Introduction Once a bootstrap image is generated it must be loaded and started on the client for which it was built Two methods of booting are provided Net Boot and Flash Boot These meth ods are explained in more detail in the following sub sections Booting a diskless client consists of two Net boot or three Flash boot distinct opera tions Step 1 Load the boot image into the client s SDRAM Step 2 Burn the boot image into flash Flash boot method only Step 3 Initiate execution of the boot image Step 1 Regardless of whether the user is using Net boot or Flash boot method the boot image must be downloaded to the diskless client Downloading is per formed across an ethernet network with the ASTRix tftp command The tftp download is initiated automatically with an ASTRix startup script or manually at the client s attached terminal console terminal The ASTRix tftp comman
71. if this file doesn t 2 7 Power Hawk Series 900 Diskless Systems Administrator s Guide 9 exist vi post cr Add the following lines to the post script file using the vi editor if cpuid 0 imedc 0 0 1 Use the wq vi command to write the file and exit vi You may then cat nvram post cr to check that the file contents are correct If the contents are not correct then use vi post to correct the contents The above post script will cause CPU 0 to initialize memory and enable ECC checking in SDRAM when ever STAR begins executing after a reset or power on To create the post script that will be executed before the client netboot image is executed issue the following commands cd nvram cr rm postO5 cr ignore any error messages if this file doesn t exist vi postO5 cr Then add the following lines to the post05 script file using the vi editor if cpuid 0 serialok 1 printf MEnabling ECC n serialok 0 eecc Use the wq vi command to write the file and exit vi You may then cat nvram post05 cr to check that the file contents are correct If the contents are not correct then use post05 to correct the contents of the post file When the client boot image is booted the post05 script above will output the Enabling ECC message to the console terminal and will then re enable ECC checking in SDRAM This step will setup an A
72. ince all the boot images are stored on the File Server s disk Client All SBCs except for the File Server are considered clients Clients do not have their own system disk Clients must rely on the File Server for such support However clients may have local non system disk drives configured The two client configurations embedded and NFS are described below 1 Embedded Client An embedded client runs self contained from an internal memory based file system they do not offer console or network services There is no swap space because there is no media that can be used for swapping pages out of main memory Applications run in single user mode init state 1 2 NFS Client NFS clients are diskless SBCs that are configured with networking and NFS Most directories are NFS mounted from the File Server In addition to NFS all standard PowerMAX OS network protocols are available Swap space is configured to be remote and is accessed over NFS Applications run in multi user mode init state 3 System Disk Net Boot or Network Boot Flash Boot Boot Image Synergy Monitor STAR ASTRix Trivial File Transfer Protocol TFTP System Run Level Init Level swap space Introduction The PowerMAX OS requires a number of system directories to be available in order for the operating system to function properly These directories include etc sbin dev usr var and opt The File Server is configured so
73. ing into ASTRix To get back into STAR from ASTRix simply enter the following command while running ASTRix star lt cr gt This command will reboot the board and return to STAR without autobooting ASTRix regardless of the current setting of the Enable ASTRix autoboot parameter Additional parameters in the STAR config command are listed below Netboot System Administration The Port A baudrate parameter should be set to 9600 and it is recom mended that you also set the Port B baudrate to 9600 especially if you plan to hook up a second terminal to Port B The Share console port parameter MUST be set to Y It is highly recommended that you do NOT modify the Board serial num ber parameter This parameter is used to create unique Ethernet hardware addresses for all of the on board Ethernet controllers The PCI 0 and 1 Memory Space config command parameters should beset to the values shown above If they are not setup with these values then use the STAR config command to set them to the values shown above After you have successfully netbooted the client you may want to change the PCI Memory Space layout from the above default values Please consult the Power Hawk Series 900 Console Reference Manual Pubs No 0830060 section PCI Memory Space Configuration for more information on this subject It is highly recommended that you do NOT set the Enable quiet boot mode to Y Th
74. ing significant amounts of additional free memory The address space aging interval should also be increased System tunables that govern the address space aging interval are INIT AGEQUANTUM MIN AGEQUANTUM MAX AGEQUANTUM LO GROW RATE HI GROW RATE In order to ensure longer address space aging intervals all of these tunables may be set to a higher than default value The GPGSLO tunable value can be decreased in order to lower the free memory level at which process swapping will become aggressively active The DISSWAPRES tunable disables virtual swap reservations by setting the amount of available virtual swap space to an artificially large value The DISSWAPRES tunable allows more user page identities objects to be created than what can be accommodated for in virtual swap space Since typically applications do not tend to access all the pages that may potentially be considered writable and therefore require a virtual swap reservation this tunable may allow for a larger number of applications to run simultaneously on a system by not requiring virtual swap space for every potentially writable user page However when the DISSWAPRES tunable is enabled it becomes possible for page allocations to block forever since the pageout and process swap daemons may not be able to swap out an adequate number of user pages in order to free up pages for additional allocations At this point the system will enter a state where little or no useful
75. ional considerations for adding entries to the vroot files add table The client s usr and sbin svstem directories are shared completelv with the File Server hence these directories do not appear under a client s virtual root and mav not be used in the vroot files add table The files in the vroot files add table are copied into a client s virtual root partition and therefore require disk space on the File Server svstem In some cases it mav be more efficient to NFS mount a user s working directorv on the client svstem instead of duplicating the files in the client s virtual root directorv Because of kernel dependencies device files should be created locally in the client s virtual root directory this vroot files add file table should NOT be used for this purpose To add a device file to a client s vroot the corresponding kernel module must be enabled config r vroot path the corresponding Node 4 file under the client s vroot may need to be modified and the client s kernel must be rebuilt and rebooted mknetbstrap B r unix client profile filename 2 5 3 Modifying the Client Profile Parameters 2 28 To modify the parameter values in a netboot client profile file the client configuration must be removed and reconfigured The only exceptions to this rule are the AUTOBOOT and SWAP SIZE parameters discussed separately below Itis best that the netboot NFS client be shutdown before modifying any of its client
76. is parameter disables all STAR output to the console ter minal However if you do happen to enable quiet mode by accident then enter the following command to disable quiet mode The output of this command will NOT be echoed to the terminal serialok 1 cr After this command has been entered then execute the config command again and disable the Enable quiet boot mode parameter Even if you wish to enable ECC memory error correction and detection leave the Init mem EDC parameter set to The enabling of ECC will be accomplished through the use of a post script and is described in a fol lowing step in this section The netboot client s hardware clock must be updated to match the date on the system designated as the File Server Use the STAR date command to display and or set the current date and time To display the date time val ues enter the date command with no arguments STARO date X 2 10 2006 TUE 15 44 38 GMT 0 To set the date and time to a value that matches the File Server s date and time displayed via a date 1 command when executed on the File Server use the STAR date command where the new date and time values are spec ified in the following format date YYMMDD HHMMSS tz The help date STAR command will output more information about the date command Power Hawk Series 900 Diskless Systems Administrator s Guide It is best to use a tz timezone value of 0 GMT timezone and adjust the
77. kprivate tool script in the clients s custom conf directory and edited there KOOclient Script is executed during svstem shutdown It is executed on the client from the path etc rc0 d K00client By default this file is emptv S25client Script is executed during system start up It is executed on a client configured with NFS support from the path etc rc3 d S25client For embedded configurations it is executed from etc rcl d S25client By default this file is empty 2 5 2 2 memfs inittab and inittab Tables These tables are used to initiate execution of programs on the client system Programs listed in these files are dispatched by the init process according to the init level specified in the table entry When the system initialization process progresses to a particular init level the programs specified to run at that level are initiated It should be noted that embedded clients can only execute at init level 1 since an embedded client never proceeds beyond init level 1 NFS clients can execute at init levels 1 2 or 3 Init level 0 is used for shutting down the system See the online man page for inittab 4 for more information on init levels and for information on modifying this table The mem s inittab table is a part of the memory based file system which is a component of the boot image Inside the boot image the files to be installed in the memory based file system are stored as a compressed cpio file When the 2 23 Power Ha
78. l actually be referencing the etc chroot file that exists in the File Server s etc directory File Server System root letc virtual root gt etc i nodename chroot shared etc chroot Client System root etc server virtual root gt etc shared etc server etc nodename chroot chroot Figure 1 2 Power Hawk Networking Structure As new files are added and removed from the File Server s etc and var directories the symbolic links under the client s virtual root may become stale The configuration util ity mknetbstrap can be used to update the links in these directories to match the cur rent state of the File Server The directories dev and tmp are also created under the client s virtual root but do not share any files with the File Server Device files may have kernel dependencies and so these files are not shared The directory tmp is created as an empty directory The directory users is also empty and may be used to access user files across NFS Once the client system is up and running the files in the memory based root file system required for booting are no longer needed and are removed to free up memory 1 13 Power Hawk Series 900 Diskless Systems Administrator s Guide Permission to access remote files on the File Server is automatically granted During client configuration the etc dfs dfstab see dfstab 4 and usr etc diskless d cluster co
79. lude the client in its rw and root attribute specifications For example if the virtual root directory for the client named client is in home vroots clienti and the home vroots directory is currently shared then the sample entry below would be changed as shown below from usr sbin share F nfs d home vroots home vroots vroots to usr sbin share F nfs o root client1 d home vroots home vroots vroots There is no need to add an rw attribute since when not specified it defaults to read write access permissions to all If no entry is currently shared that covers the client s virtual root directory then a specific entry for each client is appended to the d stab diskless file For example for the client named client1 whose virtual root directory is home vroots clientl the following entry is generated usr sbin share F nfs o rw client1 root client1 d home vroots clientl home vroots clientl vroot After the files are updated the shareall F nfs command is executed to update the File Server s shared file system table Introduction When a client s configuration is removed all references to the client and it s virtual root directory are removed from both the d stab and d stab diskless files and the unshare 1M command is executed to update the shared file system table 1 4 6 Swap Space Embedded systems generally do not have swap space Nevertheless some aspects of swap space confi
80. m s files add table are memory resident and diminish the client s available free memory This is not the case for a system where the boot image is stored in flash using the raw write method with the ASTRix fw flash write command since pages are brought into DRAM memory from flash only when referenced Please refer to the Flash Boot System Administration chapter for more details on flash booting Changes to the memfs files add file are processed the next time mknetbstrap 1M is invoked A new memfs image and boot image is then created A client must be rebooted using the new boot image in order for these changes to take effect You can verify that the file has been added to the mem s cpio image using the following command on the File Server d virtual rootpath etc conf cf d memfs cpio cpio itcvl grep lt file gt 2 5 2 6 vroot files add Table 2 26 This custom client configuration table may be used to optionally specify a set of non system files that are located on the File Server to be automatically copied by mknetbstrap 1M into a a client s virtual root directory so that they can be subsequently accessed from the client system This custom client configuration file may only be used by NFS netboot clients the embedded netboot clients are unable to access their virtual root on the File Server system This table is processed by mknetbstrap 1M whenever this table has been modified since the last invocati
81. mbedded do not depend upon the File Server system once they are up and running NOTE A netboot client may download the boot image and instead of booting from it may burn the boot image into its User Flash Memory for later booting This is called Flash booting and it is described in a separate chapter Refer to Chapter 3 Flash Boot System Administration for more information on Flash Booting Prior to net booting verify that the following steps have been completed 2 33 Power Hawk Series 900 Diskless Systems Administrator s Guide 1 Verify that the networking parameters have been setup on the client board with the STAR config command If the client is to automatically boot up after reset or power cycle then ver ify that STAR is configured to automatically boot ASTRix and that an ASTRix startup script has been setup to download and boot the client image from the File Server See the SBC Client Board Configuration sec tion for details 2 Verify that the boot image has been created See Creating the Boot Image on page 2 32 3 Verify that the File Server is up and running in run level 3 2 6 3 1 Netboot Using ASTRix 2 34 Once the boot image is generated and the File Server is accepting network requests is up and is at init state 3 you can test the network booting of a client by one of the following methods If the client is configured to autoboot ASTRix and an ASTRix startup script is setup to tftp
82. n array of integers The data structures associated with drivers are used as buffers for holding data being moved between user data space and the device as flags for indicating error device status as pointers to link buffers together and so on data terminal ready DTR data transfer The signal that a terminal device sends to a host computer to indicate that a terminal is ready to receive data The phase in connection and connection less modes that supports the transfer of data between two DLS users Glossary 3 Power Hawk Series 900 Diskless Systems Administrator s Guide device number diagnostic DLM DRAM driver entry points driver embedded The value used by the operating system to name a device The device number contains the major number and the minor number A software routine for testing identifying and isolating a hardware error message is generated to notify the tester of the results Dynamically Loadable Modules Dynamic Random Access Memory Driver routines that provide an interface between the kernel and the device driver The set of routines and data structures installed in the kernel that provide an interface between the kernel and a device The host system provides a boot image for the client system The boot image contains a UNIX kernel and a file system image which is configured with one or more embedded applications The embedded applications execute at the end of the boot sequence
83. n the client s SBC Example SWAP SIZE 192 ETHER SUBNETMASK This parameter specifies the ethernet interface subnetmask in decimal dot notation XXX XXX XXX XXX Example ETHERNET SUBNETMASK 255 255 255 0 GATEWAY IPADDR The IP address in decimal dot notation xxx xxx xxx xxx of a gateway system that provides this client with access to the File Server This parameter is optional and it is only necessary to setup this parameter when the File Server and this netboot client do not reside on the same physical subnetwork Example GATEWAY IPADDR 129 158 64 40 2 4 1 3 Hosts Tables For each loosely coupled client profile file created in the etc profiles directory an entry for that client s onboard ethernet networking interface must be added to the systems hosts 4 file etc hosts The client hostname added to the etc hosts file must match the client profile filename of the client For example if a new client profile file named etc profiles fred has just been created then an entry with a hostname of fred must be added to the etc hosts file The corresponding IP address for each new etc hosts entry should be chosen based on local rules for the ethernet subnet This IP address should match the value entered for this client SBC under the STAR config command s Target IP address parameter as described under step 5 in the SBC Client Board Configuration section Power Hawk Series 900 Diskless Systems Administr
84. n the connection less datagram delivery service UDP twisted pair Ethernet 10base T An Ethernet implementation in which the physical medium is an unshielded pair of entwined wires capable of carrying data at 10 Mbps for a maximum distance of 185 meters Glossary 11 Power Hawk Series 900 Diskless Systems Administrator s Guide twisted pair Ethernet 100base T UDP unbuffered I O upstream user space yellow pages Glossary 12 An Ethernet implementation in which the physical medium is an unshielded pair of entwined wires capable of carrying data at 100 Mbps for a maximum distance of 185 meters User Datagram Protocol provides low overhead connection less datagram delivery service I O that bypasses the file system cache for the purpose of increasing I O performance for some applications The direction of STREAMS messages flowing through a read queue from the driver to the user process The part of the operating system where programs that do not have direct access to the ker nel structures and services execute The UNIX operating system is divided into two major areas the user programs and the kernel Drivers execute in the kernel and the user programs that interact with drivers execute in the user program area This space is also referred to as user data area See definition for NIS Network Information Services Symbols letc clients 2 21 etc hosts 2 15 etc profiles 2 13 2 28 Numerics 100base T
85. nf dfstab diskless tables are modi fied to allow a client either read or read write access to files which reside on the File Server The dfstab diskless file is generated when the first client is configured Sample entries from this file are listed below These entries should not be modified share share share share share share F F F F F F nfs nfs nfs nfs nfs nfs O ro root SCLIENTS d etc etc O ro root SCLIENTS d dev dev O rw SCLIENTS root SCLIENTS d var var ro root CLIENTS d sbin sbin O rw SCLIENTS root SCLIENTS d usr usr Oo rw SCLIENTS root SCLIENTS d opt opt shared etc shared dev shared var sbin usr opt 1 14 The dfstab diskless file is referenced from a command line entry in dfstab generated when the first client is configured For every client configured thereafter the client s name is added to the CLIENTS variable For example after configuring two clients named client and client2 the following line appears in the dfstab table CLIENTS clienti client2 usr sbin shareall F nfs usr etc diskless d cluster conf dfstab diskless In addition an entry to make each client s virtual root directory accessible is generated at configuration time If a parent directory of the client s virtual root directory is already cur rently shared has an entry in sharetab 4 then the matching entry in dfstab 4 if found is modified to inc
86. nt The boot image contains everything needed to boot a diskless client The components of the boot image are unix kernel binary compressed cpio archive of a memory based file system abootstrap loader that uncompresses and loads the unix kernel Each diskless client has a unique virtual root directory Part of that virtual root is a unique kernel configuration directory etc con for each client The boot image file unix bstrap in particular two of its components the kernel image unix and a memory based file system memfs cpio are created based on configuration files that are part of the client s virtual root The makefile etc diskless d sys conf bin d bstrap makefile is used by mknetbstrap 1M to create the boot image Based on the dependencies listed in that makefile one or more of the following steps may be taken by mknetbstrap 1M in order to bring the boot image up to date 1 Build the unix kernel image and create new device nodes 2 31 Power Hawk Series 900 Diskless Systems Administrator s Guide 2 Create and compress a cpio image of the files to be copied to the memfs root file system 3 Insert the loadable portions of the unix kernel the bootstrap loader the compressed cpio image and certain bootflags into the unix bstrap file The unix kernel portion in unix bstrap is then compressed When mknetbstrap is invoked updates to key system files on the File Server i e etc inet hosts will cause the
87. nt of file space required for the client virtual root directories and t tpboot boot images On the File Server system in the etc profiles directory create a client profile file for each netboot client 2 1 Power Hawk Series 900 Diskless Systems Administrator s Guide You can use netbootconfig 1M to print out a template of a netboot client profile The client profile file name must be equivalent to the client s Ethernet hostname For example to create a netboot client profile for a client with a hostname of wilma do the following netbootconfig P etc profiles wilma vi etc profiles wilma Edit the resulting client profile file to be relevant to the specific characteristics of the client SBC The parameters listed in the client profile are described in The Client Profile File on page 2 13 Additional description of these parameters in the file usr etc diskless d profiles conf netboot client profile README Update the etc hosts file on the File Server SBC with the hostnames of all the netboot clients The hostname s added to the hosts file should match the filename s of the client profile file s The IP addresses for each client should correspond to the IP address of the first onboard Ethernet controller dev gteO for Series 910 920 dev mve0 for Series 940 On the File Server system execute netbootconfig 1M to configure the build environment of each diskless client to be configured See the Configuring
88. od depending upon their needs 3 3 The Flash Filesystem Method To convert the netboot client into a Flash filesystem booted client take the following steps NOTE This procedure assumes that the user has already setup and booted the client successfully as a netboot client using the procedures described in the Netboot System Administration chapter 1 Reset or power cycle the client board and wait for STAR to autoboot into ASTRix 2 Abort the startup script with a lt Ctrl gt c keyboard sequence if an auto matic netboot startup script has previously been setup to netboot the client 3 Tftp the netboot image into the client s memory from the File Server SBC tftp g r client bstrap HOST cr where client should be replaced with the name of the client profile file that is located on the File Server SBC in the etc profiles directory 4 Burn the boot image into a Flash filesystem If you choose to use the Boot Flash for storing the boot image then Flash filesystem number 3 should be used Power Hawk Series 900 Diskless Systems Administrator s Guide ffsw lt client gt bstrap 3 cr NOTE Multiple versions of a client bootimage may be stored into the same Flash filesystem by using a different name For example to store a second version of a boot image into Flash filesystem 3 while keeping the previous version mv lt client gt bstrap lt client gt bstrap2 cr ffsw lt client gt bs
89. on can also be automatically mounted and unmounted during the system s boot shutdown if desired Ifthe disk client node is another diskless client the mount points may be added to the memfs root file system via the memfs files add table and the automatic mounting may be achieved via the node s v stab file or the rc scripts shown below 1 Add directories to be used for mount points to the memfs filesystem etc clients disk client dir custom conf mkprivate memfs files add vimemfs files add Example entry d rem 050 0755 2 Add an entry to the client s startup script to automatically mount the partition etc clients disk client dir custom conf mkprivate S25client vi S25client Example entry if the disk server is up mount remote file system mount point rem 050 if ping disk server gt dev null then sbin mount F nfs disk server disk0s0 rem 0s0 fi 3 Addanentry to the client S shutdown script to automatically unmount the partition 4 etc clients disk client dirs custom conf d mkprivate KOOclient vi KOOclient Example entrv umount rem 0s0 10base T 100base T ARP AUI asynchronous Glossary See twisted pair Ethernet 10base T See twisted pair Ethernet 100base T Address Resolution Protocol as defined in RFC 826 ARP software maintains a table of translation between IP addresses and Ethernet addresses Attachment Unit In
90. on of mknetbstrap 1M Although non system files can be copied manually into a client s virtual root directories the use of this table provides an automated method that provides the following advantages This file table makes it easier to recreate a client s virtual root environment when a client is removed R and s options and then recreated C option with netbootconfig 1M Netboot System Administration Entries in this file table may be setup to have mknetbstrap 1M auto matically re copy the specified File Server source files into the client target virtual root directories every time this table is processed with the a option see below The format for each entry in this file is Path on server Path on client Options Lines beginning with the pound sign will be ignored The fields in this table are described below Path on server This is the pathname of a file or directory located on the File Server system that is to be copied into the client s virtual root When the pathname is a directory then the contents of this directory will be recur sively copied into the client s vroot directory Path on client This is the pathname of a file or a directory as it will be accessed from the client system If a directory in this path does not currently exist in the client s virtual root directory then it is created This path must begin with one of the system directories already under the client s vroot users
91. on the File Server since all the boot images and the system files are stored on the File Server s disk A kernel utility used in a driver The term function is used interchangeably with the term kernel function The use of functions in a driver is analogous to the use of system calls and library routines in a user level program A SBC running a full fledged PowerMAX OS system containing disks networking and the netboot development environment Called a File Server since it serves clients with boot images filesystems or whatever else they need when they are running The single board computer of the File Server A name that is assigned to any device that has an IP address A term used for the File Server It refers to the prerequisite Power Hawk system interprocess communication IPC A set of software supported facilities that enable independent processes running at the same time to share information through messages semaphores or shared memory Glossary 5 Power Hawk Series 900 Diskless Systems Administrator s Guide interrupt level interrupt vector ICMP ISO kernel buffer cache kdb loadable module MTU Driver interrupt routines that are started when an interrupt is received from a hardware device The system accesses the interrupt vector table determines the major number of the device and passes control to the appropriate interrupt routine Interrupts from a device are sent to the device s interrupt vect
92. opy the netboot image into the client s memory from the File Server SBC tftp g r client bstrap SHOST cr Power Hawk Series 900 Diskless Systems Administrator s Guide where client should be replaced with the name of the client profile file that is located on the File Server SBC in the etc profiles directory 4 Burn the boot image into Flash using the ASTRix Flash write command fw 2 0 lt client gt bstrap 5 fw Oxfa000000 lt client gt bstrap Replace client above with the same name used in step 3 Note Flash area 2 and the flash offset value of 0 MUST be used in the flash write fw command above in order for this method of Flash booting to function properly The above Flash address MUST be used in order for this method of Flash booting to function prop erly The fw command will replace overwrite any filesystem that was previously setup in Flash filesystem 3 in the Boot Flash Also note that the ffsis ASTRix command will therefore display the 3 BOOT area as Unused If Flash Booting is no longer desired the Flash filesystem 3 may be re created when a ffsw ASTRix command is issued for Flash filesystem 3 In this case the first ffsw command issued for Flash filesystem3 will recognize that the Flash filesystem is not present and this command will prompt the user to see if a new filesystem should be created 6 Place the flashboot program from the File Server into NVRAM The flash boot prog
93. or activating the interrupt entry point for the device Internet Control Message Protocol an integral part of IP as defined in RFC 792 This protocol is part of the Internet Layer and uses the IP datagram delivery facility to send its messages The Internet Protocol 791 is the heart of the TCP IP IP provides the basic packet delivery service on which TCP IP networks are built International Organization for Standardization A set of buffers used to minimize the number of times a block type device must be accessed Kernel debugger A kernel module such as a device driver that can be added to a running system without rebooting the system or rebuilding the kernel Maximum Transmission Units the largest packet that a network can transfer memory file system image Glossary 6 A cpio archive containing the files which will exist in the root file system of a client sys tem This file system is memory resident It is implemented via the existing memfs file system kernel module The kernel unpacks the cpio archive at boot time and populates the root memory file system with the files supplied in the archive memory management modem netboot netload network boot network load netstat NFS NFS client Glossary The memory management scheme of the UNIX operating system imposes certain restric tions on drivers that transfer data between devices A contraction of modulator demodulator A modulator convert
94. or s Guide config utility REPE E eR 2 19 Adt neob ee sees e e 2 20 Custom Configuration Files 0 eee 2 20 S25client and KOOclient rc Scripts 2 23 memfs inittab and inittab 2 23 Vistab Table soho teu t ee Sisal at oh 2 24 kernel modlist add 2 25 memfs files add Table 2e epe ERR 2 25 vroot files add Table 00 2 26 Modifying the Client Profile Parameters 2 28 Launching Applications eee 2 29 Launching an Application for Embedded 2 29 Launching an Application for NFS 2 30 Booting and e 2 31 The Boot Image posee A RE eed eee tes Be 2 31 Creating the Boot Image 2 32 Examples on Creating the Boot Image 2 33 Net Booting o IN EU 2 33 Netboot Using ASTRix 1 0 ee 2 34 Shutting Down the Client 2 35 Chapter Flash Boot System Administration Introduction ere ee pte ee E EAR 3 1 Flash Characteristics coser never Y Ue PERPE PER EA AC 3 2 The Flash Filesystem 858 3 3 The Raw Write Method eee en 3 5 Chapter 4 Debugging Tools System Debugging Tools 4 1 kdb s
95. ose contents are not lost across system reboots or power cycles Power Hawk Series 900 systems come with at least 32 MB of Boot Flash The actual STAR ASTRix boot code uses a small portion of this 8 MB The rest of the Boot Flash space is free for users to store their own data Additional User Flash can be optionally added to the board by the factory in 32 MB incre ments up to a total of 96 MB NOTE For Power Hawk Series 920 systems please refer to the Boot and Flash Memory section of the Raptor DX VMEbus Dual G4 Dual PMC amp StarFabric User Guide for more information about User Flash modules and their use For Power Hawk Series 940 systems please refer to the Manta QX VMEbus Quad Single PMC amp StarFabric User Manual While Flash reads behave much like normal memory reads Flash writes are accomplished through use of a set ASTRix flash write commands that support bulk erasure and repro gramming of Flash with any desired set of user data that will fit onto the device The rest of this chapter discusses using some portion of Boot or User Flash to store a cli ent s netboot image so that the client may be thereafter booted directly from Flash without the need to re download the client s boot image from the file server with tftp Flash booting would typically be done as part of the transition from the software develop ment and testing phase of a project to the production or deployed phase While in the development phase
96. otparamd is running penfsd is running The svstem is running in client server bootserver and pc server modes If there is a console attached to the client and the client appears to boot successfullv but cannot be accessed from any other system verify that the inetd 1M daemon is running on the client 2 6 4 Shutting Down the Client From the client s console the client may be shutdown using any of the system shutdown commands e g shutdown 1M or init 1M A client configured with NFS can be shutdown from the File Server using the rsh 1 command For example the following shutdown 1M command would bring the system configured with the ethernet hostname fred to init state 0 immediately rsh fred Sbin shutdown g0 y i0 By default clients configured in Embedded mode do not require an orderly shutdown but an application may initiate it 2 55 Power Hawk Series 900 Diskless Systems Administrator s Guide 2 36 3 Flash Boot System Administration Introduction Ev bee eee e els 1 1 3 2 Flash Characteristics eee 1 2 3 3 The Flash Filesystem 1 3 3 4 The Raw Write Method 1 5 Power Hawk Series 900 Diskless System Administrator s Guide 3 Flash Boot System Administration 3 1 Introduction Flash is a mostly read memory mapped device that behaves like normal memory at least for reads and wh
97. p during bootup in kdb via the mknetbstrap b kdb option See the mknetbstrap 1M system manual page for more details on these kdb related options When kdb is configured into a client s kernel the k sequence will cause the system to drop into kdb The crash 1M utility command may be used to examine the system memory image of a running system by internally formatting and then displaying various control structures tables and other information The crash 1M utility may be run directly on the client system to examine the current state of the running system The user may be logged into the system through the attached console terminal or through a networked terminal connection such as telnet or rlogin A Adding a Local Disk By default clients are configured as diskless systems It may be desirable to connect a local disk drive which is used to store application specific data The following example demonstrates how to configure a disk assuming that the disk has been formatted file systems have been created on the appropriate partitions and the disk has been connected to the client Refer to the System Administration manual Volume 2 for guidelines on how to accomplish these pre requisite steps The kernel configuration may be modified using the config 1M tool and specifying the client s virtual root directory For example if the client s virtual root path is vroots elroy config r vroots elroy If necessary add an entr
98. page 2 15 for more details By default when run in create mode C option netbootconfig 1M performs the following tasks Populates a client private virtual root directory Modifies client private configuration files in the virtual root Creates the virtual rootpath client profile Modifies the d stab 4C table and executes the sharea11 1M com mand to give the client permission to access via NFS its virtual root direc tory and system files that reside on the File Server Creates the client private custom directory etc clients client hostname gt net custom conf where the client hostname is equal to the name of the client profile file For example if a client s client profile filename is Ered then the client s custom directory would be etc clients fred net custom conf By default when run in remove mode R option netbootconfig 1M performs the following tasks Netboot System Administration Removes the virtual root directory Removes client s name from the d stab 4C tables and executes an unshare 1M of the virtual root directory Removes the client private log file etc profiles client hostname gt 1log Removes the client private custom directory etc clients client hostname gt net custom conf The update option U indicates that the client s environment already exists and by default nothing is done The task to be performed must be indicated by speci
99. path tmp tmp virtual rootpath users users etc shared etc var shared var dev shared dev The etc and var directories under the client s virtual root contain some files that are client specific therefore these files are not shared Each client has its own unique version of these files These directories also contain some files which have the same content for the File Server and all client virtual roots on the File Server These files are shared between the File Server and all of the File Server s clients Because the etc and var directories contain a mix of both shared and non shared files these directories require special han dling within the client s virtual root The files etc nodename not shared and etc chroot shared will be used to illustrate how shared and non shared files are handled in etc and var Introduction The etc nodename file is simply created as a real file in the client s virtual root under the virtual rootpath etc directory The virtual rootpath etc directory is mounted on the diskless client under the etc directory The etc chroot file is created in the client s virtual root not as a real file but as a symbolic link to the file name shared etc chroot On the File Server there is no such directory as shared On the client system shared is used as the mount point for mounting the File Server s actual etc directory Thus any reference on the diskless client to etc chroot wil
100. phabetically by package name Package MMA Package Dependencies Narje Package Description See Note base Base System Release 6 0 or later cmds Advanced Commands Ip nsu gte GT64260 Ethernet Driver 910 920 nsu mve MV64460 Ethernet Driver 940 nsu sym Symbios 53C885 Fast Ethernet Driver nsu dfs Distributed File System Utilities inet inet Internet Utilities nsu 1 Printer Support ncr Internal NCR SCSI Driver netcmds Commands Networking Extension Ip inet nfs Network File System Utilities nsu inet rpc dfs nsu Network Support Utilities rpc Remote Procedure Call Utilities inet Note All packages are dependent on base package Introduction 1 6 Licensing Information The system installed on the File Server carries a license for the number of processors allowed to be booted The license also carries a limit for the number of users allowed to log on to the File Server diskless client SBCs are limited to a maximum of 2 users each To print the processor and user limits set for your machine use the g option of the keyadm 1M command Power Hawk Series 900 Diskless Systems Administrator s Guide 1 20 2 1 2 2 2 3 2 4 2 5 2 6 2 Netboot System Administration Configuration ee 1 1 2 1 1 Installing a Loosely Coupled 1 1 2 1 2 Installing Additional Boards
101. profile parameters As an example to modify most parameters in netboot client wilma s client profile file take the following steps to remove modify and re create and client s configuration 1 Remove the current client configuration for wilma but preserve any client customization files for client wilma netbootconfig R s wilma 2 Edit wilma s client profile file as needed vi etc profiles wilma Netboot System Administration 3 Re create the configuration for client wilma netbootconfig C wilma The client profile file parameters that MAY be modified without the need to remove and reconfigure the client are described below These parameters only apply to NFS netboot clients For these parameters the actual client profile parameter value within the client s profile file are NOT modified only the actual objects that these parameters act upon are modified AUTOBOOT This parameter is implemented as a hidden file named autoboot directly under the client s virtual root directory This hidden file may be created and removed to enable or disable respectively the automatic shutdown of the client when the File Server shuts down In this case it is not necessary to modify the actual client profile file in order to modify this setting See the section The Client Profile File on page 2 13 for more details on the AUTOBOOT parameter SWAP SIZE For NFS clients a different sized dev swap file file from the one specified in the
102. ram is a small executable that will be booted by ASTRix via the boot 5 command This flashboot program will jump from memory to the start of the Flash executable to begin execution of the client Flash boot image To store the flashboot program into NVRAM cd nvram cr tftp g r flashboot SHOST cr 7 Change the boot command configuration slot number 5 so that it boots the flashboot program boot 5 n nvram flashboot s c 8 If you wish to only manually Flash boot then remove the startup script rm nvram startup and proceed to step 8 Flash Boot System Administration If you wish to autoboot from Flash then create a new startup script rm nvram startup vi nvram startup Add the following lines to the file using vi echo About to Flash boot the client sleep 3 boot 5 Use wq lt gt to write the file and exit vi The sleep 3 line above will provide time for you to abort out of the Flash autoboot sequence with a Ctrl c keyboard sequence should you need to make any configuration changes to the client To test the Flash boot configuration with autoboot configured issue the ASTRix reboot command reboot cr The board should reset and enter STAR STAR should then autoboot ASTRix and ASTRix should execute the nvram startup script which will boot the client from Flash If there is a problem with the Flash autoboot configuration then you may abort the ASTRix startup sc
103. rds SCSI devices Glossary 10 System disk system initialization System Run Level swap space target TELNET TCP Trivial File Transfer Protocol TFTP Glossary The PowerMAX OS requires a number of system directories to be available on a system disk in order for the operation system to function properly These directories include etc sbin dev usr and var The routines from the driver code and the information from the configuration files that initialize the system including device drivers A netboot system is not fully functional until the files residing on the File Server are accessible init 1M init state 3 is the initdefault and the only run level supported for netboot systems In init state 3 remote file sharing processes and daemons are started Setting initdefault to any other state or changing the run level after the system is up and running is not supported Swap reservation space referred to as virtual swap space is made up of the number of real memory pages that may be used for user space translations plus the amount of secondary storage disk swap space available See definition for client The Network Terminal Protocol provides remote login over the network Transmission Control Protocol provides reliable data delivery service with end to end error detection and correction Internet standard protocol for file transfer with minimal capability and minimal overhead TFTP depends o
104. re this custom file 1 Make the changes to the shared file type is either nfs or emb vi etc clients cluster conf custom conf type lt customfile gt 2 For each client to share these changes a Verify that the custom file is symbolically linked to the file edited above ad etc clients lt client gt net custom conf 1s 1 lt customfile gt b If the file is not currently symbolically linked then re link it mkshared lt customfile gt c Verify that the file is now symbolically linked 15 1 lt customfile gt For example to make private changes to the KOOclient script for a netboot client named wilma cd etc clients wilma net custom conf mkprivate KOOclient vi KOOclient Changes to the customization files are processed the next time the boot image generating utility mknetbstrap 1M is invoked If mknetbstrap 1M finds that a customization file is out of date compared to a file or boot image component it will implement the changes indicated If applicable some changes do not affect the boot image the boot image component will be rebuilt and a new boot image will be generated The customization files are described below in terms of their functionality 2 22 Netboot System Administration 2 5 2 1 S25client and KOOclient rc Scripts Commands added to these rc scripts will be executed during system initialization and shutdown The scripts must be written in the Bourne Shell sh 1
105. refix protocol RFS random I O Glossarv 8 Network Information Service formerlv called vellow pages or vp NIS is an administra tive svstem It provides central control and automatic dissemination of important adminis trative files Non Volatile Random Access Memorv This tvpe of memorv retains its state even after power is removed The state where an unrecoverable error has occurred Usuallv when a panic occurs a mes sage is displaved on the console to indicate the cause of the problem Protocol Data Unit The PowerPC G4 7450 microprocessor Part of the PowerPC family of microprocessors an architecture based on Motorola IBM s 32 bit RISC design CPU core Point to Point protocol is a method for transmitting datagrams over point to point serial links A character name that uniquely identifies a driver s routines to the kernel The prefix name starts each routine in a driver For example a RAM disk might be given the ramd prefix If it is a block driver the routines are ramdopen ramdclose ramdsize ramdstrategy and ramdprint Rules as they pertain to data communications Remote File Sharing I O operations to the same file that specify absolute file offsets raw I O raw mode rcp read queue rlogin routines rsh SBC Glossary Movement of data directly between user address spaces and the device Raw I O is used primarily for administrative functions where the speed of a specific operation is
106. ript with a Ctrl c keyboard sequence and make any required changes If the board is configured to manually boot the system without using an ASTRix startup script then test the Flash boot configuration with the ASTRix boot command boot 5 cr ASTRix should load and boot the client boot image from Flash 3 7 Power Hawk Series 900 Diskless Systems Administrator s Guide 4 Debugging Tools 4 1 System Debugging Tools 1 1 4D REC e n o t t aet e o des 1 1 4 3 eras iiie vL C READS ERR eU DUREE RU NONE ERAT 1 2 Power Hawk Series 900 Diskless System Administrator s Guide 4 Debugging Tools 4 1 System Debugging Tools 4 2 kdb This chapter covers the tools available for system debugging on a diskless client Tools covered in this include the following kdb crash In the loosely coupled architecture the only attachment between the file server and the diskless client is via an ethernet network connection There is no way to remotely access a diskless system s memory in a loosely coupled configuration A client is referred to as a netboot client and is configured via a netboot client profile file in the etc profile directory For more information on netboot clients see Chapter 2 Netboot System Administration The state of a diskless system may be examined as follows a Enter kdb by typing a k sequence on the client s console The client s boot image must have been built with kdb support b
107. s an SBC with attached disks where the boot image and a virtual root par tition for each configured diskless system is created The virtual root is both the environ ment used to build the boot image and it is also mounted by diskless systems that maintain an NFS connection to the File Server Embedded diskless configurations do not maintain such an NFS connection When the virtual root is mounted by the diskless system it is used to hold system commands and utilities as well as user defined files and application programs The virtual root can be viewed as a resource for additional disk space for a diskless system Once a boot image is created it must be copied from the File Server to the diskless system There are two supported mechanisms for transferring a boot image to a diskless system 1 A diskless system that is configured to boot from the network will read the boot image via an ethernet network connection to the File Server The firmware uses the Trivial File Transfer Protocol TFTP over an ethernet connection to download the boot image 2 The boot image may have already been burned into flash ROM In this case the board s firmware ASTRix is configured to execute the boot image from flash ROM Closely related to the technique for copying a boot image to a diskless SBC is the tech nique for initiating the boot sequence on the diskless SBC There are two techniques for initiating the boot sequence on a diskless system In some cases
108. s data character by character between the device and the user pro gram Character drivers are usually written for use with terminals printers and network devices although block devices such as tapes and disks also support character access The process of reading and writing to from a terminal A SBC board usually without a disk running a stripped down version of PowerMAX OS and dedicated to running a single set of applications Called a client since the client may maintain an Ethernet connection to its File Server and use that File Server as a type of remote disk device utilizing it to fetch applications data and to swap unused needed pages to from memory The circuit board that connects a device such as a terminal or disk drive to a computer A controller converts software commands from a driver into hardware commands that the device understands For example on a disk drive the controller accepts a request to read a file and converts the request into hardware commands to have the reading apparatus move to the precise location and send the information until a delimiter is reached cyclic redundandancy check CRC datagram data structure A way to check the transfer of information over a channel When the message is received the computer calculates the remainder and checks it against the transmitted remainder Transmission unit at the IP level The memory storage area that holds data types such as integers and strings or a
109. s digital signals from the computer into tones that can be transmitted across phone lines A demodulator converts the tones received from the phone lines into digital signals so that the computer can pro cess the data The process of a client SBC downloading into its own memory and then executing a boot image file that is retrieved from a File Server SBC by using the TFTP network protocol On client SBC boards networking is configured with the STAR config command and an ASTRix nvram startup script may be created and automatically executed after a reset in order to download and execute a client boot image via TFTP with ASTRix tftp and boot commands The process of a target loading a boot image as discussed under netboot but without subsequently executing it On Power Hawk Series 900 client boards netloading is invoked with the ASTRix tftp command See definition for netboot See definition for netload The netstat command displays the contents of various network related data structures in various formats depending on the options selected Network File System This protocol allows files to be shared by various hosts on the net work In a NFS client configuration the host system provides UNIX file systems for the client system A client system operates as a diskless NFS client of a host system Glossary 7 Power Hawk Series 900 Diskless Systems Administrator s Guide NIS NVRAM panic PDU PowerPC G4 PPP p
110. s system configuration The files used for client customizing include Netboot System Administration K00client to execute commands during system start up S25client to execute commands during system shutdown memfs inittab to modify system initialization and shutdown inittab to modify system initialization and shutdown nfs clients only vfstab to automatically mount file systems nfs clients only kernel modlist add to configure additional modules into the unix kernel memfs files add to add files to the memfs root file system vroot files add to make a copy of specific non system files in the client s virtual root directory nfs clients only When a client is configured using netbootconfig 1M a directory is created specifically for that client under the etc clients directory The client s custom configuration files are installed under this client s custom conf directory etc clients client dirs custom conf and are initially linked to the files in the cluster s custom conf directory usr etc diskless d cluster conf custom conf nfs emb When the client is a netboot client then the name of the client dir will be of the format client profile filename gt net So for example if the client profile file named fred is for a netbooted client then the corresponding private client directory name will be etc clients fred net custom conf The files in these client private directories are initially shared wit
111. s that provide a basic I O system BIOS a boot Flash and system diagnostics for Power Hawk Series 900 single board computers ASTRix startup script As part of the boot process ASTRix can automatically perform ASTRix commands and or user defined functions written in a startup script that is stored in NVRAM nonvolatile RAM A special startup script is used for netbooting client SBCs in loosely coupled con figurations SMTP The Simple Mail Transfer Protocol delivers electronic mail small computer system interface SCSI The American National Standards Institute ANSI approved interface for supporting specific peripheral devices SNMP Simple Network Management Protocol Source Code Control System SCCS A utility for tracking maintaining and controlling access to source code files special device file The file that identifies the device s access type block or character the external major and minor numbers of the device the device name used by user level programs and security control owner group and access permissions for the device synchronous data link interface SDLI A UN type circuit board that works subordinately to the input output accelerator IOA The SDLI provides up to eight ports for full duplex synchronous data communication system A single board computer running its own copy of the operating system including all resources directly controlled by the operating system for example I O boa
112. sh if available as well as the Boot Flash area The main disadvantage of using the Flash file system method is that unlike the Raw Write method the entire client boot image is loaded into memory and uncompressed The Flash Boot System Administration entire memory resident root filesystem remains in memory after booting This attribute of Flash filesystem booting is thus the same as network booting The main advantage of using the Raw Write method is that even though the real physical Flash bank size is 32 MB the boot image is conceptually divided into a series of 128 KB sized virtual banks or pages by the PowerMAX OS kernel These virtual Flash pages are copied into memory only on an as needed basis and the Flash backed memory area is freed when it is no longer in use This Flash paging support provides for better utilization of client memory as the root file system continues to reside in Flash after booting thus freeing up the memory that would otherwise have been used to implement a memory resident root filesystem The main disadvantage of using the Raw Write method is that when the Raw Write method is used the boot image object must be written to a specific location within the Boot Flash Therefore only one boot image may be stored in Flash at any point in time The following sections discuss how to setup a Flash boot image using either the Flash file system or the Raw Write methods The user is free to choose either meth
113. sh Boot booting Glossary 4 burning Glossary 5 characteristics 3 2 file system 3 2 3 3 memory Glossary 5 overview 3 1 Raw Write 3 2 3 5 Flash Boot definition 1 5 description 1 9 1 11 system administration 3 1 hardware overview 1 6 host Glossary 5 hosts tables 2 15 idtuneobj 2 20 init level 1 5 Glossary 11 inittab 1 12 2 21 2 23 K KOOclient rc script 2 21 2 23 Index 2 kdb 4 1 kernel configuration 2 18 kernel modlist add 2 19 2 21 2 25 keyadm 1 19 launching applications embedded clients 2 29 NFS clients 2 30 licensing information 1 19 local disk 1 Looselv Coupled Svstem LCS configuration l 2 definition l 4 disk space l 18 hardware prerequisites 1 17 installation 2 1 overview 1 1 software prerequisites 1 18 memfs file system 1 8 1 12 memfs cpio 1 7 1 8 memfs files add 2 21 2 25 2 29 memfs inittab 2 21 2 23 mknetbstrap 2 19 2 28 2 31 2 33 4 2 N Net Boot configuration client 2 13 modify 2 12 overview 2 1 definition 1 5 description 1 9 1 10 procedure 2 33 system administration 2 1 toolset 1 3 using ASTRix 2 34 netboot Glossary 7 netbootconfig 1 4 2 13 2 16 2 17 netload Glossary 7 network boot Glossary 7 network load Glossary 7 networking structure 1 13 NFS client configuration 2 14 definition 1 4 Glossary 7 launching an application 2 30 make NFS server B 1 overview 1 2 NFS mount scheme 1 12 nfssrv B 1 paged flash 3 2 Point to Point protocol Glossarv 8 Power
114. sk s i b maj minor sbin mknod dev rdsk disk s i c maj minor Adding a Local Disk fsck ufs only and mount each partitions if SFSTYPE xfs then echo gt Fsck ing partition dev rdsk disk s i etc fs SFSTYPE fsck y dev rdsk disk s i X dev null fi create a mount point directory usr bin mkdir disk s i mount the partition echo sMounting dev dsk disk ss i disk s i n etc fs ESTVPE mount dev dsk disk s i disk s i done break disk sbin expr disk 1 done verify the partitions are mounted echo gt Verifying mounted file systems sbin df kl A 5 Power Hawk Series 900 Diskless Systems Administrator s Guide B Make Client System Run in NFS File Server Mode To become an NFS server a client system must have an attached local disk In becoming an NFS server the client can share the data in the local disk partitions with other systems that have network access to the client See Appendix A Adding a Local Disk for instructions on how to add a local disk to a diskless client We will refer to the client with the local disk as the disk server and the system s that want to access this client disk as disk clients Note It is important that the hostnames that are used for the disk server and the disk clients correspond to network interfaces that are accessible to both the disk clients and disk
115. stra tor s Guide 0891089 for details that are specific to setting up Power Hawk Series 700 systems in LCS configurations that may or may not include Series 900 systems 1 1 1 Loosely Coupled Systems LCS The LCS configuration see Figure 1 1 is supported when the only attachment between the fileserver and the diskless system is from an ethernet network Inter process commu nication between processes running on separate single board computers is limited to standard networking protocols across ethernet There are two possible ways of configuring a diskless client system The difference between these client configurations is whether the client system maintains an NFS connection to the fileserver after boot such that file system space is available for the client system on the File Server It is important to note that the type of client system configuration selected will impact the resource requirements of the File Server as will be explained in more detail later The two client configurations are Embedded client Embedded clients are either stand alone systems which have no attach ments to other SBCs or they are not configured with networking and therefore do not use existing network attachments once the system is up and running The embedded applica tions must be a part of the original boot image which is downloaded onto the client system and those applications begin execution at the end of the boot sequence 1 1 Power Hawk Series 90
116. sue the following ASTRix commands on the client cd nvram cr rm startup cr ignore any errors if this file didn t exist vi startup cr Then add the following lines to the startup file using the vi editor the com mented out route lines are not always necessary see the comments below CLIENT route add net XXX XXX XXX XXX gW XXX XXX XXX XXX V netmask 255 255 xxx xxx dev ethO while 1 do echo tftp g r SCLIENT bstrap SHOST tftp g r SCLIENT bstrap HOST case in 0 break sleep 5 echo retrying esac done echo Booting client CLIENT netboot image sync boot 5 Modify the CLIENT line above using vi so that CLIENT is set to the hostname of this client For example if wilma is the hostname of this cli ent then change this line to CLIENT wilma Enter wq in vi to write the nvram startup file and exit the vi editor The two route lines in the above startup script above should only be used when the client accesses the file server through a gateway For example if the file server s network is 129 148 42 0 and it is accessed locally through the gateway 129 148 43 196 with the client residing on the 129 148 43 0 network then the above routing lines in the startup script should be uncommented and modified to be route add net 129 148 42 0 gw 129 148 43 196 netmask 255 255 255 0 dev ethO If the client and file server reside on the same network then
117. t the client is connected to and that does not conflict with any other IP addresses used in the network Introduction If the client SBC s access to the File Server is through a gateway then a routing table entry must be setup with the ASTRix route command Usually this would be done in the ASTRIx startup script so that this setup would be automatically executed every time ASTRix was initiated following a reset or power cycle After STAR has been initialized with the proper network parameters and any required routing table entry has been added with the ASTRix route command a transfer of the boot image across the Ethernet connection may be initiated via the ASTRix tftp com mand Refer to the chapter on Netboot System Administration for more information on Net Booting 1 4 4 2 Flash Boot Flash Boot is a method of loading and executing a kernel image from flash Flash Boot is the preferred method of booting a diskless client in the production or deployed phase of an application There are two advantages to booting from flash Hlash Boot provides very fast boot times because there are no rotational delays that would normally be associated with reading from a local system disk and no networking delays that would normally be associated with downloading a bootable image from a file server when using the Net Boot method The root file system resides as a read only image in flash thus providing greater system stability because the root
118. t be rebooted using the new boot image in order for these changes to take effect Like the other customization files these tables may be updated in one of the two subdirectories nfs or emb Changes made under the usr etc diskless d custom conf client shared directory apply globally to all NFS or embedded clients that share this File Server If the changes are specific to a particular client then a private copy of the shared file should first be created in the client s private customization directory by using the mkprivate tool and then edited in that client S custom conf directory 2 5 2 5 memfs files add Table When the mknetbstrap 1M utility builds a boot image it utilizes several files for building the compressed cpio file system The set of files included in the basic diskless memory based file system are listed in the files devlist nfs netboot and filelist nfs netboot for NFS clients and devlist emb netboot and filelist emb netboot for embedded clients under the usr etc diskless d sys conf memfs d directory Additional files may be added to the memory based file system via the memfs files add table located under the usr etc diskless d cluster conf custom conf directory Like the other customization files this tables may be updated in one of the two subdirectories nfs or emb Changes made under the usr etc diskless d cluster conf custom conf directory apply globally to all nfs or embedded clients that share this File Serv
119. t point directory entries to the memfs root file system via the memfs files add custom file For example to add the directories arbitrarily named dsk0s0 and dsk0s1 etc clients lt client dir custom con mkprivatememfs files add vi memfs files add Example entries d dsk0s0 0777 d dsk0s1 0777 5 Enable Automatic Mounting of a Disk Partition by adding entries to the client s vfstab file Note that the mount point directory name must match the directory name specified in the memfs files add file in step 4 above cd etc clients client dir gt custom conf mkprivate vfstab vivfstab Example entries dev dsk 0s0 dev rdsk 0sO0 dsk0s0 ufs 1 yes dev dsk 0s1 dev rdsk 0s1 askosl ufs 1 yes 6 Generate a new netboot image mknetbstrap r all client Embedded clients steps 3 5 3 The disk management tools must be added to the memfs file system The list of tools is documented in the file usr etc diskless d sys conf memfs d add disk sh Inexecuting the following commands we grep the HH db db db db Adding a Local Disk list of commands from this file and append them to the mem s files add tables 4 cd etc clients client dir gt custom conf mkprivatememfs files add sbin grep ff usr etc diskless d svs conf X memfs d add disk sh cut c2 memfs files add Verify that the following entries were appended to memfs files add sbin expr 0755
120. terface available as special order only An event occurring in an unpredictable fashion A signal is an example of an asynchro nous event signal can occur when something in the system fails but it is not known when the failure will occur asynchronous I O operation An I O operation that does not of itself cause the caller to be blocked from further use of the CPU This implies that the caller and the I O operation may be running concurrently asynchronous I O completion block data transfer block device An asynchronous read or write operation is completed when a corresponding synchronous read or write would have completed and any associated status fields have been updated The method of transferring data in units blocks between a block device such as a magnetic tape drive or disk drive and a user program A device such as a magnetic tape drive or disk drive that conveys data in blocks through the buffer management code Compare character device Glossary 1 Power Hawk Series 900 Diskless Systems Administrator s Guide block driver block I O block boot boot device boot image file bootstrap buffer cache character device Glossary 2 A device driver such as for a magnetic tape device or disk drive that conveys data in blocks through the buffer management code for example the bu structure One driver is written for each major number employed by block devices A data transfer method used b
121. that these directories are available on one or more locally attached SCSI disk drives Since clients do not have locally attached system disk s they will NFS mount these directories from the File Server an NFS Client or create them in a memory file system which is loaded with the ker nel an Embedded Client A client server kernel boot method that uses standard TFTP protocols for kernel loading from the File Server Any client can be configured to initiate a net boot operation from the File Server A client boot method where the boot image executed comes from the client s own Flash memory This is the object that is downloaded into the memory of a diskless client It contains a UNIX kernel image and a memory based root file system The memory based file system must contain the utilities and files needed to boot the kernel In the case of an NFS client booting must proceed to the point that remote file systems can be mounted For an embedded kernel the memory based file system is the only file system space that is available on the diskless system Users may add their own files to the memory based file system A board resident ROM monitor utility that provides a basic I O system BIOS a boot ROM and system diagnostics for Power Hawk Series 900 single board computers SBCs Internet standard protocol for file transfer with minimal capability and minimal overhead TFTP depends on the connectionless datagram delivery s
122. tory may be generated in any file system partition on the file server except for those used for the root and var file systems Virtual roots are created on the host for all clients Clients running embedded systems will utilize their virtual root for configuring the clients s kernel and building the boot image 1 4 2 Boot Image Creation and Characteristics One of the primary functions of the virtual root is as the development environment for building the boot image that will be downloaded to diskless client systems After a client s virtual root development environment has been created users have the opportunity to tune the development environment in various ways including that of adding in their own applications and data The boot image file known as unix bstrap is composed primarily of two intermedi ate files unix and mem s cpio These are located in the same directory as unix bstrap unix is the client s kernel as built by idbuild 1M memfs cpio is a compressed cpio archive of all the files which are to be the contents of that client s memfs root filesystem This archive was compressed using the tool rac 1 Conversely if the user wants to examine the contents rac 1 must be used to decom press it The final boot image unix bstrap will contain a compressed version of the text and data regions of the unix kernel These were extracted from the unix file It will also con tain bootstrap code which decompresses the kernel an
123. trap2 3 cr Alternatively if you have User Flash available and wish to store the client boot image into User Flash instead of Boot Flash then use the ASTRix command ffsls cr to view the available Flash filesystem areas and to pick a Flash filesystem that resides in a User Flash area instead of Boot Flash filesystem area 3 So for example to use User Flash filesystem area 5 use the fol lowing ASTRix command to burn the boot image to User Flash ffsw lt client gt bstrap 5 5 Change the boot command s configuration slot number 5 so that it boots from the Flash boot image boot 5 n ffs3 curtis bstrap s c Or possibly enter something like boot 5 n ffs3 lt curtis gt bstrap2 s c if you are using a second boot image Alternatively if you are using User Flash to store the image then enter the Flash filesystem number that you used in step 4 For example the command boot 5 n ffs5 lt curtis gt bstrap s c will setup to boot the boot image from User Flash filesystem number 5 6 If you wish to only manually Flash boot then remove the startup script rm nvram startup and proceed to step 7 If you wish to autoboot from Flash then create a new startup script Flash Boot System Administration rm nvram startup vi nvram startup and add the following lines to the file echo About to Flash boot the client sleep 3 boot 5 Use wq lt gt to write the file and exit vi
124. uring system startup to ward off the network congestion that occurs when many clients simultaneously download their boot images from a common File Server 3 2 Flash Characteristics The Flash on Power Hawk Series 900 boards is what is called a paged flash or a banked flash Flash is accessed through two 32 MB windows one for User Flash and one for Boot Flash where the Boot Flash is always accessible A particular single bank of 32 MB User Flash may be selected for access through the 32 MB User Flash window where this selec tion is accomplished through the ASTRix fbs User Flash bank select command PowerMAX OS supports two methods for storing and booting a client boot image in Flash the Raw Write method and the Flash file system method The Raw Write method uses the ASTRix fw Flash write command to write the client boot image to Flash in a direct raw write manner The Flash filesystem method uses the ASTRix ffsw Flash filesystem write command to write the client boot image into a Flash file system area The format of this area is an ASTRix supported Flash filesystem that may contain multiple files The advantages of using the Flash filesystem method are The ASTRix Flash filesystem support is used for storing the client boot image As such multiple boot images may be stored into the same Flash filesystem and queried with the ffsls Flash filesystem Is command The user may store client boot images into User Fla
125. wk Series 900 Diskless Systems Administrator s Guide memfs inittab file is modified a new mem s cpio image and a new boot image will be created the next time mknetbstrap 1M is invoked client must be rebooted using the new boot image in order for any changes to take effect Any programs to be initiated on an embedded client must be specified to run at init level 1 NFS clients may use the memfs inittab table for starting programs at init levels 1 3 However part of the standard commands executed at init level 3 onan NFS client is the mounting of NFS remote disk partitions At this time an NFS client will mount its virtual root The memfs based etc directory is used as the mount point for the virtual root etoc directory that resides on the File Server This causes the mem s inittab table to be replaced by the inittab file This means that any commands to be executed in init state 0 system shutdown or commands which are to be respawned in init state 3 should be added to both the memfs inittab and the inittab file if they are to be effective After configuring an NFS client system the inittab table contains entries that are needed for the basic operation of a diskless system configuration The default entries created by the configuration utilities in the inittab file should not be removed or modified Changes to inittab are processed the next time mknetbstrap 1M is invoked The inittab table is copied into the client s virtual
126. work is being accomplished Therefore caution is advised when using the DISSWAPRES tunable The DISSWAPRES tunable may be useful when a fixed set of applications and their corresponding virtual address space working sets are known to fit into the amount of available real memory and secondary storage swap space if any even though their total virtual swap space requirements exceed the system s virtual swap space capacity Introduction 1 5 Configuring Diskless Systems 1 5 1 Loosely Coupled System Hardware Prerequisites A loosely coupled configuration requires the following hardware At least one VME card chassis One Power Hawk Series 900 or Series 700 single board computer witha minimum of 128 MB of DRAM for use as the server SBC One Power Hawk Series 900 or Series 700 single board computer with a minimum of 128 MB of DRAM for each client SBC One SCSI 2 GB 4 GB or higher is preferred disk drive for PowerMAX OS software installation connected to the Synergy PSCx SCSI PMC interface card on the server SBC One supported SCSI CD ROM device connected to the PSCx SCSI PMC interface card for installation of system software on the server SBC At least one system console terminal which may be a video display terminal such as a Wyse 150 vt100 or comparable device connected to Serial Port on the server SBC Additional system console terminals may be attached to any client SBC s Serial Port A for debug purposes
127. y drivers for block access devices Block I O uses the sys tem buffer cache as an intermediate data storage area between user memory and the device The basic unit of data for I O access A block is measured in bytes The size of a block dif fers between computers file system sizes or devices The process of starting the operating system The boot process consists of self configura tion and system initialization The device that stores the self configuration and system initialization code and necessary file systems to start the operating system A file that can be downloaded to and executed on a client SBC Usually contains an operating system and root filesystem contents plus all bootstrap code necessary to start it The process of bringing up the operating system by its own action The first few instruc tions load the rest of the operating system into the computer A staging area for input output I O processes where arbitrary length transactions are col lected into convenient units for system operations buffer consists of two parts a mem ory array that contains data from the disk and a buffer header that identifies the buffer A section of computer memory where the most recently used buffers i nodes pages and so on are stored for quick access A device such as a terminal or printer that conveys data character by character character driver character I O client controller Glossary The driver that convey
128. y to the adapters table Adapter information must be added to the adapters table for VME adapters i e via PCI adapters i e ncr are auto configurable and should not be added to the adapters table If this is a VME adapter add an entry for it in the adapters table using the Adapters Add menu option of config 1m 2 Configure kernel modules Use the Modules function of the config 1M tool to enable the follow ing modules gd generic disk driver scsi device independent SCSI interface support ncr internal SCSI adapter interface driver ufs unix file system sfs unix secure file system If a Resilient File System XFS is required for a client instead of enabling ufs and Sfs enable xfs resilient file system xfsth resilient file system threaded Note that the kernel modlist add table in the client s custom conf directory etc clients client dir gt custom conf may instead be used to enable kernel modules 1 Power Hawk Series 900 Diskless Systems Administrator s Guide Note The procedural steps below differ depending whether the client was configured with NFS support or as embedded NFS Clients steps 3 6 3 Configure Disk Device Files Check that an appropriate device node entry Node 4 exists and is uncommented for the disk being added The following is such an entry from the Node file vroots elroy etc conf node d gd gd dsk O D ner 0 0 0640 2 4 Add moun

Power HawkTM Series 900 Diskless Systems

Contents

Download Pdf Manuals

Related Search

Related Contents