Home

Shelf Manager User Guide

1. OEM Proprietary Destination Address VLAN TAGs N A 9 Change the IP settings for channel 1 clia setlanconfig 1 ip 10 1 1 10 D 03 05 07 09 OB OD Pigeon Point Shelf Manager Command Line Interpreter IP set successfully Clia setlanconfig 1 subnet_mask 255 255 0 0 Pigeon Point Shelf Manager Command Line Interpreter Subnet Mask set successfully clia setlanconfig 1 dft_gw_ip 10 1 1 1 Pigeon Point Shelf Manager Command Line Interpreter Default Gateway Address set successfully Priv Level OEM Priv Level OEM Priv Level Priv Level OEM Priv Level Priv Level OEM 10 If the second network interface is used for RMCP get the current IP settings for channel 2 clia getlanconfig 2 Pigeon Point Shelf Manager Command Line Interpreter 11 Change the IP settings for channel 2 clia setlanconfig 2 ip 10 2 1 10 Pigeon Point Shelf Manager Command Line Interpreter IP set successfully clia setlanconfig 2 subnet_mask 255 255 0 0 Pigeon Point Shelf Manager Command Line Interpreter Release 3 2 0 90 July 31 2013 Pigeon Point Shelf Manager User Guide Subnet Mask set successfully clia setlanconfig 2 dft_gw_ip 10 2 1 1 Pigeon Point Shelf Manager Command Line Interpreter Default Gateway Address set successfully 3 4 6 Assigning VLAN IDs Starting from release 2 6 0 the Shelf Manager supports Virt
2. Release 3 2 0 219 July 31 2013 Pigeon Point Shelf Manager User Guide 8 Re programming the ShMM 700 This section describes how to update the firmware on the ShMM 700 using the reliable upgrade facilities Comparisons and contrasts with the ShMM 500 ShMM 1500 style of firmware upgrades are included but this section is also intended to be usable for readers without prior background on the ShMM 500 ShMM 1500 firmware upgrade facilities 8 1 In This Section This section contains the topics listed below J ust click on a topic to go to it Firmware Reliable Upgrade Procedure Overview Flash Partitioning and Image Layout Reliable Upgrade Procedure Status File Reliable Upgrade Utility Reliable Upgrade of A2F 060 Firmware Reliable Upgrade Utility Use Scenarios Reliable Upgrade Examples 8 2 Firmware Reliable Upgrade Procedure Overview For the ShMM 700 a reliable upgrade procedure for the firmware images on a running and functioning ShMM is provided An arbitrary combination of the following four software images can be upgraded reliably the U Boot firmware the Linux kernel the Linux root file system RFS and the application package If an attempted reliable upgrade fails an automatic rollback to the previous version of the firmware takes place An upgrade failure can be caused for example by installation of a faulty kernel image that cannot properly boot up the ShMM or a faulty Shelf Manager application that fails to sta
3. cccecccsecsesteeeseetseeeeees 82 3 4 6 ASSIGNING VLAN ADS civic 2opiceys egasscccts sivees souicicessuaedaceussee apshst pe teeaSeb ea SatbeevesSoo aes 91 3 4 7 Assigning IP Addresses to the Shelf Manager via DHCP sscccccssssseceesseeeees 92 3 5 CONFIGURING THE FRU INFORMATION ccceeeeeeeeeeeeseeceeeeeceaeeeeaaeeeeeeeseeeeseaeeeeaeeeeneeee 95 3 5 1 Accessing the Shelf FRU Information 2 cccccccecceecceeseeceeneeceeeeesnseeeeeeeseeeesseneteas 95 3 5 2 Setting up the Shelf FRU InfOrrmation 1 21 cccccceceececsseceeneeceeececsueseeneeseeseetseneteas 97 3 5 3 Setting up the Shelf FRU Information Using the CLI ccccccscccsssscccesssteeseennnees 99 3 5 4 Other FRU Information Repositories 2 ccccccceecceesseceeeeeceeecessneeeeeceeneetseneetsaes 100 3 6 CONFIGURING CARRIER AND SHELF ATTRIBUTES USING HPDL 0 ceeeeeereeereeereeeeeees 101 3 6 1 Compiling HPDL Definitions 0 0 eccccecccecscececececeneeeeneceeneecenececeeeesaseseeseecenneessanens 101 3 6 2 Compiling SDRS iss esthetic 102 3 6 3 Deploying HPDL Data to the SAMM File SYSte M 2 ccccsccccseceeesteceeeeeesnneetsaes 102 3 6 4 Deploying HPDL Data to FRU Information Areas s 103 3 7 CONFIGURING THE COOLING MANAGEMENT STRATEGY sccsceseceseceneeeseeeeresereeeneeeeeeses 106 3 7 1 Default Cooling Management Strategy ccccccccesseceeseceeeeesssneseneeseeeeecsnneeseaes 107 3 7 2 Configuring a Specific Cooling Management St
4. A configuration parameter for the signal Channels in the form Reset lt n gt specifies the number of the ShMM GPIO used to reset the device This GPIO is chosen by the designers of the ShMM carrier For distinct multiplexers different GP1Os must be used Signals with internal names in the form Channel lt n gt _Isolated report the isolation state of a specific channel of the multiplexer or switch starting from 1 The value of the signal is 1 if the channel is currently isolated by the isolation algorithm and 0 otherwise These signals can be exposed as IPMI sensors using normal HPDL mechanisms The signal IsolatedChannelMask reports the bitmask that has 1s for currently isolated channels bit 0 of the bitmask corresponds to the first channel bit 1 to the second channel etc In this mode the isolation algorithm operates as follows When the error ETIMEDOUT is detected by the Linux 12C subsystem while trying to access a device on a subordinate bus this bus is recorded as faulty in the Shelf Manager and the multiplexer switch is held in reset via the corresponding GPIO This operation causes all the channels to be deselected on the multiplexer switch which allows the Shelf Manager to access devices on the main master only 12C bus even if one of the subordinate buses has a fault The multiplexer switch channel that caused the error is noted and values of the signals Channel lt n gt _ Isolated and lsolat
5. b lt ip_address gt board lt ip_address gt Specify board trace mode in which IPMI messaging is traced on another IPM controller that is connected to the network The trace daemon obtains the trace using an RMCP session via a HPM 2 IPMI Trace Payload with the target IPMC The parameter lt ip_address gt specifies the IP address of the target IPMC m lt mask gt mask lt mask gt In board trace mode specify the bit mask of IP MI channels that are to be traced on the target IPMC This bit mask is represented by the numeric parameter lt mask gt with prefix 0x used to specify a hexadecimal number The following channel numbers can be used Release 3 2 0 171 July 31 2013 Pigeon Point Shelf Manager User Guide Channel 0 bit mask 1 IPMB 0 Channel 7 bit mask 0x80 IPMB L Channel 15 bit mask 0x8000 System Payload Interface The default value is 1 trace IPMB 0 only i lt number gt In board trace mode specify the payload instance instance lt number gt number to use for establishing the RMCP session This number must match the IPMI Trace Payload instance number used by the target IPMC for IPMB tracing refer to the board documentation for this number The default value is 1 U lt username gt In board trace mode specify the RMCP user name username lt username gt for establishing the RMCP session with the board The default user na
6. 0 cc ccceeeeeeeeeeeceaeeeeaeeeeeeeeseaeeesaeeeeneeeeaes 210 7 8 RELIABLE UPGRADE EXAMPLEG ccceeeeeeeeeeeeneeeeceeeeeeeeeeaaeeeseeaaeeeseeaaeeeseeaaeeeeeeaeeeeeeaaes 211 7 8 1 EXGIMDIO TD NA ENT TA E A A be jenna Peper tweeter dyadic eens 211 7 8 2 EXQIMDIO 2 A A EE E ENE privet Les E E ANE eves E E 214 7 8 3 e E NE OEE ENE E A A E A E A E 217 8 RE PROGRAMMING THE SHMM 700 ccsccseseeesseeeeseeeeeeeeessaeseseeeensneeeseeeeseaesesenenseees 220 8 1 IN THIS SECTHON rinsio kanaan aaa aaaea aea an taaa Eae ena abaa aaae 220 8 2 FIRMWARE RELIABLE UPGRADE PROCEDURE OVERVIEW cccecceeeeeeeeeeeeeeeeeseaeeeeneeeeaes 220 8 3 FLASH PARTITIONING AND IMAGE LAYOUT c ccceeeeeeeeeeeeeeeeeeeeaeeeseeaaeeeseeaaeeeeeeaeeeeeeaaes 222 8 4 RELIABLE UPGRADE PROCEDURE STATUS FILE esseseeseeereeresrserssrrssrnssriesrissrrnse 224 8 5 RECIABLE UPGRADE UTILITY i ceecceseteeatiategdnee ct csensceecceeece beacasda ediad Euer 225 8 6 RELIABLE UPGRADE OF A2FO60 FIRMWARE ec eeeeeeneeeeeeeaeeeeeenaeeeseeaaeeeeeeaeeeeeeaaes 231 8 7 RELIABLE UPGRADE UTILITY USE SCENARIOS 0 cc ceeeeeeceeeeeeeaeeeeeaeeeeeeeeeaeeesaeeneaeesaes 232 8 8 RELIABLE UPGRADE EXAMPLES cisscscssssdeecanesebeaenscee canvass Heaeesds ahaaa A 233 8 8 1 EEX EAITIOIO TT E cs ives cat oi aes putes cpa ad pak ped vag na snot E a egeeb en subeseees 233 8 8 2 EXIT E N OE E goes eda eS aco E veyed pan putives aces ibe Seivenee 240 8 8 3 EEX LITHO O css cosas fed
7. IPMB_RETRIES Number 3 The number of attempts to send an No IPMB request before finally giving up if no response is received to this request IPMB_RETRY_TIM Number 0 The amount of time in seconds that No ROUT the Shelf Manager waits for a response after sending an IP MB request before retrying the request IPMB_RETRY_TIM Number 500 The millisecond part of the retry No EQUT MSEC timeout value If the retry timeout is less than a second this configuration variable contains the actual timeout while the value of the configuration variable IPMB_RETRY_TIMEOUT is 0 Release 3 2 0 55 July 31 2013 IPMC_PRESERVE_ ON_REVISION_CH ANGE Pigeon Point Shelf Manager User Guide Boolean DEFAULT TRUE DESCRIPTION Setting this variable to TRUE preserves the Shelf Manager s identification of an IPM controller after a firmware upgrade if only the Firmware Revision and or Auxiliary Firmware Revision information is changed If the variable is set FALSE any change in Get Device ID response data during the Shelf Manager s regular polls is considered to signal the presence of a different IPM controller CAN BE OBTAINED FROM SHELF FRU INFO Yes ISOLATE_MUX_AD DRESS Number 0x70 The 7 bit 2C multiplexer address on platforms where SHMM_GPI08 is used to control access from the Shelf Manager to the multiplexer on the master only I2C bus via the original mechan
8. eth0 The name of the network adapter Yes ER used for RMCP based communication RMCP_NET_ADAPT String 16 None The name of the alternate network Yes BRA adapter used for RMCP based communications if cross connect links are supported by the hardware eth1 is the recommended value if the parameter is specified RMCP_WITHOUT_S Boolean FALSE RMCP is available in the absence of No ee the Shelf FRU only if this configuration parameter is defined and set to TRUE RUN_HPI_SNMP_S Boolean FALSE Specifies whether the HPI SNMP No UBAGENT subagent should be started automatically SDR_READ_RETRI Number 3 The maximum number of times the Yes ES Shelf Manager retries the Read Device SDR command Release 3 2 0 July 31 2013 Pigeon Point Shelf Manager User Guide DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO SEL_FILE_COMPR Boolean TRUE If TRUE the Shelf Manager No eee compresses the SEL before writing it to the ShMM Flash this reduces the size of the file that is periodically written to the Flash and thus increases the lifetime of the Flash SEL_FILE_JOURN Boolean TRUE If TRUE the Shelf Manager uses No ALING ENABLED journaling when writing the SEL to the ShMM Flash this reduces the size of the data periodically written to the Flash and thus increases the lifetime of the Flash The reduction is even greater if SEL compression is simultaneously enabled SEL_FILE_WRITE
9. lable upgrade initiation s1265 6 Write upgrade watchdo Last upgrade stages log O initial PASSED 1 before storage mount PASSED 2 after storage mount PASSED 3 after marking image set invalid PASSED 4 after u boot burn PASSED 5 after kernel burn PASSED 6 after rootfs burn PASSED 7 after storage umount PASSED 8 after cleaning previous apps set PASSED 9 after writting new apps set PASSED 10 after sync PASSED 11 after processing var and etc PASSED 12 after boot select PASSED 13 after reboot FAILED Release 3 2 0 249 g 0 gt 1 ALLOW July 31 2013 Pigeon Point Shelf Manager User Guide 14 after confirm FAILED The last stages are marked as FAILED in the reliable upgrade status The ShMM has automatically rolled back to the confirmed firmware because the upgrade failed Release 3 2 0 250 July 31 2013 Pigeon Point Shelf Manager User Guide 9 HPl based Shelf Manager Upgrade In addition to the reliable upgrade facility a Shelf Manager with IntegralHP also supports upgrades over the HPI interface HP defines a special type of management instruments used for upgrades and called Firmware Upgrade Management Instruments FUMIs IntegralHP creates a FUMI for each resource that represents a physical Shelf Manager In a redundant configuration there are two Shelf Manager related FUMIs one for the active and one for th
10. lt INFO gt shmm700_hal c 1285 Write storage state available ALLOW lt INFO gt shmm700_hal c 369 Resetting UBIFS cache lt INFO gt shmm700_hal c 1913 Clean on defice 1 1 dev mtd2 untouched lt INFO gt backend c 73 lt INFO gt backend c 76 tmp file lt INFO gt backend c 73 lt INFO gt backend c 76 tmp file tmp_s lt INFO gt backend c 73 lt INFO gt backend c 76 tmp file tmp_s lt INFO gt shmm700_hal available ALLOW lt INFO gt backend c 73 lt INFO gt backend c 76 tmp file tmp_s lt INFO gt backend c 80 lt INFO gt backend c 82 lt INFO gt shmm700_hal lt INFO gt shmm700_hal 0 Burning image 0 Burning image 4 1402348 bytes entry kernel to 0 Burning image 4 2283039 bytes was written dev 2 1 was written not available gt Leaving file 4 393216 bytes was written from file tmp_u boot bin to dev 1 1 from file from file gt not entry rfs to dev 4 1 c 1285 Write storage state available 0 Burning image 4 2110743 bytes was written from file entry app to dev 8 1 8 Erasing user data on candidate bank 1 Copy ssh keys c 1189 Write confirmed 0 gt c 1194 Write candidate 1 gt 1 AL LOW OW The log contains no confirm message the confirmed and candidate image set numbers are the Same they were before rel lt INFO gt shmm700_hal lt INFO gt frontend c 4
11. cccceesseceessseeceeseseeeeseneeeeeeaaes 200 Table 23 Flash Partitioning for 64MB Flash SAMM 1500S ccceesseceeseseeceeeeseeeeseneeeesenaes 201 Table 24 Standard Source File Names for Upgrade Components ccceeceeeeeeeeeeeeeeneeeeaes 205 Table 25 Flash Partitioning for SAMM 700s 64MB Flash eccceeeeeeeseeeeeeeeeeeeeeenaeeeeaeeteees 222 Table 26 File Names for Firmware Images Stored as UBIFS Files cccceccceeeeeeeeeeeeeneeteees 223 Table 27 Reliable Upgrade Stages cccccssccccssssceeseseececseseeeeesesaeeeseesaeeeeseaeesecesaeeessesaeeseseaaes 224 Table 28 Protocols Recognized in the URLs by the Reliable Upgrade Utility ee 227 Table 29 Standard Source File Names for Upgrade Components cccceecceeeeeessteeeeneeeeees 229 Release 3 2 0 8 July 31 2013 Pigeon Point Shelf Manager User Guide 1 About This Document This document describes the Pigeon Point Shelf Manager This PDF file requires Adobe Acrobat Reader 7 0 for full functionality For a free reader go to http www adobe com 1 1 Shelf Manager Documentation This document is one of two documents in the Shelf Manager documentation set These documents are available in P DF file format The complete set includes Table 1 Shelf Manager Documentation DOCUMENT DESCRIPTION Pigeon Point Shelf Manager This document describes the overall configuration User Guide and use of the Pigeon Point Shelf Manager Pigeon
12. cs Enforce checking Shelf FRU Info checksums If a checksum is invalid an error message is produced and the Shelf FRU Info is not used If a checksum is valid no message is produced If the option es is not present in the command line the value of the parameter VERIFY_SHELF_FRU_CHECKSUM from the file etc shelfman conf determines Shelf Manager behavior in this area w wt Release 3 2 0 129 July 31 2013 Pigeon Point Shelf Manager User Guide Enable disable the watchdog timer If neither of the w or wt options is present in the command line the value of the parameter WATCHDOG_ENABLED from the file etc shelfman conf determines Shelf Manager actions in this area The wt option puts the watchdog timer in test mode that is the actual timer is not turned on or strobed but the strobing thread in the Shelf Manager still runs and a warning message is printed if the interval between subsequent strobes exceeds 500 ms In normal operation if this interval exceeds one second the Shelf Manager is reset 1 s c Set logging destination use s to choose syslog and e to choose console If this option is not Specified the values of the parameters CONSOLE_LOGGING_ENABLED and SYSLOG_LOGGING_ENABLED from the file etc shelfman conf determine Shelf Manager actions in this area g lt ip_address gt Set the default gateway IP address This address is only used if no gateway address is set in the LAN Configuration Para
13. representing the physical and the logical Shelf Managers are activated even if the Shelf FRU Information cannot be found This option should be used with caution because the power consumption of the payload of the physical Shelf Manager IPM controllers may potentially exceed the power capability of the corresponding Slot in the shelf ALARM_CUTOFF_T Number 600 The alarm cutoff timeout time after Yes TIMEOUT which the alarm cutoff is deactivated in seconds Release 3 2 0 36 July 31 2013 ALLOW_ALL_ COMM ANDS_FROM_IPMB Pigeon Point Shelf Manager User Guide Boolean DEFAULT FALSE DESCRIPTION If setto TRUE most of the commands allowed from the RMCP interface are allowed from IPMB 0 as well except for session related commands For example Cold Reset and user management commands are accepted from IP MB 0 in this case In this case a malicious IPM controller can seriously jeopardize the functionality of the shelf CAN BE OBTAINED FROM SHELF FRU INFO Yes ALLOW_CHANGE_E Boolean TRUE If setto TRUE the Event receiver Yes aera address for the Shelf Manager can be Set to an address other than 20h LUN 0 If setto FALSE any attempt to change event receiver address for the Shelf Manager is rejected ALLOW_CLEARING Boolean FALSE If setto TRUE the critical alarm Yes iene condition can be cleared by the CLI command clia alarm clear ALLOW_POWER_UN
14. rootfs lt URL gt a lt URL gt application lt URL gt base lt string gt copy config copy etc copy var copy all erase all skip checksums v lt verbose level gt verbose lt verbose level gt The main command line parameters have the following meanings and at least one of the following parameters must be specified u uboot lt URL gt specify the location of the new U Boot image k kernel lt URL gt specify the location of the new Linux kernel image r rootfs lt URL gt specify the location of the new Linux root file system image a application lt URL gt specify the location of the new application package image Release 3 2 0 225 July 31 2013 Pigeon Point Shelf Manager User Guide The following additional parameters are optional verbose lt verbose level gt specify the verbosity level see below base lt string gt specify the string used as a prefix for all URLs in the command copy config copy the SSH keys the Shelf Manager configuration files and the Subdirectory var nvdata from the current image set to the new image set copy etc copy the entire ete directory and the subdirectory var nvdata from the current image set to the new image set copy var copy the SSH keys and the entire var directory from the current image set to the new image set copy all copy the entire var directory and the entire etc directory f
15. u u boot bin k sentry kernel r sentry rfs hook etc_copy rupgrade_tool copying u boot bin from tmp to dev mtdchar8 using cp protocol rupgrade_tool copying sentry kernel from tmp to dev mtdchar7 using cp protocol rupgrade_tool copying sentry rfs from tmp to dev mtdchar9 using cp protocol rupgrade_tool invoking scripts step4h etc_copy At this point the step4hshm hook Script is invoked it stops the Shelf Manager and copies non volatile information to the provisional Flash etc upgrade step4hshm Stopping Shelf Manager etc upgrade step4hshm Erasing var and etc copying var nvdata etc upgrade step4hshm Operation copy etc and var nvdata etc upgrade step4hshm Copying completed rupgrade_tool image s copy OK rupgrade_tool watchdog started rupgrade_tool selected provisional flash rupgrade_tool reboot Restarting system Here the reliable upgrade procedure resets the ShMM This causes U Boot to start from the provisional Flash x Resetting Integrated Peripherals U Boot 1 1 2 May 12 2005 21 27 13 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 08000412 DRAM 64 MB Flash 16 MB In serial Out serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 Booting image at bfb00000 Image Name IPS Linux 2 4 26 Created 2005 06 24 13 29 50 UTC Image Type IPS Linux Kernel Image gzip comp
16. DESCRIPTION This parameter specifies the minimum number of missing inactive fan trays that triggers the remaining fan trays to go to TURBO mode thatis run at full Speed in the default HPDL cooling management algorithm If this parameter is set to O this feature of Setting fans to their maximum speed due to missing inactive fan trays is disabled entirely If the shelf implements zoned cooling the specified number of missing inactive fan trays is counted relative to each zone and setting TURBO mode affects only the fan trays that participate in cooling of the corresponding zone CAN BE OBTAINED FROM SHELF FRU INFO Yes UNCONDITIONAL_ SDR_REREAD_ON_ VERS ION_CHANGE Boolean FALSE If the value of this parameter is TRUE the Shelf Manager unconditionally re reads SDRs from an IPM controller when it receives a Version Change event from that controller even if the Sensor Population Change Indicator in the Get Device SDR Info response does not change This is done for the benefit of ATCA boards that are not fully IP Ml compliant in this respect Yes USE_DHCP Boolean FALSE Requests assignment of RMCP accessible and private IP addresses for the Shelf Manager from a DHCP server the configuration parameter DHCP_SERVER_ADDRESS Can be used to specify the IP address of the DHCP server Yes Release 3 2 0 69 July 31 2013 USE_SECOND_CHA NNEL Pigeon
17. In the example above 1 is the number of the string to set multiple strings use sequential string numbers 1 2 etc for subsequent strings 7 Set at least one alert policy that sends alerts to the destination channel Fh the IP MI System Interface clia setpefconfig alert_policy 1 1 8 F 1 1 Where the fields in the example have the following meanings see the Pigeon Point Shelf Manager External Interface Reference document for detailed descriptions of the setpefconfig and other CLI commands e 1 number of alert policy table entry policy number policy enabled always send alert to this destination destination channel destination selector alert string selector eoeoeee ee ee rH 8 Set event filters for relevant events For example the following event filter can be used to capture hot swap transitions to state MO of FRU 0 on IPMC 9Ch clia setpefconfig event_filter 1 80 1 1 0 9C FF FO FF FF FF FF OF FF 0 00 0 FF FF 0 where e 1 filter number 80 filter configuration enabled software configurable filter 1 filter action alert 1 alert policy number o event severity unspecified 9c Slave Address or Software ID from Event Message FF channel Number LUN to match FO type of sensor FF sensor FF event trigger FF Event Data 1 Event Offset Mask FF Event Data 1 Event Offset Mask OF Event Data 1 AND Mask FF Event Data 1 Compare 1 o Event Data 1 Compare 2 Release 3 2 0 119 Jul
18. Loading init Ramdisk from Legacy Image at 46000000 Image Name ShMM700 RFS 3 2 0 Image Type ARM Linux RAMDisk Image gzip compressed Data Size 2277767 Bytes 2 2 MB Load Address 00000000 Entry Point 00000000 Verifying Checksum OK Loading Kernel Image OK OK Starting kernel shmm700 login root 6 Allow the ShMM to start up Please note that the settings that were changed in the U Boot firmware are not necessarily propagated to the Linux environment The reason for this is that the Shelf Manager needs to maintain its own copy of the network configuration data in order to manage failover situations If this is the first time the Shelf Manager has been booted or if the Flash devices have been reset to factory default prior to bootup then the Shelf Manager uses the network settings provided by U Boot to set up this networking context and thus the changes you made in U Boot are propagated forward Otherwise the network configuration data is stored on the ShMM Flash file system in the following files var nvdata ch1_param parameters for channel 1 var nvdata chl_param_v2 additional parameters for channel 1 Shelf Manager release 2 6 and higher var nvdata ch2_param parameters for channel 2 var nvdata ch2_param_v2 additional parameters for channel 2 Shelf Manager release 2 6 and higher In this latter case the following steps are required to configure the network settings in the Shelf Manager co
19. tmp sentry rfs lt INFO gt frontend c 449 Downloading file file tmp sentry app lt INFO gt frontend c 535 Initiating partial upgrade lt INFO gt frontend c 735 uboot tmp _tmp_uboot lt INFO gt frontend c 745 kernel tmp _tmp_kernel lt INFO gt frontend c 755 rfs tmp _tmp_rfs lt INFO gt frontend c 765 app tmp _tmp_app lt INFO gt frontend c 820 Calling backend to handle partial upgrade sbin backend uboot tmp _tmp_uboot kernel tmp _tmp_kernel rootfs tmp _tmp_rfs application tmp _tmp_app lt INFO gt shmm700_hal c 1285 Write storage state not available gt available ALLOW UBI device number 1 total 484 LEBs 31657472 bytes 30 2 MiB available 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB lt INFO gt shmm700_hal c 369 Resetting UBIFS cache Erasing 64 Kibyte 30000 75 complete lt INFO gt shmm700_hal c 1913 Clean on defice 1 1 Leaving file dev mtd2 untouched lt INFO gt backend c 730 Burning image lt INFO gt backend c 764 393216 bytes was written from file tmp _tmp_uboot to dev 1 1 Erasing 64 Kibyte 30000 75 complete 512 0 records in 512 0 records out 262144 bytes 256 0KB copied 1 571406 seconds 162 9KB s lt INFO gt backend c 730 Burning lt INFO gt backend c 764 1402348 bytes was written from file tmp _tmp_kernel to dev 2 1 lt INFO gt backend c 730 Burning lt I
20. RSp oad type oad type oad type oad type Set Set G RAKP RAKP RAKP RAKP Session Session t Device ID Get G Device ID Get t Device ID Device ID Rsp RSP Req Rsp nd Message Send Message Get Device ID C G lose Session lose Session To stop the capture session terminate the tshark process e g by pressing C 5 4 5 Analyzing a Trace Using the GUI To selecta message in the Wireshark GUI main window simply scroll to the message that interests you in the upper frame of the main window Release 3 2 0 187 July 31 2013 Pigeon Point Shelf Manager User Guide Figure 14 Message 244 is Selected 192 168 165 Capturing File Edit View Go Capture Analyze Statistics Help Se oO a a aF EB aaah BUSS g Filter s Expression Clear Apply No Time Source Destination Protocol Info sca RAD rari SR a ERR RRR AS ow 237 6 761618 I2C 1 0x76 IPMI ATC Req Get Device ID seq Ox3e 238 6 762032 12C 2 0x76 IPMI ATC Req Get Device ID seq 0x3e 239 6 762581 I2C 2 0x77 IPMI ATC Req Get Device ID seq 0x3f 240 6 762993 I2C 1 0x77 IPMI ATC Req Get Device ID seq 0x3f 241 6 763541 I2C 1 0x78 IPMI ATC Req Get Device ID seq 0x00 242 6 763952 I2C 2 0x78 IPMI ATC Req Get Device ID seq 0x00 243 6 764502 I2C 2 0x79 IPMI ATC Req Get Device ID seq 0x01 244 6 764913 I2C 1 0x79 IPMI ATC Req Get Device ID seq 0x01 245 6 7654
21. SHELF MANAGER User Guide Release 3 2 0 J uly 31 2013 Pigeon Point Shelf Manager User Guide 2002 2013 Pigeon Point Systems LLC All rights reserved Pigeon Point Shelf Manager and ShMM 500 ShMM 500R ShMM 500RE ShMM 1500R ShMM 700R This document is furnished under license and may be used or copied only in accordance with the terms of such license The content of this manual is furnished for informational use only is subject to change without notice and should not be construed as a commitment by Pigeon Point Systems Pigeon Point Systems assumes no responsibility or liability for any errors or inaccuracies that may appear in this book Except as permitted by such license no part of this publication may be reproduced stored in a retrieval system or transmitted in any form or by any means electronic manual recording or otherwise without the prior written permission of Pigeon Point Systems The Pigeon Point Shelf Manager uses an implementation of the MD5 Message Digest algorithm that is derived from the RSA Data Security Inc MD5 Message Digest algorithm The Pigeon Point name and the stylized lighthouse logo as well as Monterey Linux and IntegralHPI are trademarks of Pigeon Point Systems LLC Linux is a registered trademark of Linus Torvalds Release 3 2 0 2 July 31 2013 Pigeon Point Shelf Manager User Guide Table of Contents 1 ABOUT THIS DOCUMENT iccsseccctcicd creck fecceheded ect cbepeccchcted sa
22. The following is a parsed version of sample SNMP Trap in the default IP MI PET v1 0 format The original event which is reported by this trap follows Event at Tue Aug 5 16 29 21 2008 from 0x20 0 0 sensor 0x12 0x85 event 0x6f asserted OxCO 0x00 0x00 This trap describes a system reconfigure event which is generated during Shelf Manager startup and switchover In the example below a bold hexadecimal represents the Tag of a field and an underlined hexadecimal represents the length of a field message length 125 30 7B 02 01 00 04 06 70 75 62 6C 69 63 A4 6E 06 09 2B 06 01 04 01 98 6F 01 01 40 04 CO A8 00 02 02 01 06 02 03 12 6F 00 43 04 61 A3 ED 13 30 4D 30 4B 06 OA 2B 06 01 04 01 98 OF 01 01 01 04 3D 1D 3A 30 76 62 F2 11 DD 00 80 00 50 C2 3F FB 56 00 07 13 ED A3 61 00 00 20 20 01 20 85 00 00 CO 00 00 00 00 00 00 00 19 00 00 40 OA 00 00 80 4B 01 54 65 73 74 53 74 72 69 6E 67 00 Cl Here is a detailed description of the content of this trap ASN 1 tag and remaining length oo SNMP version field len 1 O SNMP v1 70 75 62 6c 69 63 community field public PDU Trap command tag and remaining length 30 02 04 A4 oO j bh Oy oD CJ Release 3 2 0 122 J uly 31 2013 Pigeon Point Shelf Manager User Guide 06 09 2B 06 01 04 01 98 6F 01 01 AgentID Enterprise OID 1 3 6 1 4 1 3183 1 1 40 04 co as 00 02 Agent IP v4 192 168 0 2 02 01 06 Generic Status len
23. Upper Critical Threshold Lower Non Critical Threshold Lower Non Recoverable Threshold Lower Critical Threshold Lower Non Critical Threshold If and only if the symbolic constant HYSTERESIS has been specified in the Sensor Release 3 2 0 112 J uly 31 2013 Pigeon Point Shelf Manager User Guide Initialization field the following fields from the replacement SDRs specify the replacement hysteresis values for the target sensor Positive Hysteresis Negative Hysteresis Hysteresis values must be specified in the raw format with the Ox prefix e The field OEM from the replacement SDR always replaces the corresponding attribute of the target sensor even if not specified e The field Id String replaces the target sensor name only if specified in the replacement SDR The example below illustrates a typical local sensor configuration text definition It redefines thresholds for the two temperature sensors on the physical IPM controller It is assumed that the physical IPM controllers have IPMB 0 addresses 10h and 12h in the target shelf Full Sensor Record Owner Id 0x10 Sensor Number 2 Sensor Initialization THRESHOLDS Lower Non Critical Threshold 0xb0 Lower Critical Threshold 0xc0 Lower Non Recoverable Threshold 0xd0 Upper Non Critical Threshold 0x40 Upper Critical Threshold 0x48 Upper Non Re
24. e ftp lt server gt lt dir gt lt user gt lt pwd gt Copy froma remote FTP server This protocol assumes that all of the specified lt s re gt files are to be copied to the ShMM from the FTP server host specified as the host name or the IP address by lt server gt All the images must reside in the directory specified by lt dix gt on the remote FTP server The FTP connection is made using the account specified by the lt user gt parameter with the password Specified by the optional lt pwd gt parameter If no lt pwd gt is supplied the utility prompts for a password e scp lt server gt lt dir gt lt user gt lt id_file gt port Copy from a remote SSH server using the SCP protocol This protocol assumes that all of the specified lt s re gt files are to be copied to the ShMM from the secure shell server host specified as the host name or the IP address by lt server gt All the images must reside in the directory specified by lt dir gt on the remote SSH server The path is absolute The user name for the SSH connection is specified by the lt usexr gt parameter It is not possible to supply a password in this mode but the lt id_file gt parameter allows specifying the secure shell identity file to be used for password less authorization The lt port gt parameter can be used to specify a non standard port number for the remote SSH server If the lt port gt parameter is not Supplied the default SSH service port number
25. lt bus_type gt BUSED DUAL_STAR REDUNDANT_DUAL_STAR lt cross_link_num gt 0 1 The lt cross_link_num gt Specifies the link number used for communication with the peer Shelf Manager in a radial configuration and is used only with the Redundant Dual Star topology The default value is 0 which means that no cross link exists and the Shelf Managers communicate with Release 3 2 0 148 July 31 2013 Pigeon Point Shelf Manager User Guide each other over the bused IPMB 0 Currently only link number 1 can be specified as the crosslink as a result the only legal values for lt cross_link_num gt are 0 and 1 Remaining aspects of radial IP MB 0 support in SnMM 700 based shelves are the same as in ShMM 1500 based shelves except for programming upgrading the routing element FP GA s which is described in the next section 4 5 3 1 Upgrading the Routing Element FPGA s The routing element FPGA s can be programmed upgraded using the ripmb_prog utility which is included as part of the standard ShMM 700R RFS image In addition to programming the FPGA s this utility can be used to obtain the version of the currently installed image s IMPORTANT before running ripmb_prog to program upgrade the routing element FPGA s terminate the Shelf Manager by issuinga clia terminate command Failing to do so may lead to FPGA access errors and unpredictable behavior If ripmb prog is invoked only to obtain the version of the currently in
26. the Shelf Manager makes a call to rupgrade_tool f which completes the upgrade procedure Release 3 2 0 210 July 31 2013 Pigeon Point Shelf Manager User Guide e The user optionally calls rupgrade_tool w in order to find out the status of the upgrade session As explained above this option may be especially useful for a remote upgrade session where the progress of the upgrade cannot be observed directly from the messages printed to the serial console as is the case for a local upgrade e After the completion of the reliable upgrade the user can revert to the original images if he she detects that the new images are not acceptable for any reason To do this the user calls rupgrade_tool sR hook flip If necessary the above sequence can be easily automated by developing a simple script designed to run on a remote network host Alternatively an operator can carry out the reliable upgrade manually either locally from the serial console or remotely over the network 7 8 Reliable Upgrade Examples This section provides reliable upgrade examples for both ShMM 500 and ShMM 1500 The reliable upgrade procedure works similarly on these two types of ShMMs but there may be minor differences in the appearance of the command output 7 8 1 Example 1 This example shows a reliable upgrade of all three components U Boot kernel and RFS image copying etc and var nvdata non volatile directories to the provisional Flash All images are t
27. 0 and operational state for each FRU activated deactivated etc e Generating events typically directed to the Shelf Manager for exceptional conditions detected by any sensor based on its configured event generation settings e Negotiating with the Shelf Manager for resources needed by the FRU s including power and interconnects BMR firmware running ona Carrier IPMC additionally represents its installed AMCs or an intelligent RTM to the Shelf Manager including negotiating for power resources on their behalf Furthermore a Carrier IPMC negotiates with its modules and on carrier switching resources regarding interconnect configurations Release 3 2 0 14 July 31 2013 Pigeon Point Shelf Manager User Guide The AMC management architecture is purposely designed to 1 avoid impacting existing ATCA Shelf Managers and 2 minimize the resources required to implement a Module Management Controller since board real estate and cost are at a premium on AMCs There are Pigeon Point BMR variants for AdvancedTCA IPM Controllers as well as for Carrier IPMCs and MMCs The principal BMR variants are based on the Microsemi S martF usion intelligent mixed signal FPGA with variants based on the Atmel AVR ATmega and Renesas H8S microcontroller families for some controller types Please see httpo www pigeonpoint com products html for more details on these offerings as well as on the Pigeon Point solutions for MicroTCA controllers Pigeon Poi
28. 3338383 rupgrade is called to strobe the reliable upgrade WDT once again just before starting the Shelf Manager or confirming the upgrade if automatic start is disabled via the environment variable start_rc2_ daemons lt gt 09 28 46 973 134 Pigeon Point Shelf Manager ver 3 2 0 Built on Mar 17 2013 22 20 24 lt gt 09 28 47 012 134 Lock log print buffer at Oxla2d74 lt gt 09 28 47 013 134 Pthread lock log print buffer at 0x1a7594 KKK lt I gt 09 28 47 044 134 Reading configuration file etc shelfman conf shmm700 login lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW The reliable upgrade WDT is strobed once again just to make sure it won t fire during the confirm procedure lt INFO gt shmm700_hal c 1189 Write confirmed 1 gt 0 ALLOW rupgrade changes the confirmed image set index to the candidate image set index lt INFO gt shmm700_hal c 1202 Write upgrade_state in progress 2 gt confirmed 4 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 0 ALLOW And finally rupgrade turns the reliable upgrade WDT off The reliable upgrade is complete Finally the user checks the status of the reliable upgrade by calling rupgrade s xrupgrade s lt INFO gt frontend c 37 Last upgrade log lt INFO gt shmm700_hal c 1285 Write storage state not available gt available ALLOW lt INFO gt shmm700_hal c 369 Resettin
29. ARM9 based processor Both the ShMM 1500R and the ShMM 700R have different physical dimensions and a different approach for connecting to the ShMM carrier Therefore distinct ShMM carrier board designs are required for Shelf Manager Solutions based on each of the ShMM 500R ShMM 1500R or ShMM 700R ShMM variants However the Pigeon Point Shelf Manager provides the same functionality and high level interfaces for all three ShMM variants The Pigeon Point Shelf Manager is adaptable to manage CompactP Cl platforms as well This document focuses primarily on AdvancedTCA contexts but provides Compact Cl specific comments where appropriate 2 1 In This Section This section contains the topics listed below J ust click on a topic to go to it e Intelligent P latform Management An ATCA Overview e Pigeon Point Board Management Reference Hardware and Firmware e Pigeon Point Shelf Manager and ShMM 2 2 Intelligent Platform Management An ATCA Overview The Pigeon Point products are the first Intelligent P latform Management building blocks designed from the ground up for modular platforms like AdvancedTCA in which there is a strong focus ona dynamic population of Field Replaceable Units FRUs and maximum service availability Release 3 2 0 11 J uly 31 2013 Pigeon Point Shelf Manager User Guide The Intelligent P latform Management Interface IP MI specification provides a solid foundation for the management of such platforms but requires
30. Boolean FALSE This variable affects the behavior of Yes RELATED_FRU_IN _CRITICAL_STAT E the Shelf Manager with respect to powering up FRUs that are in state M3 when the shelf is in Critical thermal alert state If set to TRUE the FRU can be powered on if the Critical alert state is caused by temperature sensors that belong to a different FRU If set to FALSE no FRU can be powered up when the Shelf is in the Critical alert state In the case of a Critical alert caused by a Shelf wide sensor no FRU can be powered up irrespective of the value of this variable Release 3 2 0 37 July 31 2013 ALLOW_RESET_ST ANDALONE Pigeon Point Shelf Manager User Guide TYPE Boolean DEFAULT FALSE DESCRIPTION If setto TRUE the command Cold Reset is accepted even if the Shelf Manager does not have an available backup and reboots the Shelf Manager By default the command Cold Reset is accepted only ina dual redundant configuration and causes a Switchover CAN BE OBTAINED FROM SHELF FRU INFO Yes ALLOWED_CIPHER SUITES Number OxF FFF FFF F decimal 1 The bit mask of the IP MI 2 0 cipher Suite IDs thata LAN clientis allowed to use when establishing a session to the Shelf Manager By default all Supported cipher suites can be used If security considerations require the set of allowed cipher suites can be restricted using this parameter A
31. If TRUE instructs the Shelf Manager to retrieve configuration parameters from a special set of records in the Shelf FRU Information If these parameters are successfully retrieved they override the values from configuration files Not all configuration variables can be overridden from the Shelf FRU Information for example this variable cannot be overridden for obvious reasons and to be effective must be specified in the configuration file CAN BE OBTAINED FROM SHELF FRU INFO No SHORT_SEND_MSG Boolean TRUE Determines the type of the Send No SRESEONSE Message response provided by the Shelf Manager required by the current PICMG 3 0 if TRUE or compatible with the previous versions of the Shelf Manager if FALSE SPECIAL_HPI_EN Boolean TRUE If TRUE IntegralHPI uses the entity No TITY_FOR_AMC_C typ e ia SAHPI_ENT_SUBBOARD_CARR IER_BLADE for AMC carrier boards otherwise the entity type SAHPI_ENT_PICMG_FRONT_B LADE is used SWAPPED_CROSS_ Boolean FALSE Swaps the names of network Yes CONNECTS adapters used for cross connects on the ShMC with the odd hardware address Release 3 2 0 65 J uly 31 2013 SWITCHOVER_ON_ BROKEN_LINK_BA CKUP_DELAY Pigeon Point Shelf Manager User Guide TYPE Number DEFAULT DESCRIPTION This parameter specifies the delay on the backup side during a switchover caused by a broken physical network link When the active Shelf Manager
32. New section 3 13 1 adds a parsed example of an SNMP Trap in default IP MI PET v1 0 format New section 3 14 describes configuration of the IntegralHP subsystem C 13 Release 2 6 4 Section 2 4 4 lists the full set of ShMM variants including those that have reached an end of life status Describes new ShMM 1500 models with encryption code removed Section 3 2 1 adds descriptions of the HP DL carrier specific option CARRIER_FRU_LOCATION Release 3 2 0 258 July 31 2013 Pigeon Point Shelf Manager User Guide Section 3 3 adds new configuration parameters BOARD_LAN PARAMETERS CHANNEL LIST INITIALIZATION SCRIPT PEF_USE_KEYED_ALARMS SEL_FILE_COMPRESSION_ENABLED SEL_FILE JOURNALING ENABLED SEL_FILE WRITE_DELAY SHELF_MANAGER_CONFIGURATION_IN_SHELF_FRU_INFO TIMEPROTO TIMESERVER New section 3 3 2 describes a new mechanism for retrieving configuration variables from Shelf FRU Information New section 3 4 2 4 describes bonded usage of the two network interfaces New section 3 6 4 1 describes how to specify location of the Carrier FRU Information Section 3 14 2 adds a note about the special version of the HPI client library Section 4 3 adds a description of the option i for Shelf Manager command line invocations adds a note about Shelf Manager configurations with encryption code present or removed New section 4 10 describes how to use a new facility in which the Shelf Manager stores and assigns LAN configuration par
33. Release 3 2 0 4 July 31 2013 Pigeon Point Shelf Manager User Guide 5 4 6 Analyzing a Trace Using Command Line TOOIS ccccccccssccessseceeseeeeseessnees 190 6 RE INITIALIZING THE SHMM 0 ccccccsccssseeceseeeeeeeeesnaeeesneeeeeeeeseaeseseeeenseeesseeesseaesaseneneeees 193 6 1 IN THIS SECTION sis2s8h cones ania Getidaeiis dati doatn edited tite 193 6 2 RE INITIALIZING THE U BOOT ENVIRONMENT ccseeceeeeeeeeeeeeeeeaeeeeeeaaeeeeeeaeeeeeeaeeeeseaaes 193 6 3 RE INITIALIZING THE FILE SYSTEM cccceesceceeccececeeeeeaeeeeeeeseaeeeeaaeeeeaaeseeeeeseaeeseaaeseeneesaes 194 6 4 RESETTING THE LOGIN PASSWORD eeeeeeceeeeeeeeeeeeeaeeeeeeaaeeeeeeaaeeeseeaaeeesneaaeeeeneaeeeenenaes 195 7 RE PROGRAMMING THE SHMM 500 1500 ccssccssseeeeeeeeeseeseseeeenseeeesseeeseaeseseeeenseees 196 7 1 IN THIS SECTION ai ged ceeees ican ade covet aa dencen afectvatesctadeas adia raanda ninae aara iieiea anaiei 196 7 2 FIRMWARE RELIABLE UPGRADE PROCEDURE OVERVIEW ccccccesseeeeeeeeeeeeeeseaeeeeneeseaes 196 7 3 FLASH PARTITIONING ceres snn aauct ividecocthatcuc uyhsdanevhesdaptnscandcvbedusesuphadeecehelvactuybeestexs 197 7 4 THE VAR UPGRADE FILE SYSTEM ccccecceeeeeceeeeeceaeeeeaaeeeeeeeceaeeesaaeeeeaeeseeeeeseaeeeeaaeeeeeees 202 7 5 RELIABLE UPGRADE PROCEDURE STATUS FILE 0csesceeeeeeeeeeeeeaaeeeeeeeseaeeeseaeeeeneeseaes 202 7 6 RELIABLE UPGRADE UTILITY iiriricsesnn iaa 203 7 7 RELIABLE UPGRADE UTILITY USE SCENARIOS
34. ShMM variant Hardware level status updates are communicated between the Shelf Manager instances via the Hardware Redundancy Interface HRI which is also implemented differently in each of the three ShMM variants Release 3 2 0 16 July 31 2013 Pigeon Point Shelf Manager User Guide Figure 2 Pigeon Point Shelf Manager Architecture and Context SanH G3LOSNNOO SSOHYO KEY MANAGEMENT SUBSYSTEMS I I RDEDIORTOTE RASS SOP ate ORR re em em 1 i cue I I CLI sul I t A w g 1 ret li t O25 11 I i ate I I i I NEZ oe r 1 See a a a aaa I IOo I OME aaan Peal oaot a i nt aan aea anaa ead Il o an l LO Z Em iz S4yuo Orsi I I 7 02 1 Y oWes iit SISE I m n Ei E739 Imi 25l 1 Z Naw ee e e eee e e ee ee eee eee ee ee ee ee ee 7 4 1 gm ee ee ee 1 12 i ID I 1 1 I I 1 I I I I i 1 I I I 1 I As shown in Figure 2 ShMC cross connects allow both ShMMs to be connected with both Base Interface Hubs This improves system availability because either hubs or ShMMs can switchover independently if necessary ShMC cross connect support requires the use of both ShMM Ethernet links and is implemented in essentially all modern ShMM based shelves that support Shelf Manager redundancy In the now obsolete ShMM 300 family one of the two ShMM Ethernet links was typically routed directly between the two ShMMs and used for the SRI With this arrangement ShMC cross connect support was not
35. at www python org After producing the binary SDR definition file you should place it on your ShMM in the directory var nvdata under the name user_sdr For example you can use FTP gt ftp 192 168 191 Connected to 192 168 1 191 192 168 1 191 Release 3 2 0 109 July 31 2013 Pigeon Point Shelf Manager User Guide 220 shmm 191 FTP server Version wu 2 6 2 1 Wed Oct 5 21 30 04 GMT 2005 ready Name 192 168 1 191 username anonymous 331 Guest login ok send your complete e mail address as password Password 230 Guest login ok access restrictions apply Remote system type is UNIX Using binary mode to transfer files ftp gt cd var nvdata 250 CWD command successful ftp gt put user_sdr local user_sdr remote user_sdr 227 Entering Passive Mode 192 168 1 191 246 195 150 Opening BINARY mode data connection for user_sdr 226 Transfer complete 124 bytes sent in 8 5e 05 secs 7 6e 03 Kbytes sec ftp gt quit 221 You have transferred 124 bytes in 1 files 221 Total traffic for this session was 1165 bytes in 1 transfers 221 Thank you for using the FTP service on shmm 191 221 Goodbye A newly installed SDR definition file var nvdata user_sdr takes effect only after a restart of the Shelf Manager the Shelf Manager log during that restart could look like the following example daemon f shelfman lcs cs sf eth0 1F lt I gt 03 45 09 041 152 Pigeon Point Shelf Manager ver 3 2 0
36. be changed by editing the configuration file etc snmpd conf 514 tcp RSH server Activated by the etc inetd conf file to disable the RSH Server comment out the she11 line in that file Runian rshd help to see list of supported arguments that can be specified Release 3 2 0 125 July 31 2013 Pigeon Point Shelf Manager User Guide PORT NUMBER TYPE NETWORK SERVICE DESCRIPTION 623 udp ShM RMCP interface Activated by the ete re common Script if the U Boot variable start_rc2_daemons is Set to y to disable the RMCP interface comment out the corresponding line in the script This interface is available only on the active ShMM and only if the Shelf Manager is operational The Shelf Manager only accepts connections targeted to configured RMCP address es 1040 tcp ShM Redundancy interface This is the inter ShMM communication interface needed in redundant configuration It only accepts connection targeted to the IP address Specified as the configuration parameter REDUNDANT_IP_ADDRESS in the Shelf Manager configuration file 4743 tcp HPI server This service is available when an instance of an on ShMM OpenHPI or IntegralHP running on the ShMM IntegralHPI is activated as a part of the Shelf Manager if the configuration variable ENABLE_INTEGRALHPT is setto TRUE and the HPI dynamic library is present OpenHP I on ShMM version is started by the etc re common Script so itis possible to comment out correspo
37. causes a TELCO alarm is counted separately in this case so that if a specific alarm is raised due to multiple event conditions it stays raised while any of these conditions stay asserted Release 3 2 0 59 J uly 31 2013 Pigeon Point Shelf Manager User Guide DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO PET_FORMAT Number 0 Specifies the format of the Platform Yes Event Traps that are sent by the Shelf Manager as the Alert action initiated by event processing in the Platform Event Filtering facility The values are defined as follows O the default IP MI format defined by IPMI Platform Event Trap Format v1 0 Specification 1 plain text format all the event details are sent as plain ASCII text in a single variable 2 multi variable format each event field is encoded as a separate variable PET_OEM_WITH_S Boolean FALSE If TRUE the event severity and alert No earner string is added to a trap message in the PPS OEM Plain Text and Mult Variable PET formats The default IPMI PET format always includes these fields from the corresponding event filter POWER_UNLISTED Boolean TRUE Allow the FRUs not listed in the power No FRUS management table in the Shelf FRU Information to be activated and powered up PROPAGATE_RMCP Boolean FALSE If TRUE the active Shelf Manager Yes _ADDRESS propagates the RMCP IP address to the backup Shelf Manager which co
38. customers By default this variable is FALSE and a standard Pigeon Point Client ID format is used as described in section 3 4 7 of this document CAN BE OBTAINED FROM SHELF FRU INFO Yes ENABLE_DIRECT_ SHELF_FRU_WRIT E Boolean FALSE This variable controls whether the Shelf Manager allows direct writes to the Shelf FRU Info FRU Device ID 254 on the IPM controller at 20h on IPMB 0 via the IPMI command Write FRU Data By default direct writes are prohibited as mandated by PICMG 3 0 R2 0 ECN 002 which requires that Shelf FRU Info writes use a special locking protocol so that only one writer is active at once Setting this variable to TRUE enables direct writes for compatibility with System Manager applications that rely on the pre ECN 002 behavior No ENABLE_INTEGRA LHPI Boolean FALSE This variable determines whether the Shelf Manager exposes the Hardware Platform Interface HP defined by the Service Availability Forum www saforum org Setting this variable to TRUE enables the separately licensed IntegralHP implementation within the Shelf Manager and makes the Shelf Manager accessible via HPI No Release 3 2 0 49 July 31 2013 ENABLE _LOCKS_L OGGING Pigeon Point Shelf Manager User Guide TYPE Boolean DEFAULT DESCRIPTION If this variable is TRUE default value the Shelf Manager records all requests to acquire and release locks
39. introduces the option for SEL truncation to be done automatically under control of the System Manager C 4Release 2 4 0 Section 3 4 6 adds a description of the Request Identifier Request ID byte of the DHCP Client Identifier along with coverage of the default inclusion of the Shelf Address in the DHCP Client Identifier Section 3 3 introduces the new configuration parameter RMCP_WITHOUT_SHELF_FRU Section 4 4 1 provides a description of the new Redundancy and CPLD sensor that is provided by both active and backup Shelf Managers Section 7 3 deletes the table describing the Flash partitions fora ShMM 500 with 32MB Flash devices this configuration is not offered C 5Release 2 4 1 Section 3 2 1 changes the default value of the U Boot environment variable hostname to shmm300 or shmm500 based on the ShMM type Section 3 2 4 a new section that describes how to establish the secondary RC script Section 3 3 contains the content of the former section 3 4 Setting up Shelf Manager Configuration File Section 3 3 uses the new configuration parameter name DHCP_SERVER_ADDRESS instead of the original name PREFERRED_DHCP_SERVER Section 3 5 2 adds caveats about using the eepromw utility Section 3 5 3 a new section that describes how to setup the Shelf FRU Information using CLI commands Section 3 11 adds a description of how ShMM POST results are reported using the IP MI command Get Self Test Results C 6Release
40. 0 95 July 31 2013 Pigeon Point Shelf Manager User Guide configuration supports the approach where redundant Shelf FRUs are represented by separate IPM controllers as FRU 1 as well as the two SEEPROMs approach The following key configuration variables are set as follows for that case LOCAL_SHELF_FRU TRUE SHELF_FRU_IN_EEPROM TRUE In this case however support for SEEPROMs must also be provided by the carrier specific module in the Shelf Manager If none of the above approaches is supported by the shelf and there are no sources of Shelf FRU Information represented by separate IPM controllers the system integrator must resort to the Flash file as storage for Shelf FRU Information In that case the following changes to the default configuration should be done e setthe variable SHELF_FRU_IN_EEPROM to FALSE e setthe variable MIN_SHELF_FRUS to 1 The last change is necessary because there is only one copy of the Shelf FRU Information on each ShMM The variable LOCAL_SHELF_FRU must retain its default value of TRUE The following table summarizes the configuration variable settings that correspond to the various Shelf FRU Information source possibilities described above Table 10 Shelf FRU Information and Configuration Variable Settings SOURCE OF THE SHELF FRU INFORMATION SETTINGS OF CONFIGURATION VARIABLES Non volatile storage likely SEEPROMs LOCAL_SHELF_FRU FALSE accessed via IPM controllers on I
41. 0x19 b Frame 1 7 bytes on wire 7 bytes captured b Inter Integrated Circuit Data b Intelligent Platform Management Interface 0000 02 18 e6 20 bO 01 2f seed TCP 192 168 1 65 lt live capture in pro Packets 274 Displayed 274 Ma Profile Default A The upper frame shows a list of captured messages that provides basic information about each IPMB message The list can be sorted by the contents of each field as a key Release 3 2 0 173 July 31 2013 Pigeon Point Shelf Manager User Guide In an IPMB trace analyzer usage this frame displays the full list of collected IPMB messages For each message the following basic information is provided Table 17 IPMB Message List Basic Information GUI FRAME FIELD INTERPRETATION No The index number of the message in the trace Time Message date and time stamp displayed as an offset from the first packet The time stamp obtained on the targetis available in the detailed packet information in the bottom frame Source The IPMB the message was captured on This field is either L2C 1 for IPMB A or I2C 2 for IPMB B Destination Destination 12C address of the message in 7 bit format This field is for IPMB events Protocol The type of the captured message This field can be one of the following IPMI ATCA for an IPMI ATCA formatted message EVENT for an PMB event I2C READ 0rI2C WRITE fora raw lC transaction Info Basic information about the capture
42. 14 35 24 115 236 Set Port State disable ipmc FC chan 2 it 0 ports 1 lt 1 ext 0 group 0 lt I gt 14 35 24 130 217 ipmc_cooling_scan_sensors Reread sensors on the SA OxFC lt I gt 14 35 24 134 213 Controller 20 FRU 1 ATCA state set to M3 prev M2 cause 1 locked 0 lt I gt 14 35 24 137 213 Controller 20 FRU 1 ATCA state set to M4 prev M3 cause 0 locked 0 lt I gt 14 35 24 147 231 sdrrep finished reading SDRs for SA FC added 29 preserved 0 deleted 0 time 7 seconds lt I gt 14 35 24 152 214 Hot Swap event SA 0x20 FRU 1 sensor_number 0x02 M2 gt M3 cause 1 lt I gt 14935224 167 236 SA Oxfc FRU 0 is OPERATIONAL lt I gt 14 35 24 179 214 Hot Swap event SA 0x20 FRU 1 sensor_number 0x02 M3 gt M4 cause 0 lt I gt 1435524211 235 Tasklet ACTIVATE 20 2 is_local_address 1 lt W gt 14 35 24 213 235 ocal FRU 2 at SA 0x20 is not listed in Shelf FRU Info still activate lt I gt 14435 24 236 213 Controller 20 FRU 2 ATCA state set to M3 prev M2 cause 1 locked 0 lt I gt 14 35 24 241 214 Hot Swap event SA 0x20 FRU 2 sensor_number 0x03 M2 gt M3 cause 1 lt I gt 14 35 24 257 218 Adjust power for 20 01 from 0 to 5000 lt I gt 14 35 24 258 218 Sending sync point lt I gt 14 35 24 269 237 SA 0x20 FRU 1 is OPERATIONAL lt I gt 14 35 24 301 218 Adjust power for 20 02 from 0 to 20000 lt I gt 14 35 24 302 218 Sending sync point Release 3 2 0 135 July 31 2013 Pigeon
43. 192 168 1 253 option bootfile name tftpboot ssh_script_eth0O_ShM2 host client03 option dhcp client identifier 0 0 0 0 0 0 0 21 fixed address 192 168 1 143 option tftp server name 192 168 1 253 option bootfile name tftpboot ssh_script_eth1_ShM2 host client04 option dhcp cl nt identifier 0 0 0 0 0 0 0 0 H Release 3 2 0 94 J uly 31 2013 Pigeon Point Shelf Manager User Guide fixed address 192 168 1 144 host client0O5 option dhcp client identifier 0 0 0 0 0 0 0 1 fixed address 192 168 1 145 3 5 Configuring the FRU Information This section describes configuring the Field Replaceable Unit FRU Information 3 5 1 Accessing the Shelf FRU Information According to the ATCA specification the Shelf FRU Information should be redundant at least two copies per shelf and each copy may be represented by separate IPM controllers as FRU 1 Some ATCA shelves adopt this approach fully For such shelves the default configuration file must be changed the variable LOCAL_SHELF_FRU must be set to FALSE This enables a shelf wide search for potential sources of Shelf FRU Information on IP MB 0 On most ATCA shelves implementing this approach the Shelf FRU Information is accessed via two IPM controllers with known IPMB addresses In this case itis possible to limit the search for Shelf FRU Information to the two pre determined IP MB locations To do this two configu
44. 2 4 2 Section 3 3 1 adds in a new subsection clarifications for the role of a carrier specific Shelf Manager configuration file in determining the effective values of Shelf Manager configuration variables Release 3 2 0 256 July 31 2013 Pigeon Point Shelf Manager User Guide C 7Release 2 4 4 Section 3 3 4 adds in a new subsection a description of the new configuration parameter VERBOSITY_CONSOLE C 8Release 2 5 0 Section 3 3 adds new configuration parameters DETECT_DEADLOCKS EXIT_IF_HEALTHY LOST _IN_STANDALONE_MODE EXTERNAL_EVENT_HANDLER INNER_SEQUENCE_NUMBER_IN_SEND_MSG_ RESPONSE IPMC_PRESERVE_ON_REVISION_CHANGE New section 3 12 describes the new external event handling facility including how to configure and use it New section 4 8 describes the deadlock detection facility Entire document removes coverage of first generation ShMM 300 mezzanine Starting with release 2 5 0 the ShMM 300 is not supported on new releases of the Shelf Manager C 9Release 2 5 2 Section 3 3 adds new configuration parameters COOLING_MANAGEMENT HPDL HPDL_ON_SUBSIDIARY_FRUS ISOLATE_MUX_ON_GPIO8 PET_FORMAT SWAPPED_CROSS_CONNECTS ENABLE_DIRECT_SHELF_FRU_WRITE Modifies the description of the configuration parameter SWITCHOVER_ON_HANDLE_OPEN to reflect that programmatic deactivation also triggers a switchover Section 3 4 2 2 describes the second instance of the Shelf Manager IP Connection record in the Shelf FRU Informatio
45. Built on May 17 2013 11 03 27 lt gt 03 45 09 055 152 Limits code 10af4040 10ba95c0 end_data 10c355d4 start_stack 10c36e64 sp 10c36874 eip 10bal838 lt gt 03 45 09 063 152 Stack limits curr 800000 max fffffffft lt gt 03 45 09 067 152 Data limits curr ffffffff max fffffffFf lt gt 03 45 09 090 152 Lock log print buffer at 10c10e90 lt gt 03 45 09 094 152 Pthread lock log print buffer at 10c14ed0 kkk lt I gt 03 45 09 380 152 Enabling the CPLD Active bit workaround lt I gt 03 45 09 435 152 User SDR size 124 lt I gt 03 45 09 453 152 20 120 Type 1 lt I gt 03 45 09 467 152 a0 2 Type 1 If the SDR definition file var nvdata user_sdr is not present the configurations of the local sensors are unmodified from the default established in the Shelf Manager and the log looks like the following example daemon f shelfman lcs cs sf ethO 1F lt I gt 17 11 28 004 522 Pigeon Point Shelf Manager ver 3 2 0 Built on May 17 2013 19 03 33 lt gt 17 11 28 009 522 Limits code 400000 529030 end_data 10062000 start_stack 7fff7df0 esp 7fff 7758 eip 2ab0d2e4 Release 3 2 0 110 July 31 2013 lt gt lt gt lt gt lt gt KKK lt W gt It iS Pigeon Point Shelf Manager User Guide 17 11 28 010 1711728010 17511528 014 172 11 28 014 Stack limits curr 1ff000 max 7ffffffFf Data limits curr 7fffffff max 7fffffff Lock log
46. Control Protocol RMCP required by AdvancedTCA command line web browser Simple Network Management Protocol SNMP e Includes a Shelf Adaptation Layer based on the Pigeon Point Hardware Platform Description Language HPDL that makes it straightforward to adapt the Shelf Manager for operation across a wide range of ATCA shelf architectures e Optionally supports an additional built in shelf external interface that complies with the Hardware Platform Interface HPI a set of application programming interfaces APIs for managing hardware platforms that is defined by the Service Availability Forum www saforum org The Pigeon Point Shelf Manager can also be used in Compact CI shelves 2 4 2 Support for Dual Redundant Operation The Pigeon Point Shelf Manager can be configured with active backup instances to maximize availability Figure 2 shows how both Shelf Manager instances are accessible to the System Manager with primarily the active instance interacting at any given time Only the active instance communicates over IPMB 0 with the IPM Controller population in the shelf The two Shelf Manager instances communicate over TCP IP with the active instance posting incremental software level state updates to the backup instance via a Software Redundancy Interface SRI implemented between them As a result the backup can quickly step into the active role if necessary The SRI is a USB ora high speed UART based serial link depending on the
47. EVENTS has been specified in the Sensor Initialization field the following fields from the replacement SDR replace the corresponding attributes of the target sensor e For Threshold Based sensors e Lower Threshold Reading Mask e Upper Threshold Reading Mask e Threshold Assertion Event Mask e Threshold Deassertion Event Mask e Settable Threshold Mask e Readable Threshold Mask For Discrete sensors e Assertion Event Mask e Deassertion Event Mask e Discrete Reading Mask e The fields Sensor Units 1 Base Unit Modifier Unit cannot currently be replaced e The following sensor attributes are replaced from the replacement SDR fields if at least one of these fields is specified in the replacement SDR with a non zero value e Linearization M Tolerance B Accuracy Accuracy exp R exp B exp Analog Characteristic Flags Nominal Reading Normal Maximum Normal Minimum Sensor Maximum Reading e Sensor Minimum Reading e fand only if the symbolic constant THRESHOLDS has been specified in the Sensor Initialization field the fields listed below from replacement SDRs specify replacement threshold values for the target sensor Note however that only thresholds supported by the sensor implementation can be redefined Some local sensors do not support all possible thresholds Here is a list of replaceable threshold types all of which must be specified in the raw format with the Ox prefix Upper Non Recoverable Threshold
48. Erasing user data on candidate bank lt INFO gt backend c 860 Copy etc lt INFO gt shmm700_hal c 1161 Write confirmed 0 gt 0 ALLOW lt INFO gt shmm700_hal c 1166 Write candidate 1 gt 1 ALLOW lt INFO gt shmm700_hal c 1235 Write upgrade watchdog 0 gt 1 ALLOW At this point the telnet session is closed after a certain inactivity period after several seconds itis possible to reconnect to the target again and check the status of the reliable upgrade by invoking rupgrade s telnet 192 168 1 123 Trying 192 168 1 123 Connected to 192 168 1 123 Escape character is shmm700 login root Password rupgrade s lt INFO gt frontend c 87 Last upgrade log lt WARN gt shmm700_hal c 322 Device ubil boot is mounted to folder boot lt INFO gt shm 700_hal c 1255 Write storage state not available gt available ALLOW lt INFO gt shmm700_hal c 368 Resetting UBIFS cache lt INFO gt shmm700_hal c 1863 Clone file dev mtd2 of device 1 1 lt INFO gt shmm700_hal c 1642 Cleaning image var upgrade boot ulImage 1 lt INFO gt shmm700_hal c 1863 m m m lt INFO gt shmm700_hal c 1898 File dev mtd0 of device 1 0 cloned to m m lt INFO gt shmm700_hal c 1898 File var upgrade boot uImage 0 of device 2 0 cloned to file var upgrade boot uImage 1 of device 2 1 Release 3 2 0 242 July 31 2013 Pige
49. FTP server is shmm700 sentry shmm700 app The upgrade procedure is started from a telnet session Note Since only the application package image is explicitly updated the U Boot kernel and RFS images are automatically copied from the confirmed locations to the candidate locations Release 3 2 0 240 July 31 2013 Pigeon Point Shelf Manager User Guide The ShMM must be able to access the FTP server over the network that is its network adapter must be up and configured and a route must exist from the ShMM to the FTP server In the example below the ShMM is configured with the network address 192 168 1 123 which is in the same network with the FTP server telnet 192 168 1 123 Trying 192 168 1 123 Connected to 192 168 1 123 Escape character is shmm700 login root Password The parameters to rupgrade indicate that only the application package is being upgraded and that the copy protocol is FTP accessing a specified IP address and file with the user name admin and password ADMINPWD rupgrade skip checksums copy etc a ftp admin ADMINPWD 192 168 1 253 shmm700 sentry shmm700 app lt INFO gt frontend c 449 Downloading file ftp admin ADMINPWD 192 168 1 253 shmm700 sentry shmm700 app lt INFO gt frontend c 732 Initiating partial upgrade lt INFO gt frontend c 765 app tmp _shmm700_sentry shmm700 app lt INFO gt frontend c 784 copy etc lt INFO gt frontend c 820 Calling backend to
50. File shelfman conf acb3 gz 3 3 2 2 Using the Shelf Manager Command Line Interface to Place Configuration Variables in the Shelf FRU Information An extended version of the CLI command rudataw can be used to dynamically update Shelf FRU Information with configuration information The corresponding configuration file optionally compressed must be first downloaded to the ShMM The corresponding command line has the following syntax clia frudataw p lt ipmb address gt lt fru id gt lt in file gt clia frudataw p c lt ipmb address gt lt fru id gt The parameters lt ipmb address gt and lt fru id gt specify the IPMB address and FRU device ID of the FRU Information to update For the Shelf FRU Information use IPMB address 20h and FRU device ID 254 The parameter lt in file gt indicates the configuration file which is possibly compressed to be Stored in the target FRU Information The option c removes clears the configuration data from the target FRU Information For example the following command can be used to update the Shelf FRU Information with a new version of the configuration file clia frudataw p 20 254 shelfman conf acb3 gz Release 3 2 0 12 J uly 31 2013 Pigeon Point Shelf Manager User Guide 3 3 3 Verbosity Level Description This parameter controls the verbosity of output sent by the Shelf Manager to the system log This parameter also controls the verbosity of output to the console if the co
51. Number 3 The minimum number of seconds No DELAY between subsequent updates of the SEL file on the ShMM Flash Setting this number to a higher value causes SEL updates to be less frequent this reduces the total amount of data written to the Flash and thus increases the lifetime of the Flash but increases the probability of event loss if both redundant Shelf Managers go out of service SEL_HIGH_WATER Number 0 This value is the high watermark for Yes MARE the algorithm that controls automatic purging of the SEL if the actual percentage of free entries in the SEL falls below this value or the SEL overflows the Shelf Manager starts a thread that purges old records from the SEL in order of decreasing age SEL_LOW_WATERM Number 0 This value is the low watermark for Yes ARK the algorithm that controls automatic purging of the SEL if the thread that purges old records from the SEL starts it purges records until the percentage of occupied entries in the SEL falls below this value Release 3 2 0 63 J uly 31 2013 Pigeon Point Shelf Manager User Guide TYPE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO The time in seconds between No successive polls of local Shelf Manager sensors by the Shelf Manager SESSION_SEQUEN Number 128 This is the window of acceptable No ee WINDOW RMCP sequence numbers the wider this window is the more tolerant is the Shelf Manager to
52. Pigeon Point Shelf Manager User Guide The Shelf FRU Information in radial shelves must contain special records that identify the shelf as implementing a radial IP MB 0 topology and specify the routing of the IPMB links to the IPM controllers These records must conform to the ATCA specification section 3 8 1 1 Radial IP MB 0 is implemented differently for SAMM 500s and for ShMM 1500s For ShMM 500s there is a single logical bus for each of the two constituent buses of IPMB 0 that is IPMB A and IPMB B and radial IPMB is actually implemented in a pseudo radial in which the radial segments are only implemented at the physical level For ShMM 1500s there are distinct logical buses for each radial segment Support for radial IPMB is not yet implemented in the Linux kernel for the ShMM 700 though on the hardware level the same approach as on the ShMM 500 should work fine 4 5 1 Operation in ShMM 500 based Shelves with Radial IPMB 0 For pseudo radial IP MB 0 implementations on the ShMM 500 the Shelf Manager uses a single logical IPMB interface on each of IPMB A and IPMB B to communicate with all the IPM controllers but there is a separate GP 10 control on the carrier for each of the radial segments that can be used to turn that segment on and off Only some carriers provide this support If both 1 the Shelf FRU Information contains the radial IPMB 0 description record and 2 the carrier Supports radial operation the Shelf Manager on ShMM 50
53. Point Shelf Manager User Guide TYPE Boolean DEFAULT FALSE DESCRIPTION This parameter applies only if two network interfaces on the ShMM are used for RMCP communication If TRUE the two network interfaces on the ShMM are used in parallel mode if FALSE they are used in redundant mode CAN BE OBTAINED FROM SHELF FRU INFO Yes VERBOSITY Number 7 The Shelf Manager verbosity level No VERBOSITY_CONS Number N A The Shelf Manager verbosity level for No OLE console output VERIFY_SHELF_F Boolean TRUE Enable verification of checksums in No RU CHECKSUM Shelf FRU Information records if set to FALSE Shelf Manager ignores checksums WATCHDOG_ENABL Boolean TRUE This configuration parameter is used No BY to enable or disable the hardware watchdog timer When enabled the watchdog will cause the Shelf Manager to reboot if the watchdog expires WATCHDOG_TIMEO Number 2 The hardware watchdog timeout in No UT seconds If the software does not Strobe the watchdog at least once during this period the hardware resets the ShMM The supported range for this parameter is 2 30 seconds which is valid for all Supported ShMM platforms Increasing the default value may be necessary if additional software besides the Shelf Manager application runs on the ShMM increasing the CPU load Also see section 4 4 3 for the possible reasons to increase the watchdog timeout The downside of increasing
54. Release 2 7 3 e Section 3 3 adds new configuration parameter COOLING_NO_POWER_DOWN_IN_CRITICAL_ALERT e Section 3 15 adds more information about configuration of the SNMP service C 20 Release 2 7 4 e Section 3 3 adds new configuration parameter REDUNDANCY_COMPRESSION_THRESHOLD C 21 Release 2 8 0 e Section 3 3 adds new configuration parameters NO_MO_ Ml EVENT_AT STARTUP PET_OEM WITH _SEVERITY_STRING RUN_HPI_SNMP_SUBAGENT SPECIAL _HPI_ENTITY_FOR_AMC_ CARRIER e Section 3 3 adds a description of the corrupted_images variable Release 3 2 0 260 July 31 2013 Pigeon Point Shelf Manager User Guide Section 3 3 adds coverage of the etc localtime file Section 3 3 adds a description of double active state detection Section 4 12 adds descriptions of the ShMM tests accessible via Diagnostic Initiators Section 9 adds a description of HP l based upgrades C 22 Release 2 8 1 Section 3 3 adds a new configuration parameter HPDL_NETWORK_ELEMENT_ID_EEPROM_OFFSET Section 3 3 fixes a typo in the configuration parameter name SHELF_MANAGER_CONFIGURATION_IN_SHELF_FRU_INFO C 23 Release 2 8 2 Section 3 3 adds a new configuration parameter DHCP_SPECIAL_CLIENT_ID_FORMAT C 24 Release 3 0 0 Section 2 adds a description of the ShMM 700 Section 2 2 updates Figure 1 Section 2 4 1 adds HPDL coverage Section 2 4 2 updates Figure 2 Section 2 4 4 adds text and a new column for the ShMM 700 in Table 3
55. SNMP server boa HTTP server and shel f man the Shelf Manager after Linux boots If the U Boot variable start_rc2_ daemons is set to y the secondary RC script should also define command line options for automatic invocation of the shelfman daemon It may also provide other configuration services 3 3 Setting Up Shelf Manager Configuration File The Shelf Manager configuration file she1fman conf is located in the etc directory Each line in the file is either a comment line starting with or a lt name gt lt value gt pair representing the assignment for the configuration parameter The name and the value are Separated with the equal sign The configuration parameter name is case insensitive Each configuration parameter is one of the following types Boolean number string or IP address The values of string type of configuration parameters are case sensitive The format of the value conforms to the type of the configuration parameter as shown in the following table Table 6 Configuration Parameter Types and Descriptions CONFIGURATION PARAMETER DESCRIPTION TYPE Boolean A Boolean can be represented by either the strings FALSE false Of TRUE true or by their numerical representations of O and 1 respectively Number A whole possibly signed numeric value hexadecimal notation 0x is also Supported String A string quoted always with double quotes or unquoted Quoted strings may contain
56. The TIMESERVER variable can be changed by the Shelf Manager if the Shelf Manager is configured to receive its RMCP IP address via DHCP In that case depending on the configuration of the DHCP server the Shelf Manager can also receive the IP address of the NTP server via DHCP It stores this address in the tmp timeserverip file which is used by the script etc timesync to override the current value of the TIMESERVER variable Note that during each reboot of the Shelf Manager the clock is reset to J anuary 1 1970 Therefore the system time stays incorrect until the script etc timesync succeeds in obtaining correct time from the time server The variable timezone contains the name of the current time zone followed by its offset from Greenwich Meridian Time GMT The offset is positive for time zones to the west of Greenwich and negative for time zones to the east of Greenwich This variable is propagated to the Linux level as the environment variable Tz The default value of this variable is UTC i e Universal Coordinated Time which matches Greenwich time Note that textual timestamps generated on a ShMM correspond to the local time zone That is these timestamps include the offset defined by the environment variable Tz The time sent by time servers is GMT time if the time zone on the Shelf Manager is not set or not Set correctly the time obtained from the time server is interpreted incorrectly The three letter name of the time zone is not u
57. User Guide TEST NAME TEST PARAMETERS NAME TYPE TEST DESCRIPTION 1 DESCRIPTION ETHER INT 2 mac INT32 MAC ID to The RX and TX channels are Internal test internally looped back while the test is Ethernet Acceptable values are 0 in progress which may result in data Loopback and 1 loss and or a broken network Test Default value is 0 connection ETHER PHY 3 mac INT32 MAC ID to The RX and TX channels are Ethernet PHY test internally looped back within the PHY Loopback Acceptable values are 0 while the testis in progress which Test and 1 may result in data loss and or a Default value is 0 broken network connection ETHER ARP 4 mac INT32 MAC ID to The test sends an ARP request with Ethernet ARP test the specified IP address and waits for Test Acceptable values are 0 a reply in the specified timeout range and 1 Default value is 0 ip INT32 IP Address to resolve during the test timeout INT32 Timeout value in seconds for the test FLASH CRC 5 mtd INT32 Software The test calculates a checksum for Flash image number MTD the software image for ShMM Checksum partition number for ShMM 500 1500 located in the Test 500 1500 for the test corresponding MTD partition and Acceptable values are compares it with the checksum in the e 2 Active Linux kernel image header e 3 Active U Boot e 4 Active RFS e 7 Inacti
58. _ubifs_ al t Pigeon Point Shelf Manager User Guide DESCRIPTION ShMM 700 only U Boot command that loads alternative Linux kernel and RFS images from the Flash drive into RAM Images are selected on the basis of the alt_image variable value The kernel image is loaded at the Lloadaddr address and the RFS image is loaded at the rfsaddr address This command is invoked by ubifs command if one of the images from the current image set Selected by the image_sel value is broken and there is no reliable upgrade running log_max Specifies the size limit for the syslog file in bytes Default is 250000 When the log file size reaches the limit the file is renamed as a backup file and a new syslog file is started replacing any existing backup so that the maximum space used by the overall syslog is twice the log_max value log_remote Specifies the IP address for the remote syslog facility By default this option is not set The syslog daemon on Linux systems can be configured to receive messages from remote hosts normally using the x option so using the Log_remote Setting it is possible to send the ShMM system log to a remote system logging Specifies if messages log file should be maintained in ram or Flash Default is ram which is the recommended option mac_override ShMM 700 only This variable allows overriding MAC address configuration for both Ethernet controllers for the Linux kernel This variable is us
59. any given time RMCP communication passes through only one adapter initially this iis RMCP_NET ADAPTER However if the Shelf Manager detects that the adapter currently used for RMCP communication becomes physically disconnected from the network link broken it automatically switches to the other alternate adapter The first adapter is turned off the RMCP IP address is transparently moved to the other adapter and three ARP notifications are broadcast to notify other systems about the MAC address change This change is transparent for the System Manager and does not break existing RMCP connections However if the Shelf Manager detects that the other network adapter is also physically disconnected from the network it does not perform the IP address switchover described above but performs a full switchover to the backup Shelf Manager A full switchover is also performed if the Shelf Manager detects physical disconnection of the network adapter used for RMCP communication in non cross connect configurations Detection of physical disconnection of the RMCP network adapter is controlled by the Shelf Manager configuration parameter SWILTCHOVER_TIMEOUT_ON_BROKEN_LINK The Release 3 2 0 75 July 31 2013 Pigeon Point Shelf Manager User Guide value of this parameter is the time interval in seconds during which the adapter stays physically disconnected before the Shelf Manager performs an IP address switchover or full switchover Detection and s
60. available for assignment of LAN configuration parameters by the Shelf Manager Channel numbers each in the range 1 to 7 can be separated by commas spaces or any other Separators for example 3 4 5 LAN parameters are assigned only to channels specified in this list and in the order in which they appear in this list BOARD_LAN_PARA String 256 This string value represents the listof Yes METERS SYNCHRO descriptors for FRUs that require iii synchronous assignment of LAN configuration parameters Itis meaningful only if the variable BOARD_LAN_PARAMETERS_US E_DHCP is TRUE The detailed syntax for this value is defined in section 4 10 4 Release 3 2 0 41 J uly 31 2013 BOARD_LAN PARA METERS_USE_DHC P Pigeon Point Shelf Manager User Guide TYPE Boolean DEFAULT FALSE DESCRIPTION If TRUE use DHCP to retrieve board module LAN configuration parameters If this parameter is FALSE the following options are available e Ifthe Shelf FRU Information contains one or more Board AMC LAN Configuration Parameters multirecords use these to retrieve the board module LAN configuration parameters e Otherwise if the LAN configuration parameters file exists on the ShMM use this file to retrieve the board module LAN configuration parameters e Otherwise Shelf Manager assignment of board module LAN configuration parameters is not available CAN BE OBTAINED FROM SHELF FR
61. blanks unquoted strings are terminated by the first blank The maximum string size is specified separately for each string oriented configuration parameter These values are case sensitive IP address An Internet Protocol address in decimal dot Xxx XXX XXX XXX notation Release 3 2 0 34 July 31 2013 Pigeon Point Shelf Manager User Guide Itis possible to specify a value of an environment variable as a configuration parameter value using the notation lt envvar gt in that case the value of the variable lt envvar gt is substituted when the configuration file is read Here is an example DEFAULT_RMCP_IP_ADDRESS SIPADDR After the Shelf Manager has been brought up for the first time the IP addresses are stored with the IPMI LAN configuration parameters The LAN configuration parameters can be accessed or modified via any of the RMCP CLI web or SNMP external interfaces and take precedence over the shelfman configuration file when the Shelf Manager is restarted This is to ensure the persistency of any modifications that are made to the LAN IP Addresses and gateway via those interfaces If the Shelf Manager IP Connection record in the Shelf FRU Information contains an IP address it takes precedence over all other settings of the external or RMCP IP address Preferably the Shelf FRU Information should either not specify this address atall orsetitto 0 0 0 0 to ensure that addresses can be controlled through
62. boot up These names are separated by space characters These tests do not take much time and can be run on a regular basis The value of the environment variable post_poweron contains names of tests that are executed after power on reset only vs on each boot up These names are separated by space characters As the POST tests are executed the results are logged in a textual form in a dedicated area in SDRAM Results for each particular test have the following form lt 4 gt POST lt test name gt lt test specific output gt PASSED FAILED The POST framework provides interfaces for accessing the results of the POST tests in U Boot and in Linux U Boot supports a log show command which can be used to access the POST test results This command outputs the contents of the POST log buffer onto the serial console shmmxx00 log show lt 4 gt POST uart UART 0 test failed lt 4 gt FAILED lt 4 gt POST cre PASSED In a POST enabled configuration the Linux kernel shares its internal message log buffer with the POST log buffer This causes POST results to be automatically displayed on the serial console during the kernel bootstrap Interactively the user can access the kernel log buffer and thus the POST test results using the dmesg command implemented by busybox The defined names for POST tests are e memory SDRAM tests recommended for execution on the first boot up after power on e crc Flash checksum
63. broken If the link remains broken for at least the number of seconds given in this parameter a switchover takes place if the link is restored during this timeout period no Switchover takes place If the value of this parameter is 1 no automatic Switchovers take place on broken RMCP links CAN BE OBTAINED FROM SHELF FRU INFO Yes SYSLOG_LOGGING _ENABLED Boolean TRUE Output log messages to the system log SYSTEM_MANAGER _TRUNCATES_SEL Boolean FALSE If TRUE the Shelf Manager algorithm for truncating the SEL automatically is disabled the System Manager is responsible for truncating the SEL by monitoring the value of the sensor SEL State of the type Event Logging Disabled on the Shelf Manager and removing events from the SEL by sending the Delete SEL Entry command to the Shelf Manager TACHOMETER_THR ESHOLD_UPDATE_ DELAY Number 15 This parameter applies to fan trays managed by the Shelf Manager and implementing adaptive setting of tachometer thresholds depending on the current fan level This parameter Specifies the delay in seconds between setting a new fan level that is higher than the current fan level and increasing the tachometer threshold value the delay is needed to allow the fans to reach the higher speed Release 3 2 0 67 July 31 2013 TASKLET_RETRIE S Pigeon Point Shelf Manager User Guide Number DEFAULT DESCRIPTION Th
64. corresponding parameter is set to 0 0 0 0 inthe IPMI LAN Configuration Parameters for channel 1 If a non zero gateway IP address is provided in the LAN Configuration Parameters the value provided in the Shelf Manager configuration file is ignored CAN BE OBTAINED FROM SHELF FRU INFO No DEFAULT_GATEWA Y_IP_ADDRESS2 IP address None The default IP address used for the gateway for shelf external RMCP based communication on the second network interface if the corresponding parameter is set to 0 0 0 0 inthe IPMI LAN Configuration Parameters for channel 2 Ifa non zero gateway IP address is provided in the LAN Configuration Parameters the value provided in the Shelf Manager configuration file is ignored No DEFAULT_RMCP_1I P_ADDRESS IP address None The default IP address used for shelf external RMCP based communication it is switched over between the redundant instances of the Shelf Manager This IP address is used only if the corresponding parameter is setto0 0 0 0 inthe IPMI LAN Configuration P arameters for channel 1 and in the Shelf Manager IP Connection record in Shelf FRU Information If a non zero IP address is provided in the LAN Configuration Parameters and or Shelf FRU Information the value provided in the Shelf Manager configuration file is ignored No Release 3 2 0 45 July 31 2013 DEFAULT_RMCP_TI P_ADDRESS2 Pigeon Point Shelf Manager User
65. design v0 44 0 0 A2F A2F firmware version 0 9 A2F Last reset cause HARD A2F Device type A2FO60M3E FG256 A2F MSS clock frequency 40 MHz A2F Fabric clock frequency 20 MHz A2F Fast delay calibration 1330 cycles per 100uS A2F eNVM 128 KB 00000000 00020000 A2F eSRAM 16 KB 20000000 20004000 A2F Extram start 200022D0 RUPG booting from image 1 candidate Here U Boot reports that a reliable upgrade is in progress and the candidate image set is being used The candidate image set index is 1 and the confirmed image set is 0 Release 3 2 0 245 July 31 2013 Pigeon Point Shelf Manager User Guide In serial Out serial Err serial Net FECO 00 18 49 01 8f 26 FEC1 00 18 49 01 8 f 27 FECO FECL Hit any key to stop autoboot 2 Power is turned off here After some time power is turned back on The candidate Flash state has been lost because of the power loss so the ShMM reverts back to the confirmed Flash PowerPrep start initialize power Battery Voltage 1 32V No battery or bad battery detected Disabling battery voltage measurements Mar 21 201223 23 40 FRAC 0x92926152 memory type is DDR2 ait for ddr ready 1 power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory access ddr2 0x40000000 finish simple test U Boot 2009 08 Mar 21 2012 23 22 30 Freescale i MX28 family CPU 297 M
66. either toggle the least significant bit in the IP address for usb1 on the Shelf Manager with the even hardware address or toggle the least non zero bit of the network mask in the IP address for usb0 on the Shelf Manager with odd hardware address The result is the same This guarantees that usb0 on Shelf Manager with the odd hardware address and usb1 on Shelf Manager with the even hardware address are in the same logical network and are not equal Here is an example of deriving IP addresses for the USB network interfaces under the assumption that the following definitions are in etc shelfman conf REDUNDANCY_IP_ADDRESS 192 168 1 2 REDUNDANCY_NETMASK 255 255 255 128 The least significant non zero bit in the network mask is the 7th bit where smaller bit numbers are less significant To toggle this bitin an IP address itis sufficient to add 128 if this bit is set to zero in the IP address or subtract 128 if this bit is set to 1 in the IP address To toggle the least significant bit in an IP address itis sufficient to add 1 if the IP address is even or Subtract 1 if the IP address is odd Since REDUNDANCY_IP_ADDRESS is even the computations are the same whether the file etc readhwaddr is present or not On the ShMM with the even hardware address the assignment of IP addresses looks like this e usb0 192 168 1 2 no changes e usbl 192 168 1 130 toggling the least significant non zero bit of the netmask On the ShMM with
67. fw revision is absent the Auxiliary Firmware Revision is not defined According to IPMI v2 0 R1 0 Section 20 1 Get Device Id Command the Auxiliary Firmware Revision is a 4 byte data item The file var nvdata aux fw revision should contain the eight hexadecimal digits that represent the 4 bytes in question No separators are allowed The first two hexadecimal digits represent the most significant byte of the Auxiliary Firmware Revision For example assume that the file var nvdata aux fw revision contains the string adblefcd When the Shelf Manager is started and the RMCPTA connection is established it is possible to obtain the Auxiliary Firmware Revision via RMCP The parameter lt IPMB address gt below represents the IPMB address of the alternative controller Here is a dialogue with the Pigeon Point internal tool RMCPTA but any RMCP client can be used to make this query RMCPTA 1 gt TargetFwd lt IPMB address gt RMCPTA 1 gt GetDeviceld Completion Code 0x00 OK Device ID 0x00 Device Revision 0x0 Device Mod normal operation Device SDR present Firmware Rev 2 30 IPMI Version 1 5 Device Support IPMB Req Gen FRU Sensor anufacturer ID 0x0400A Product ID 0x0000 AUX FW Rev OxAOBIEFCD AO Bl EF CD 3 10 Setting Up the Clock When the system is brought up for the first time the clock is not set and must be initialized Initially the clock is set to J anuary 1 1970 The
68. iaa aE A A EE NEEE Aaa Aai 258 GAS RELEASE 2 6 4 A A A 258 G4 RELEASE 2 6 4 A A A 259 GAB RELEASE 2 6 4 e EA E A laggauiy ayueesescieaanseentenebasiensle 259 Release 3 2 0 5 July 31 2013 C 16 C 17 C 18 C 19 C 20 C 21 C 22 C 23 C 24 C 25 C 26 Pigeon Point Shelf Manager User Guide RELEASE 2 V O miaire eere d tat laden exces a a exceed ca bauladehsnnced aaaea aa a eaa 259 RELEASE 2 Va lommen a ree e a a a Aaaa aAa aaae ESA 260 RELEASE 2 V2 rompe ne e aa e a a E aie aeaa Aa EE E NaS 260 RELEASE 2 O Mer eee TAEAE E ETa E A naaa Aaa A EEE ES 260 RELEASE 227 Aronian i a a aa Aa AOne RA Aaa 260 RELEASE 2 8 0 rirse iaaea aa aia aia aaa eaaa Aar 260 RELEASE 2 8 loonie aradin i a canes aa Aa OaE eaaa A RANEE 261 RELEASE 2 622 wus vostcnenenscstuanaciiecte aged cauvaiactaezcend ca svanandti czcedcetyatideh seated a taann dit ceadedcntaadidaa seve 261 RRELEASE 3 2020 oinssocea ci vicnsesinanscdeceecus de duawndth va qend deuauanddeez ced ct duaancuhauiieadoniuaancih ceeded a Aaa 261 RELEASE Sele vacseesccccuesvandlanes seacuanssausanneaccenancenticusisdeceniavearaagesacdauuavardlunas deceneavetunaunsecneea 262 RELEASE 3 2 O iasesesiicccsivsainznssaccincnwastaabas deceuenceataauataccepcavestuasecaccepiaveedannecdecencacastaanecdeneis 262 Release 3 2 0 6 July 31 2013 Pigeon Point Shelf Manager User Guide Figures Figure 1 Management Aspects and Potential Pigeon Point Product Sites in an Example AdvancedTCA Shelt cccccsccecsceec
69. image in the provisional U Boot firmware image partition dev mtdchar3 e k Upgrade the Linux kernel image in the provisional Linux kernel image partition dev mtdchar2 e x Upgrade the root file system image in the provisional root file system image partition dev mtdchar4 lt src gt specifies an upgrade image file to be copied to the provisional Flash partition specified by lt dst gt The specifier a if used indicates that all three components U Boot kernel and root file system are to be upgraded and the source upgrade image files must have standard names These standard names are defined separately for ShMM 500 and ShMM 1500 according to the following table Release 3 2 0 204 July 31 2013 Pigeon Point Shelf Manager User Guide Table 24 Standard Source File Names for Upgrade Components COMPONENT NAME SOURCE FILE NAME SOURCE FILE NAME SHMM 500 SHMM 1500 U Boot sentry shmm500 u boot sentry shmm1500 u boot Linux kernel sentry shmm500 kernel sentry shmm1500 kernel RFS sentry shmm500 rfs sentry shmm1500 rfs The image upgrade works as follows For each of the specified source images the image is copied to the ShMM using the specified copy protocol If no a specifier is supplied the image is first copied to the RAM file system of the ShMM specifically the copy is to the tmp directory and then moved to Flash thatis copied to the destination partition in Flash and then removed from
70. images are stored in a file system and normally this is something similar to run mount_ubifs a2f upgrade run ubifs bootdelay Autoboot delay value in seconds Default setting is 3 Release 3 2 0 25 July 31 2013 Pigeon Point Shelf Manager User Guide ENVIRONMENT DESCRIPTION VARIABLE boot file Parameter that specifies what kernel image should be used by the net and nfs boot options console Setting for the kernel and init script console port and baud rate On ShMM 500 and ShMM 1500 the default is tt ySO 115200 On ShMM 700 the default is tt ySPO 115200n8 corrupted_ima ges This U Boot managed counter records the number of times that a corrupted boot image is discovered When U Boot fails to boot the kernel or the configured RFS image due to corruption of the boot image in Flash U Boot executes a recovery procedure that essentially designates the previously provisional candidate Flash device as the current persistent confirmed Flash device increments this variable and proceeds with the boot process The recovery procedure is not performed if the reliable upgrade WDT is active or if the corrupted boot image is not located in Flash Support temporarily limited to ShMM 500 1500 ethact ShMM 700 only Parameter that specifies which Ethernet controller Should be used by U Boot when handling bootp dhep nfs ping rarpboot tftpboot commands Supported values are FECO FEC1 FECO selects eth
71. installed in the shelf when the shelf is in the critical alert cooling State by default it is not powered up and stays in M3 until the critical alert goes away This Release 3 2 0 107 July 31 2013 Pigeon Point Shelf Manager User Guide behavior can be overridden by the configuration variable ALLOW_POWER_UNRELATED_FRU_IN_CRITICAL_STATE If this variable is set to TRUE a new FRU can be powered up if 1 the critical alert state is caused by a temperature sensor that belongs to a different FRU and 2 the critical alert state is not caused by a shelf wide temperature sensor In addition a fan management strategy is implemented that sets the fan level to the maximum for all fans in the cooling zone in the following cases If some fans are missing in the shelf based on the fan population specified in the Address Table in the Shelf FRU Information all fan trays that cool the same zone s as the missing ones are set to maximum If one or more of the fan tachometer sensors have a major or critical threshold crossed a fan is stopped or rotates too slowly all fan trays that cool the same zone s as the underspeed fan s are set to maximum speed The number of simultaneously underspeed fans that cause the remaining fans to go full speed is configurable via the configuration parameter TURBO_MODE MIN FAN FAILURES and the default value is 1 Setting this parameter to 0 turns off this feature The default cooling management strategy optionally su
72. interface FEC1 selects eth1 interface Default value is FECO ethaddr MAC address of the primary on chip Ethernet controller The value of this variable is set automatically by U Boot This address is passed to the kernel Ethernet driver ethladdr MAC address of the secondary Ethernet controller The value of this variable is set automatically by U Boot On ShMM 500 and ShMM 1500 this address is passed to the kernel Ethernet driver On ShMM 700 this address is not passed to the kernel Ethernet driver directly but both the U Boot and the Linux kernel FEC driver set up the MAC address for the second Ethernet controller as ethaddr incremented by 1 so both U Boot and Linux use the same MAC for the second Ethernet controller flash reset Instructs Linux to erase the Flash filesystems etc and var restoring to factory default y n The system startup script sets this variable back to n after the Flash erase Default is n gatewayip Default gateway IP address This variable can be passed as a part of the kernel command line to automatically configure routing for the network interfaces Default setting 192 168 0 1 hostname Network host name default is shmm500 for ShMM 500 shmm1500 for ShMM 1500 and shmm700 for ShMM 700 image_sel ShMM 700 only This variable contains the current image set index The variable is used in the load_ubifs command to determine which of the Linux kernel and RFS images
73. is located by default in usr local 1ib libopenhpi so to link to client applications Note The OpenHP library Libopenhpi so 2 from the Pigeon Point OpenHP 2 12 0 1 release or from the Pigeon Point OpenHP 2 14 0 0 release and higher is strongly recommended for use with client HPI applications since it contains the redundancy framework improvement to avoid the potential loss of an HPI call that is being processed during a Shelf Manager switchover The OpenHPI library Libopenhpi so 2 inthe Shelf Manager RFS already contains this improvement Also note that the posted open source OpenHPI versions 2 16 and higher already include the necessary library updates 3 15 Configuring System Services The Shelf Manager filesystem contains several network services that can be used for debug and upgrade purposes The following table contains a list of ports used by the Shelf Manager with a brief description of the service configuration with each port Release 3 2 0 124 July 31 2013 Pigeon Point Shelf Manager User Guide Table 11 Network Services Port NUMBER TYPE NETWORK SERVICE DESCRIPTION 21 tcp FTP server Activated via the etc inetd conf file to disable the FTP server comment out the tp line in that file This service is based on WU FTP daemon so the configuration is located in the etc tp files mostly in etc ftpaccess 22 tcp SSH server Activated via the etc inetd conf file to disable the SSH server comment out the ssh
74. line in that file Run sshd help to see the list of supported arguments that can be specified 23 tcp Telnet server Activated via the etc inetd conf file to disable the Telnet Server comment out the telnet line in thatfile Runtelnetd help to see the list of supported arguments that can be specified 80 tcp HTTP server with the ShM Web interface Activated by the etc re common Script if the U Boot variable start_rc2_ daemons is Set to y to disable the HTTP server comment out the corresponding line in the script The HTTP server Supports the configuration file etc httpd conf file where it is possible to specify allowed and denied IP ranges and the access login and password for example A 192 16 Allow any address that begins with 192 16 D Deny from other IP connections cgi bin username password where password is generated via the httpd m password command 161 udp SNMP server with ShM SNMP interface Activated by the etc re common Script if the U Boot variable start_rc2_ daemons is setto y to disable the SNMP server comment out the corresponding line in the script The SNMP server Supports the configuration file etc snmpd conf The validation criteria for incoming SNMP requests are hardcoded in this configuration file For SNMP v1 v2c the community string in an incoming request must be public For SNMP version 3 requests the user name in an incoming request must be overlord These parameters can however
75. modules currently only AMC modules are considered implement LAN IPMI channels there may be a need for a centralized repository of LAN configuration parameters mainly the IP addresses subnet masks and gateway addresses for these boards and modules This repository can be maintained on the Pigeon Point Shelf Manager The parameters are stored statically on the ShMM or in the Shelf FRU Information Also the Shelf Manager supports obtaining parameters from a DHCP server using the DHCP protocol One use for such LAN IPMI channels once configured is to support serial over LAN sessions via the management controllers on the boards or modules This use involves a significant number of additional IP MI parameters In the current Shelf Manager only the following three parameters are Supported IP address subnet mask and default gateway address Other parameters for instance MAC addresses may be added in future releases of the Shelf Manager 4 10 1 Structure and Composition of the Supported Parameters The parameters are specified on a per slot basis Slots are defined for both main boards and modules For main boards the slot number is the board physical slot number site number For AMC modules the slot number is the combination of the AMC carrier physical slot number and AMC site number For each slot several parameter sets may be defined Each parameter set contains parameters currently the IP address subnet mask and default gateway address f
76. of the pool of IP addresses available to the DHCP server for allocation in configurations where this pool is relatively small 2 Read from a Flash File on the ShMM After the Shelf Manager starts up it retrieves the LAN configuration parameters from a file on the ShMM var nvdata subsidiary lan_param This is the simplest approach to retrieve the parameters The file has a text format Each line defines one parameter set and has the following format lt physical slot gt lt amc slot gt lt ip address gt lt subnet mask gt lt default gateway ip address gt Comment lines starting with are allowed and ignored Also the rest of the line after a character is treated as a comment and ignored Empty lines are allowed and ignored too For each slot the file can define several parameter sets All parameter sets defined for the same Slot must be placed in the file together and have the same values to the left of the equal sign They are assigned to LAN channels on the corresponding boards or modules that require such assignments 3 Read from the Shelf FRU Info In this case there is a set of custom multirecords with Manufacturer ID 00400Ah the PPS IANA Record ID 0Ch The concatenation of the data in these multirecords optionally compressed with gzip represents the parameter data in the same format as for the text file above Management of parameter sets is also performed in the same way as for the previou
77. owner 220 waiting 223 Release 3 2 0 153 J uly 31 2013 Pigeon Point Shelf Manager User Guide lt W gt 14 09 43 866 199 ipmf_watchdog_thread_proc a potential deadlock was detected lt I gt 14 09 44 307 199 Write 148724 bytes to the var nvdata dumplog bin 1 seconds lock offset 148232 lock Ox2acd3940 owner 208 waiting 209 lock Ox2ac432f0 owner 223 waiting 217 214 220 208 lock Ox2ac43650 owner 220 waiting 223 lt W gt 14 09 44 616 199 ipmf_watchdog_thread_proc a potential deadlock was detected lock Ox2acd3940 owner 208 waiting 209 lock Ox2ac432f0 owner 223 waiting 217 214 220 208 lock Ox2ac43650 owner 220 waiting 223 lt W gt 14 09 44 926 199 ipmf_watchdog_thread_proc a potential deadlock was detected lock Ox2acd3940 owner 208 waiting 209 lock Ox2ac432f0 owner 223 waiting 217 214 220 208 lock Ox2ac43650 owner 220 waiting 223 lt W gt 14 09 45 236 199 ipmf_watchdog_thread_proc a potential deadlock was detected lock Ox2acd3940 owner 208 waiting 209 lock Ox2ac432f0 owner 223 waiting 217 214 220 208 lock Ox2ac43650 owner 220 waiting 223 lt W gt 14 09 45 546 199 ipmf_watchdog_thread_proc a potential deadlock was detected For postmortem analysis of a deadlock the utility dumplog Supplied in the Shelf Manager RFS can be used The following command line invokes this utility dumplog f var nvdata dumplog bin The result is produced o
78. parameter file are non zero The channel configuration parameters for the channel 2 are stored in the files var nvdata ch2_param and var nvdata ch2_param_v2 on the ShMM the second file is used only with the Shelf Manager release 2 6 and higher Release 3 2 0 Ty July 31 2013 Pigeon Point Shelf Manager User Guide Starting with release 2 5 2 the Shelf Manager recognizes a second instance of the Shelf Manager IP Connection record in the Shelf FRU Information as the source of the IP address netmask and default gateway for the second RMCP network interface Therefore it is possible to specify RMCP addressing information for both interfaces in the Shelf FRU Information In active active mode the IP addresses on both network interfaces are switched over to the backup ShMM as a result of a switchover 3 4 2 3 Support for Site dependent ShMC Cross connects According to the ECN 3 0 2 0 002 for revision 2 0 of the ATCA specification ShMC cross connects can be implemented with either a site independent or a site dependent configuration as shown in Figure 3 Figure 3 Implementation Options for ShMC Cross connects ShMC site ShMC site independent dependent recommended for recommended for 2 pair only 2 4 pair adaptable In the site dependent variant after a switchover in the default Shelf Manager configuration each Specific RMCP address is propagated to the corresponding network interface eth0 or eth1 on the other ShMC and
79. parameter supplied the script erases both the provisional etc and provisional var directories then copies Shelf Manager non volatile information from the directory var nvdata to the provisional var partition The script also copies the contents of etc ssh directory to the provisional etc partition to preserve SSH server keys This is the default mode of operation in this case the non volatile data is preserved but the Shelf Manager configuration file is taken from the new RFS image e conf the scriptadopts the default mode of operation except that Shelf Manager configuration files are preserved via var upgrade filesystem During the next boot configuration files are copied from var upgrade to the etc partition e erase the script erases both the provisional etc and provisional var directories they are restored from the RFS default values during the next boot the current Shelf Manager non volatile data and configuration are not preserved The only preserved configuration in this mode is the etc ssh directory which contains SSH server keys that are copied to the provisional etc partition The current Shelf Manager non volatile data and configuration are not preserved e etc_copy the script erases both the provisional etc and provisional var directories then it copies the contents of etc and the non volatile information from the directory var nvdata to the provisional Flash partitions In this case both the non volatile d
80. possible A non Ethernet link for SRI purposes was introduced with the ShMM 500R and carried forward to the SnMM 1500R and ShMM 70OR Given this history a non Ethernet SRI is often referenced elsewhere in this document as an Alternate S oftware Redundancy Release 3 2 0 17 J uly 31 2013 Pigeon Point Shelf Manager User Guide Interface This non Ethernet SRI is implemented via a pair of cross connected USB links on the ShMM 500R and ShMM 700R and with a high speed UART based link on the ShMM 1500R The HRI ensures that three key items of coordination status are available to each Shelf Manager instance e Presence each Shelf Manager instance knows whether the other instance is present in the shelf e Health each instance knows whether the other instance considers itself healthy e Switchover Request the backup instance can force a switchover if necessary The HRI is implemented differently in each of the three ShMM families Figure 2 shows other aspects of the architecture and context of the Pigeon Point Shelf Manager as well These include the Shelf Adaptation Layer and the shelf description that drives it and the main Shelf Manager subsystems that implement the features summarized in Section 2 4 1 2 4 3 System Manager Interface Another major subsystem of the Pigeon Point Shelf Manager implements the System Manager Interface System Manager is a logical concept that may include software as well as human operators in
81. power consumption of the FRU by assigning it the next lower power level In the critical alert cooling state non recoverable thermal thresholds are crossed for one or more sensors the Shelf Manager sets the fan level to maximum for the fans that serve the cooling zone s where those thresholds have been crossed In addition if the thermal condition is caused by a specific FRU the FRU is powered down If the thermal alert is caused by a shelf wide temperature sensor all FRUs are powered down as prescribed by the PICMG 3 0 specification After a FRU is powered down its further handling by the Shelf Manager depends on the value of the configuration variable COOLING_KEEP_POWERED_OFF_FRUS_IN_M1 If this variable is set to FALSE which is the default the FRU is activated but kept in state M3 until the thermal alert goes away The cooling state in which the FRU can be powered back is specified by the value of the configuration variable REAPPLY_POWER_MAX COOLING_STATE By default this state is Normal when the shelf cools down sufficiently and the shelf cooling state reaches the specified state the Shelf Manager powers the FRU back on Otherwise if the configuration variable is set to TRUE the FRU stays in the state M1 indefinitely and is not automatically powered back on when the thermal alert goes away in that case an intervention by the System Manager or by an operator is needed to activate and power on the FRU Also when a new FRU is
82. protocol was specified by the Service Availability Forum SAF http www saforum org some time ago However currently neither the mapping definition nor the subagent software itself is actively supported In order to automatically start the HPI SNMP subagent at system startup the Shelf Manager configuration variable RUN_HPI_SNMP_SUBAGENT should be setto TRUE 3 14 3 IntegralHPI Client Configuration The HP interface is a C language function call interface To use it across a network boundary a remote procedure call RPC protocol must be used IntegralHP I is compatible with the RPC protocol implemented in OpenHP www openhpi org This means that OpenHP client utilities and libraries can be used to communicate with IntegralHP in the same way as they can communicate with an OpenHP I daemon IntegralHP accepts an OpenHPI library connection to the TCP port 4743 which is the default for an OpenHPI daemon Currently the recommended approach for client configuration is the following e Download OpenHP distribution release 2 10 2 12 2 14 or 2 16 from www openhpi org configure it build and install on the client system If you have access to Pigeon Point OpenHPI Pigeon Point s distribution of OpenHP for either of the above OpenHPI releases you can use it instead of the original distribution e Setthe environment variable OPENHPI_DAEMON_HOST to the RMCP IP address of the target Shelf Manager e Use the OpenHP client library which
83. requests the switchover the backup Shelf Manager agrees to it if its physical network link is present only after this number of seconds passes following the first request from the active Shelf Manager A non zero value for this parameter can be used to prevent switchovers in scenarios where a hub board is inserted into the Shelf that activates the active Shelf Manager link several seconds after the backup Shelf Manager link CAN BE OBTAINED FROM SHELF FRU INFO Yes SWITCHOVER_ON_ HANDLE_OPEN Boolean FALSE If TRUE Switchover related behavior of the active Shelf Manager is affected by the Hot Swap state of its physical IPM controller as follows If the physical IPM controller on the active Shelf Manager goes to state M1 due to its Hot Swap Handle being open or due to programmatic deactivation a switchover to the backup Shelf Manager is initiated If the physical IPM controller on the active Shelf Manager goes to state M5 and there is no available backup Shelf Manager the physical IPM controller is not deactivated and stays in M5 indefinitely Yes Release 3 2 0 66 July 31 2013 SWITCHOVER_TIM EOUT_ON_BROKEN _LINK Pigeon Point Shelf Manager User Guide Number DEFAULT DESCRIPTION This parameter affects when or whether the Shelf Manager initiates a Switchover when the physical network link between the Shelf Manager and the System Manager the RMCP link is
84. sending switchover requests and the switchover won t happen Otherwise if the physical link to the active Shelf Manager remains broken after the expiration of this delay the switchover will finally take place The value of the delay is specified in the configuration variable SWITCHOVER_ON_BROKEN_LINK BACKUP_DELAY Usually the parameter RMCP_NET_ADAPTER2 is assigned the value eth1 while the value of the configuration parameter RMCP_NET_ADAPTER the main network adapter used for RMCP communication is eth0 as in the following sample RMCP_NET_ADAPTER etho RMCP_NET_ADAPTER2 ethi However other configurations are possible For example values ethO 1 and eth1 1 canbe used if additional permanent IP addresses need to be assigned to both network interfaces The two network interfaces are used in active standby mode if the configuration parameter USE_SECOND_CHANNEL is set to FALSE If the Shelf Manager configuration parameter PROPAGATE_RMCP_ADDRESS is defined as TRUE active standby management also applies to the network interface on the backup Shelf Manager which is assigned the RMCP derived IP address on the backup Shelf Manager That is if the backup Shelf Manager detects that the currently used adapter becomes physically disconnected from the network link broken it automatically switches to the other alternate adapter The first adapter is turned off and the RMCP derived IP address of the backup Shelf Manager is
85. stored in the FRU Information as a sequence of OEM type multirecords similar to the multirecords defined in the AdvancedTCA specification These multirecords have the Pigeon Point manufacturer ID 00040Ah and the manufacturer specific record type is 7 for HPDL data multirecords and 8 for SDR multirecords The effective binary image for HPDL and SDR data is a concatenation of the contents of all multirecords of the corresponding type that are presentin a given FRU Information Since in many cases the capacity of EEPROM devices used to store the FRU Information is small HPDL data and SDRs can be placed into the FRU Information in a compressed form The GZIP compression algorithm is supported Compressed data is detected automatically when retrieved by the Shelf Manager from the FRU Information and the data is automatically uncompressed in that case To support this deployment approach several new features have been added to the Pigeon Point FRU Compiler and to the Pigeon Point Shelf Manager CLI Command Line Interface these features enable the user to incorporate compiled HP DL data and SDRs into the FRU Information 3 6 4 1 Specifying the Location of the Carrier FRU Information The preferred location for the Carrier FRU Information is an EEPROM on the ShMM carrier In addition however the Shelf Manager supports getting the Carrier FRU Information from a file on the ShMM file system this can be useful if the ShMM carrier lacks an appropriate EE
86. the swivel chairs of an operations center The Pigeon Point Shelf Manager provides a rich set of System Manager Interface options which provide different mechanisms of access to similar kinds of information and control regarding a shelf Figure 2 shows the System Manager Interface options supported by the Pigeon Point Shelf Manager One such mechanism is the IPMI LAN Interface To maximize interoperability among independently implemented shelf products this interface is required by the AdvancedTCA Specification and supports IPMI messaging with the Shelf Manager via the IPMI Remote Management Control Protocol RMCP A System Manager that uses RMCP to communicate with shelves should be able to interact with any ATCA compliant Shelf Manager This relatively low level interface provides essentially complete access to the IPMI aspects of a shelf including the ability for the System Manager to issue IPMI commands to IPM Controllers in the shelf using the Shelf Manager as a proxy The Pigeon Point Shelf Manager also supports Simple Network Management Protocol SNMP access to the shelf This popular management protocol is supported with a custom Management Information Base MIB providing Get and Set access to a wide range of information and controls regarding the shelf In addition the Pigeon Point Shelf Manager provides two interfaces oriented towards human users rather than programmatic ones e Command Line Interface CLI This interface
87. the RAM file system If there is a d specifier supplied in the call to rupgrade_tool s the intermediate copy to the tmp directory is skipped and the image is copied directly to its destination in the Flash Use of this specifier is intended for a scenario where there is insufficient run time memory on the ShMM for an intermediate copy to the RAM file system If no d specifier is supplied the reliable upgrade procedure invokes a special script the main purpose of which is to validate images after they are copied to the tmp directory If d specifier is present no such validation is performed Currently the script etc upgrade step4vshm Supplied with the Shelf Manager does not perform specific image validation steps but does take responsibility for filling in the Flash partitions for which no images are provided in the current call to rupgrade_tool as would happen in a partial upgrade scenario These partitions are copied from the current persistent Flash to the provisional Flash For example if the current partial upgrade provides only a new RFS image the script copies the U Boot and kernel partitions from the persistent Flash to the provisional Flash As soon as the first image has been installed to its destination the utility proceeds to the second image if there is one and so on until all the supplied image files have been successfully installed to Flash A failure to successfully install an image immediately terminates the
88. the watchdog timeout is that the length of time to switch over to the backup ShMM will be also be increased in the event of a watchdog expiration Release 3 2 0 70 July 31 2013 Pigeon Point Shelf Manager User Guide By default the configuration file variables are used automatically when the ShMM is brought up for the first time The default configuration file imports several environment variables set by U Boot Table 8 U Boot Environment Variables and Descriptions U BooT ENVIRONMENT VARIABLE DESCRIPTION SIPADDR Default RMCP IP Address SIPDEVICE Default RMCP network adapter SIP1ADDR Default Redundant IP Address SIP1DEVICE Default Redundant network adapter SGATEWAY Default gateway used for RMCP communication The environment variables SCARRIER and SCARRIER_OPTIONS are set by the secondary RC script The name of this carrier specific startup script is defined by the U Boot environment variable re2 The Shelf Manager can be reset to factory default parameter values if needed 3 3 1 Carrier specific Configuration File After reading the common configuration file etc shelfman conf the Shelf Manager reads the carrier specific configuration file etc shelfman conf lt carrier name gt where lt carrier name gt is the name of the ShMM carrier used in the relevant shelf in lower case characters Settings in the carrier specific configuration file override settings for the same variable in the
89. the ShMM in this configuration The exact layout of Flash partitions provided by the FOSL depends on the type of the ShMM and the size of the Flash device installed The following tables provide a summary of the Flash partitions maintained by Linux for each of the ShMM 500 variants Table 19 Flash Partitioning for 16MB Flash ShMM 500s OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 0 0 5 dev mtdcharl0 var upgrade The second half of dev mtdblock10 the var upgrade J FFS2 file system 0 5 1 5 dev mtdchar5 Not mounted The other var dev mtdblock5 J FFS2 file system 2 1 dev mtdchar6 Not mounted The other etc dev mtdblock6 J FFS2 file system 3 1 dev mtdchar7 Not mounted The other Linux kernel image Release 3 2 0 197 July 31 2013 Pigeon Point Shelf Manager User Guide MounTED As CONTENT on STARTUP OFFSET IN FLASH SIZE IN Device NODE MBYTES IN MBYTES 0 25 dev mtdchar8 Not mounted The other U Boot firmware image 4 25 3 75 dev mtdchar9 Not mounted The other Linux root file system rfs image 8 0 5 dev mtdcharl0 var upgrade The first half of the dev mtdblock10 var upgrade J FFS2 file system 8 5 1 5 dev mtdchar0 Nar The var J FFS2 file dev mtdblock0 system 10 1 Idev mtdchar1 etc The etc FFS2 file dev mtdblock1 system 11 1 dev mtdchar2 No
90. the Shelf FRU Information Since this sensor is intended to be used for internal communication between the Shelf Manager and Pigeon Point HPI implementation its behavior set of states and event format are subject to change and itis not recommended for an ordinary user to make any direct use of it Release 3 2 0 165 July 31 2013 Pigeon Point Shelf Manager User Guide 4 12 ShMM Tests Available via the Diagnostic Infrastructure A Diagnostic Initiators DI facility has been implemented in the Shelf Manager starting with release 2 8 0 Currently PPS OEM commands are used for access to the diagnostic initiators they are documented in the Pigeon Point Shelf Manager External Interface Reference The overall infrastructure is documented in the IPM Controller Diagnostics Initiator Architecture document The Diagnostic Initiators facility is implemented on the physical Shelf Manager and provides access to several run time tests for the ShMM All tests are associated with Diagnostic Initiator 0 Cancellation is not supported for any of the tests Parameters for the tests are described in the following table The results string for most of these tests has the form Status XX where XX is a two digit hexadecimal test specific reason code 00 if the test completes successfully For the C test the results string in the case of success is a list of 2 character hexadecimal addresses for l C devices present on the bus For the IPMB test the r
91. the backup Shelf Manager to request a Switchover using the switchover command The formerly active Shelf Manager after the switchover can cease to exist or reinitialize itself as the backup Shelf Manager To reinitialize as the backup Shelf Manager the formerly active ShMM must reboot the operating system 4 4 1 Initialization of the Redundant Shelf Managers When starting the Shelf Manager tries to determine its role active or backup and establish a TCP connection with its peer Shelf Manager It does this using the following algorithm e The Shelf Manager starts listening for incoming TCP connection requests from its peer Shelf Manager ef either of the Remote Healthy and Remote Present bits is 0 the Shelf Manager assumes that there is no peer Shelf Manager and assumes the active role e Otherwise the Shelf Manager issues a connection request trying to establish a TCP connection to its peer If the connection request fails the Shelf Manager assumes the active role e lf the connection request succeeds and there is no pending incoming connection request the Shelf Manager assumes that an active Shelf Manager already exists in the shelf and assumes the backup role e If both 1 the outgoing connection request succeeds and 2 an incoming connection request is pending the situation is symmetric between the two Shelf Managers and one of the Shelf Managers should assume the active role and the other should assume the backup role The decis
92. the odd hardware address the assignment of IP addresses looks like this e usb0 192 168 1 131 toggling the least significant bit of the IP address and the least non zero bit of the netmask e usb1 192 168 1 3 toggling the least significant bit of the IP address Here is another example of deriving IP addresses for the USB network interfaces under the assumption that the following definitions are in etc shelfman conf REDUNDANCY_IP_ADDRESS 192 168 1 13 REDUNDANCY_NETMASK 255 255 255 128 Suppose also that the file etc readhwaddr is present The least significant non zero bit in the network mask is 7th bit To toggle this bitin an IP address it is sufficient to add 128 if this bit is set to zero in the IP address or subtract 128 if this bit is set to 1 in the IP address To toggle the least significant bit in an IP address itis sufficient to add 1 if the IP address is even or Subtract 1 if the IP address is odd On the ShMM with the even hardware address the assignment of IP addresses looks like this Release 3 2 0 81 July 31 2013 Pigeon Point Shelf Manager User Guide e usb0 192 168 1 12 since the file etc readhwaddr is present the least significant bit Should be set to zero e usb1 192 168 1 140 toggling the least significant non zero bit of the netmask On the ShMM with the odd hardware address the assignment of IP addresses looks like this e ush0 192 168 1 141 toggling the least significant bit of th
93. the other thread and vice versa As a result the Shelf Manager becomes unresponsive on its external interfaces though it still continues to run Any deadlock is a result of a Serious programming bug and should not happen during normal operation The deadlock detection facilities however add an additional layer of protection and ensure that Should a deadlock happen during normal operation of the Shelf Manager a switchover to the backup Shelf Manager takes place and the Shelf Manager continues operation without interruption There are two deadlock detection facilities one that is triggered by watchdog timer expiration and one that directly detects incorrect locking patterns in the Shelf Manager The watchdog based deadlock detection facility monitors the operation of the most important interface oriented subsystems of the Shelf Manager which now include the CLI processing thread and the RMCP processing threads Normally if a deadlock takes place anywhere in the Shelf Manager one or more of these threads is affected immediately or very soon as a result of deadlock propagation The threads monitored by the watchdog based deadlock detection facility must periodically strobe the internal watchdog timer The strobe period is set separately for each type of monitored thread and is 30 seconds for the CLI thread and 3 seconds for the RMCP threads If one of the monitored Release 3 2 0 152 July 31 2013 Pigeon Point Shelf Manager User Gui
94. to dynamically update any FRU Information in a shelf with HPDL data and SDRs The corresponding binary files should be first downloaded on the ShMM The corresponding command line has the following syntax clia frudataw d lt ipmb address gt lt fru id gt lt in file gt clia frudataw s lt ipmb address gt lt fru id gt lt in file gt clia frudataw d c lt ipmb address gt lt fru id gt clia frudataw s c lt ipmb address gt lt fru id gt The parameters lt ipmb address gt and lt fru id gt specify the IPMB address and FRU device ID of the FRU Information to update When updating HPDL data or SDRs in the Shelf FRU Information use IPMB address 20h and FRU device ID 254 or the addresses of actual locations of the Shelf FRU Information The parameter lt in file gt indicates the binary HPDL data or SDR file which is possibly compressed to be stored in the target FRU Information The option d indicates that HP DL data is to be updated the option s indicates that SDRs data is to be updated The option e removes clears the HPDL data or SDRs respectively from the target FRU Information For example the following commands can be used to update the Shelf FRU Information with new versions of HPDL data and SDRs ona shelf where the Shelf FRU Information is stored in two redundant EEPROMs on the backplane accessed locally by the Shelf Manager clia frudataw d 20 1 stdl14slot hpdl bin gz clia frudataw s 20 2 std
95. transparently moved to the other adapter Release 3 2 0 76 July 31 2013 Pigeon Point Shelf Manager User Guide 3 4 2 2 Active Active Usage of the Two Network Interfaces The approach to network interface redundancy outlined in the previous section seems to be insufficient for some configurations where the network connection between the Shelf Manager and the System Manager goes through several switches and may break on an Ethernet segment that is not adjacent to the ShMM This type of failure cannot be immediately recognized by the Shelf Manager Checking the accessibility of the System Manager from the Shelf Manager does not seem practical in this case since the architecture of the System Manager and its usage of IP addresses is not defined in the ATCA specification and the Shelf Manager design should not artificially limit it The solution in this case should be implemented at the System Manager level Some new features introduced in ECN 002 to the ATCA specification PICMG 3 0 R2 0 facilitate a solution The command Get Shelf Manager IP Addresses allows the System Manager to retrieve the IP addresses exposed by the Shelf Managers both active and backup Using this information the System Manager can check the availability of the Shelf Managers and the current distribution of responsibilities between them active vs backup In active active mode instead of having a single RMCP network address that is switched between the two netwo
96. upgrade procedure vs Skipping a failing image and proceeding to the next one This approach enables the user to separately upgrade the three parts of ShMM firmware U Boot kernel and RFS image However the user should bear in mind that the parts that are not explicitly updated is copied from persistent Flash PPS recommends use of one of the following upgrade approaches e Explicitly upgrading all three partitions e When fewer than 3 partitions are explicitly upgraded omitting the d specifier in that case the special script mentioned above automatically ensures that every upgrade is effectively a full upgrade covering all three partitions Release 3 2 0 205 July 31 2013 Pigeon Point Shelf Manager User Guide lt protocol gt specifies a file copy protocol used to pull each of the specified lt sre gt files to the ShMM and can be any of the following e no No copy is performed This protocol assumes that all of the specified lt s re gt files were pushed to the tmp directory prior to start of the reliable upgrade procedure This protocol choice cannot be used in conjunction with the d option e cp lt dir gt Simple copy This protocol assumes that all of the specified lt sre gt files are to be copied from the specified directory in the ShMM local file system by the ep command This protocol can be useful for instance for installation of upgrade images from an NFS mounted file system or even from a J FFS2 file system
97. via FPGA Field Yes via FPGA Field hardware redundancy Complex Programmable Logic Programmable Logic and hot swap interface via on board PLD as Programmable Logic Device Array Array indicated High speed Multiple ports SPI SPI interface s to on Supporting either SPI carrier devices or SMBus with the latter used to implement IPMB 0 if So configured J TAG interface for Yes Yes Yes processor debug and Flash programming Physical dimensions 67 60mm X 50 80mm_ 92mm x 50 80mm 67 60mm x 41 75mm Form factor definition SO DIMM 144 w Proprietary DDR3 SO DIMM 204 proprietary pin compatible assignments w proprietary pin assignments The following table provides a high level description of the full range of ShMM models including some that are no longer shipping for those in the latter category the table shows the end of life EOL date Table 4 ShMM Models PART PROCESSOR FLASH RAM COMMENTS SPEED SIZE SIZE ShMM 300R EOL June 2007 47Mhz 16Mb 32Mb Replaced with ShMM 500R ShMM 500RE 333M16F32R 333Mhz 16Mb 32Mb Entry level variant of ShMM 500R no support for ShMM redundancy along with other differences ShMM 500R 333M16F 64R EOL 333Mhz 16Mb 64Mb_ Replaced with ShMM August 2008 500R 333M32F64R ShMM 500R 333M32F 64R 333Mhz 32Mb 64Mb ShMM 500R 333M64F 128R 333Mhz 64Mb 128Mb ShMM 1500R 250M64F 128R 250Mhz 64Mb 128Mb
98. within a shelf Categories of such traffic include IPMB messages going between the ShMM and other IPM controllers in the shelf as well as IPMI traffic between the ShMM and external management applications via Ethernet The PICMG HP M 2 specification enables other traffic to be collected and analyzed as well including IP MB L traffic among a Carrier IPMC and the Module Management Controllers MMCs it represents as well as IPMI messaging between an IPMC and its payload via the Payload Interface which is called the System Interface in IP MI These tools consist of the following two parts e The IPMB trace collection daemon This application runs on the ShMM and interfaces with the IPMB specific extensions in the Monterey Linux Pigeon Point Linux kernel on the ShMM to collect IPMB trace data from IPMB 0 This daemon is also able to collect trace data from HPM 2 compliant management controllers in the shelf forwarded via RMCP connections between the daemon and those management controllers Even though this trace data is not necessarily only from IPMB the name IP MB trace daemon is still used The collected data is either stored into a file in a local or NFS mounted file system available on all ShMM variants or transferred directly to the GUI client on a separate platform available on ShMM 500 1500 only e The IPMI trace analyzer This role is typically played by a GUI application that supports viewing and analyzing an IP MB trace collec
99. 0 03 Od 2e 23 71 00 01 02 le 13 04 08 00 45 10 TETE ETETETT E 0010 00 54 00 00 40 00 3f 11 c2 79 cO a8 0l 3e 50 f0 M LES E oP 0020 66 39 02 6f 8c dO 00 40 5f 84 06 00 ff 07 06 40 LA SEENE 0030 Ob 00 00 00 02 00 00 00 17 00 89 lc 5b 20 Oc Ol sssaaa saaa EF 0040 00 00 80 02 60 02 bf Oa 40 00 00 00 02 60 01 00 eee EEPE b eth0 lt live capture in progress gt File Packets 427022 Displayed 19 Marked 0 Profile Default A 6 To stop the capture session go to the Capture submenu and select Stop In a command line context instead of steps 1 3 run the tshark utility specifying the target in the form of the network interface name e g ethO using option i Specify the filtering option R rmcp to include only RMCP traffic in the trace Options x and vV as in the case of unattended IPMB trace mode can be used to request a hexadecimal and or detailed dumps of the messages Any additional options needed for filtering and analysis of the message stream see section 5 4 6 should also be specified at this time Use the amp amp logical and operator to combine additional filtering options with the filtering option rmep Next execute the actions from step 4 the trace is then captured by tshark and sent to the command output For example tshark i ethO R rmcp 71 989290 80 240 102 57 gt 192 168 1 199 IPMI ATCA Req Get Channel Authentication Capabilities seq 0x0 71 990809 192 168 1 199 gt 80 240 102 57 IPM
100. 0 MHz A2F Fast delay calibration 1330 cycles per 100uS A2F eNVM 128 KB 00000000 00020000 A2F eSRAM 16 KB 20000000 20004000 A2F Extram start 200022F0 RUPG booting from image 1 confirmed In serial Out serial Err serial Net FECO 00 18 49 01 8f 26 FECI 00 18 49 01 8f 27 FECO FECL Hit any key to stop autoboot 0 shmm700 Note shmm500 shmm1500 and shmm700 are the U Boot prompts on each ShMM variant for user commands For a complete set of supported commands type help In this document example dialogues that are applicable to all three ShMM variants represent this prompt with shmmxx00 Example dialogues that are applicable to only one of the variants use the relevant prompt 3 2 1 U Boot Environment Variables U Boot includes a set of environment variables that should be configured prior to U Boot use The following table describes the default set of variables available Some of these variables are set automatically by U Boot and should not be modified by a user Table 5 U Boot Environment Variables ENVIRONMENT DESCRIPTION VARIABLE a2f_upg_file ShMM 700 only This variable contains the name of the A2F 060 firmware image file that U Boot looks for on the boot volume This variable is set automatically by U Boot during bootstrap and must not be changed by a user its value is used by U Boot to pass information about the A2F upgrade image file to the Linux kernel Only th
101. 00 IntegralHP requires a variant of the ShMM thathas additional memory specifically 128 megabytes of RAM and 64 megabytes of Flash These memory resources needed for IntegralHP are present on all ShMM 700s So no special ShMM configuration requirements exist for IntegralHP I oO SS oO IntegralHP is implemented as a shared library within the Shelf Manager and is turned off by default To turn it on itis necessary to set the configuration variable ENABLE_INTEGRALHPI to TRUE and ensure that the corresponding shared library Libintegralhpi so and several other libraries that IntegralHP depends on is present on the ShMM A special RFS image containing those libraries is installed on your shelf if it is enabled for IntegralHP support 3 14 1 HPI Domain Support in IntegralHP IntegralHP exposes the ATCA shelf managed by the potentially redundant Shelf Manager as a single HPI domain This domain has Domain ID 1 by default In order to allow a single HP client to Release 3 2 0 123 J uly 31 2013 Pigeon Point Shelf Manager User Guide manage multiple ATCA shelves the domain ID for a given shelf can be configured via the Shelf Manager configuration variable INTEGRALHPI_DOMAIN_ ID 3 14 2 HPI SNMP Subagent Support The IntegralHP l enabled RFS includes a public domain HPI SNMP subagent that allows a remote client to communicate with IntegralHP I using the SNMP protocol The mapping between the HPI interface and the SNMP
102. 0000 Verifying Checksum OK Release 3 2 0 247 July 31 2013 Pigeon Point Shelf Manager User Guide Loading Kernel Image OK OK Starting kernel init started BusyBox v1l 16 2 2011 11 03 18 19 59 MSK etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re etc re ounting filesystems ounted proc ounted sys ounted dev pts Boot using 0 image set Fetching evnironment settings Extracted following params re2 etc rce shmm700 hpdl ipaddr 192 168 0 22 ipdevice eth0 ipladdr 192 168 1 2 ipldevice usb0 ip2addr ip2device carrier Checking the reliable upgrade watchdog timer activated Attaching UBIO user UBI device number 0 total 512 LEBs 33488896 bytes 31 9 MiB availabl e 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB etc re etc re etc re UBI devi availabl Attach result 0 Mount ubi0 user Attaching UBI1 boot ce number 1 total 484 LEBs 31657472 bytes 30 2 MiB e 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB etc re etc re etc re etc re etc re 3297941 Attach result 0 ount ubil boot Selecting carrier from the environment Extracting shelfman rootfs patch from 0 sentry shmm700 app records in 4122 1 records out 2110464 etc re etc re etc re etc re etc re etc re etc re b
103. 0s implements isolation of faulty bus segments in the case of a persistent error on IPMB 0 This prevents such an error on one or several specific IPM controllers from causing a failure of an entire bus Isolation is performed independently for IPMB A and IPMB B Isolated segments are turned off which means that the corresponding IPM controller s is are isolated from the corresponding bus Such segments are kept in the isolated state for TPMB_LINK_ISOLATION_TIMEOUT seconds after which they are automatically turned on if the fault still exists the persistent IPMB 0 error occurs again and the faulty links is isolated again By default however the value of this configuration parameter is 1 which means never re enable the link automatically In addition isolated links can be turned on in one of the following ways e manually via the CLI command setipmbstate with parameters indicating the ShMC target address 0x20 and the corresponding link number and corresponding IPM controller is removed from the shelf that is when it goes to the state MO e automatically when the corresponding IPM controller is removed from the shelf that is when it goes to the state M0 To find out the current isolation state of radial IP MB 0 links the CLI command getipmbstate can be used When applied to the ShMC target address 0x20 this command shows the state of all radial IP MB 0 links For example the command clia getipmbstate 20
104. 0x08 0x08 0 0x08 0x08 0 0x08 0x08 0 0x08 0x08 Sensor 24 IPMB A iocal iocal Sensor 25 IPMB A 1ocal iocal Sensor 26 IPMB A iocal iocal Sensor 27 IPMB A iocal iocal IP Enabled Control Control T d IP Enabled Control Control T L IP Enabled Control Control T i IP Enabled Control T I N N I N N d B LINK 12 PMB B o fail o fail Enabled ure ure B LINK 13 PMB B o fail o fail Enabled ure ure B LINK 14 PMB B No fail IPMB B No fail No fail No fail Enabled ure ure B LINK 15 Enabled ure ure Control The output above indicates that IPMB 0 link 4 for IPMB A has been isolated as a result of a persistent IP MB failure To manually re enable the link the following CLI command can be used clia setipmbstate 20 A 41 4 5 2 Operation in ShMM 1500 based Shelves with Radial IPMB 0 On ShMM 1500 based shelves with radial IP MB 0 support there is a separate logical IP MB interface for each of the two portions IPMB A and IPMB B of each radial IPMB segment There is no need for faulty link isolation in this case since each segment is separately accessible and permanently
105. 1 EnterpriseS pecific 02 03 12 6F 00 Specific Status len 3 SenType 12h EventT ype 6Fh Event ffset 00h Direction Asserted 43 04 61 A3 ED 13 Time in ticks 0 01sec obviously wrong 30 4D ASN 1 tag at variable bindings start and remaining length 30 4B First variable header and remaining length 06 OA 2B 06 01 04 01 98 6F 01 01 01 Variable OID 1 3 6 1 4 1 3183 1 1 1 04 3p Variable data length data follows see Table 15 7 PET Variable Bindings Field of IPMlv1 5 specification 1D 3A 30 76 62 F2 11 DD 00 80 00 50 c2 3F FB 56 GUID 00 07 trap sequence 13 ED A3 61 00 00 Local Timestamp Tue Aug 5 16 29 21 2008 20 20 trap source event source IPMI 01 Event Severity Monitor 20 Sensor Device BMC 85 Sensor Number oo Entity 00h unspecified oo Entity Instance O0h unspecified_ co 00 00 00 00 00 00 00 Event Data 19 Language Code English 00 00 40 oa Manufacturer ID MSB first 00400Ah for PPS o0 oo System ID MSB first 80 4B 01 PET multirecord ASCII len Bh Text Alert String 54 65 73 74 53 74 72 69 6E 67 00 Null terminated Alert String TestS tring c1 No more fields 3 14 Configuring the IntegralHP Interface The Pigeon Point Shelf Manager optionally includes IntegralHP I an implementation of the Service Availability Forum SAF www saforum org Hardware Platform Interface HP 1 operating as a Subsystem within the Shelf Manager On the ShMM 500 and ShMM 15
106. 1 Header Checksum Holds the zero checksum of the header 5 3 Manufacturer ID LS byte first Write as the three byte ID assigned to PPS 16394 decimal or 00400Ah 8 1 PPS Record ID For this record the value 5h is used 9 1 Record Format Version For this record the value 0h is used 8 1 IPMB Topology Type The following values are defined Oh unspecified 1h Bused IPMB 0 carrier 2h Dual star radial IP MB 0 carrier 3h Redundant dual star radial IP MB 0 carrier 4h Universal redundant dual star radial IPMB 0 carrier not currently Supported Other values are reserved Release 3 2 0 147 July 31 2013 Pigeon Point Shelf Manager User Guide In a FRU Info Compiler input file the following approach is used to define the IPMB Topology record for a redundant dual star radial configuration in this example PPS IPMB Topology Carrier Type 3 In an HPDL carrier definition file the IPMB topology can be specified ina LPMB_ TOPOLOGY clause that has the following syntax lt ipmb_topology_clause gt IPMB_ TOPOLOGY lt bus_type gt lt bus_type gt BUSED DUAL_STAR REDUNDANT_DUAL_STAR For example a redundant dual star radial configuration the following definition can be used IPMB_ TOPOLOGY REDUNDANT_DUAL_STAR The radial functionality is turned on if the following two prerequisites are satisfied e the carrier is designated as a dual star or redundant dual star type
107. 1026 Shelfman Registered Shelfman Registered Shelfman Registered Shelfman Registered Shelfman Registered Shelfman Registered 131 Registered Local Temp Ext 1 temp sensor not implemented Ext 2 temp sensor not implemented sensor 0 3 3 3STBY voltage sensor 3 3MAIN voltage sensor 5V voltage sensor 0 6 Vcpp voltage sensor 12V voltage sensor 12V voltage sensor J uly 31 2013 Pigeon Point Shelf Manager User Guide lt I gt 14 35 22 188 199 FT 0 FRU successfully registered as FRU 02 lt I gt 14 35 22 190 199 Initializing Fans lt I gt 14 35 22 192 199 Registering fan RD facility lt I gt 14 35 22 194 199 Activating fan tray 0 fan level 5 power level 0 lt I gt 14 35 22 197 199 Controller 20 FRU 2 ATCA state set to M1 prev M0 cause 0 locked 0 lt I gt 14 35 22 203 199 Controller 20 FRU 2 updating blue LED for M1 lt I gt 14 35 22 206 199 Carrier version set to 82 64 01 00 lt I gt 14 35 22 243 199 Alarm sensor registered lt I gt 14 35 22 245 199 Alarm Input Sensor reading 0 initial update 0 lt I gt 14 35 22 247 199 Alarm input sensor registered lt I gt 14 35 22 253 199 SEL initialized SEL compression resources max_sel_entries 1024 lt I gt 14 35 22 256 199 SE compression is ON journal
108. 18 cB 89 04 38 8e 04 a a 8 0040 a9 eth0 lt live capture in progress gt File Packets 280 Displayed 20 Marked 0 Profile Default ib Instead of 12C related information the Source and Destination columns show the source and destination IP addresses of the captured message Instead of the three protocol layers in the case of an IPMB trace seven protocol layers are shown e Layer 1 provides low level information about a message such as the frame size arrival time relative to the start of the trace and the list of protocols used in the frame e Layer 2 provides Ethernet specific information about the frame source and destination MAC addresses and the protocol type Layer 3 provides IP specific information from the frame Layer 4 provides UDP specific information from the frame Layer 5 provides RMCP header information from the frame version sequence number and Class e Layer 6 includes additional information about the RMCP or RMCP specific fields in the message Release 3 2 0 175 J uly 31 2013 Pigeon Point Shelf Manager User Guide e Layer 7 is available only for IPMI ATCA messages and provides message header information and information about the request response data Specifically the message is identified as either a request or a reply and the request reply header fields are interpreted The data fields are interpreted as defined by IPMI and PICMG specs for a particular request or response 5 3 2 Introduc
109. 200 bootcmd run ramargs addmisc bootm kernel_start rfs_start bootdelay 3 boot file sentry shmm1500 kernel baudrate 115200 console ttys0 ethaddr 00 50 c2 3f xx yy ethladdr 00 50 c2 3f xx 2z netmask 255 255 0 0 hostname shmm1500 gatewayip 192 168 0 1 ipdevice eth0 ipladdr 192 168 1 2 ipldevice ethl rc2 etc re carrier3 ipaddr 192 168 0 22 start_rc2_daemons y flash_reset n password_reset n logging ram net tftpboot 400000 S bootfile tftpboot 1200000 ramdisk run ramargs addmisc bootm 400000 1200000 nfs tftpboot 400000 S bootfile run nfsargs addip addmisc bootm nfsargs setenv bootargs root dev nfs rw nfsroot serverip rootpath quiet quiet rc_ifconfig n ramargs setenv bootargs root dev ram rw ramdisk sentry shmm1500 rfs rootpath rootfs rmcpaddr 192 168 0 2 serverip 192 168 0 7 timezone UTC For ShMM 700 bootargs console console bootcmd run mount_ubifs a2f upgrade run ubifs bootdelay 3 baudrate 115200 netmask 255 255 255 0 bootfile sSentry shmm700 kernel loadaddr 0x42000000 console ttySP0 115200n8 rfsaddr 0x46000000 Release 3 2 0 32 July 31 2013 Pigeon Point Shelf Manager User Guide mt dids spi0 SPI 0 mtdparts1 512K uboot0 512K uboot1 256K envvars0 256K envvars1 256K u bifs_cache 30976K boot 32M user boot file sentry shmm700 kernel ramdisk sentry shmm700 rfs quiet quiet mac_override set bootargs S bootargs fec_mac eth
110. 22 is used A failure in copying an image to the ShMM causes the utility to terminate the upgrade procedure vs Skipping a failing image and proceeding to the next one For each provisional Flash partition upgraded by the s option the to be upgraded partition is given write permissions after the validity of the image has been checked and right before the src image is about to be moved to the Flash Write permissions are removed from the partition immediately after the full image has been moved to Flash Combined with the fact that all the partitions containing the U Boot Linux kernel and root file system images are read only on boot up of the ShMM this ensures that applications cannot accidentally erase the critical boot up images If the specifier R specifier has been used in the command line no copy operation is performed and itis assumed that corresponding upgrade images have been already copied into Flash None of the previously described steps is executed in this case execution of the reliable upgrade procedure begins from this point After all the specified images have been installed to their respective destinations in Flash the utility invokes a hook script that enables custom actions required by an application at the point where Release 3 2 0 206 July 31 2013 Pigeon Point Shelf Manager User Guide the upgrade images have been already installed in Flash but the upgrade procedure has not yet initiated the hardware mechanis
111. 255 0 gatewayip 192 168 0 1 shmmxx00 4 Change settings and commit to non volatile storage shmmxx00 setenv rmcpaddr 10 1 1 10 shmmxx00 setenv netmask 255 255 0 0 shmmxx00 setenv gatewayip 10 1 1 1 shmmxx00 saveenv Saving Environment to EEPROM shmmxx00 Note saveenv output is slightly different on the ShMM 700 5 Bootthe ShMM up to full operational state and log in as user root On ShMM 500 shmm500 reset U Boot 1 1 2 Apr 27 2005 19 17 09 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 00 00 00 00 00 00 00 00 00 03 03 03 DRAM 64 MB Flash 16 MB Tne serial Outs serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 Booting image at bfb00000 Image Name IPS Linux 2 4 26 Created 2005 05 07 17 35 21 UTC Image Type IPS Linux Kernel Image gzip compressed Data Size 843144 Bytes 823 4 kB Load Address 80100000 Entry Point 802bc040 Verifying Checksum OK Uncompressing Kernel Image OK Loading Ramdisk Image at bfc40000 Image Name sentry RFS Ramdisk Image shmm500 login root Release 3 2 0 85 July 31 2013 Pigeon Point Shelf Manager User Guide BusyBox v0 60 5 2005 05 07 17 27 0000 Built in shell msh On ShMM 1500 shmm1500 reset Resetting the board U Boot 1 1 4 Jun 15 2006 17 49 12 MPC83XX Clock configuration Coherent System Bus 99 MHz Core 249 MHz Local Bus 24 MHz C
112. 2C bus By resetting the 12C multiplexer or switch it is possible to implement an algorithm in the Shelf Manager to isolate faulty devices on subordinate 12C buses that are behind that multiplexer switch so that access to devices on other 12C buses can still occur successfully In the original implementation this feature was applicable only to one multiplexer or switch connected to the Shelf Manager GPIO E8 GPIO_08 on the ShMM 700R was used to reset it On ShMM 70OR carriers that have multiple master only 12C buses the number of the bus to which the multiplexer or switch was connected could be configured By default however logical 12C bus 5 physical bus 3 on the ShMM 700R was used for this functionality Starting with release 3 2 0 this feature has been re implemented to allow more than one 12C multiplexer or switch to be resettable These multiplexers or switches can also be located in the case of the ShMM 700 on multiple 12C buses The new implementation is not backwards compatible and applies only to HPDL based shelves The information about reset support and the number of the ShMM GPIO used to reset the device must be contained in the HPDL device description of the corresponding multiplexer or switch The following subsections describe the usage of this feature in the original and HPDL based mode respectively Release 3 2 0 156 July 31 2013 Pigeon Point Shelf Manager User Guide 4 9 1 Master only I2C Bus Fault Isolatio
113. 31 2013 Pigeon Point Shelf Manager User Guide unless the command line parameter skip checksums is specified This procedure is repeated for all images specified in the command line After all images have been successfully obtained and validated they are finally one by one dispatched to their destination locations that is U Boot is copied to the destination Flash partition and other images are copied to appropriate locations on the UBIFS Boot and User volumes If the reliable upgrade utility does not succeed in copying or validating at least one image it immediately terminates the overall upgrade Subsequent images if any are ignored they are not even copied to the ShMM Before replacing kernel and RFS images in the Boot volume the reliable upgrade utility cleans up the UBIFS cache partition to tell U Boot that it is not valid any more The UBIFS cache is re created by U Boot automatically When copying the kernel and RFS images to the Boot UBIFS Flash partition the reliable upgrade utility mounts the partition before copying the image and then unmounts it shortly after the completion of the copy operation Normally the Boot Flash partition stays unmounted This ensures that applications cannot accidentally erase the critical boot up images and allows U Boot to cache the location of boot images for faster loading After all the specified images have been installed at their respective destinations the utility mounts the candidate
114. 4 Trying 192 168 1 174 Connected to 192 168 1 174 Escape character is BusyBox on shmm 174 login root BusyBox v0 60 5 2005 05 07 17 27 0000 Built in shell msh Release 3 2 0 214 July 31 2013 Pige on Point Shelf Manager User Guide The parameters to rupgrade_tool s indicate that only the RFS is being upgraded and that the copy protocol is FTP accessing a specified IP address and file with user admin and no password supplied xupgrade_tool s r proto ftp 192 168 1 253 sentry shmm1500 rfs tf tpboot ru ppc admin hook etc_copy v age saved nt flash is 0 nal flash is 1 rupgrade_tool PLB is 9 rupgrade_tool EEPROM p rupgrade_tool persiste rupgrade_tool provisio rupgrade_tool copying rupgrade_tool copying 192 168 1 253 tftpboot 1 1 3 USER admin image s sentry shmm1500 rfs from ru ppe to tmp using ftp protocol 220 vsFTPd The user is asked here for a password to the FTP site that password is entered manually 331 Please specify the password PASS kkkkxk 230 Login successful TYPE I 200 Switching to Binary PASV 227 Entering Passive Mo ppc sentry shmm1500 rfs 150 Opening BINARY mode ppc sentry shmm1500 rfs 226 File send OK QUIT 221 Goodbye H ave fun mode de RI 192 168 1 253 112 206 ETR tftpboot ru data connection for tftpboot ru 4069968 bytes In the next step a special script step4vshm is invok
115. 44a0 pid 263 file msg c line 867 Jun 22 17 23 27 2007 Succeeded 2ac344a0 pid 263 file msg c line 867 Jun 22 17 23 27 2007 Released 2ac344a0 pid 263 file msg c line 876 Jun 22 17 23 27 2007 Acquiring 2acbe5ec pid 263 file ipmf_ipmb c line 1649 Jun 22 17 23 27 2007 Succeeded 2Zacbe5ec pid 263 file ipmf_ipmb c line 1649 Jun 22 17 23 27 2007 Released 2acbe5ec pid 263 file ipmf_ipmb c line 1656 Jun 22 17 23 27 2007 Acquiring 2Zacclde0 pid 263 file ipmf_ipmb_buffe line 98 Jun 22 17 23 27 2007 Succeeded 2acclde0 pid 263 file ipmf_ipmb_buffe line 98 Jun 22 17 23 27 2007 Released 2acclde0 pid 263 file ipmf_ipmb_buffe line 129 Jun 22 17 23 27 2007 Released 2acclf80 pid 263 file ipmf c line 1533 Jun 22 17 23 27 2007 CondWait Before 7d5ff 740 pid 263 file msg c line 790 Jun 22 17 23 27 2007 CondWait After 7d5ff740 pid 263 file msg c line 790 Jun 22 17 23 18 2007 Succeeded 2ac34970 pid 216 file shelffru c line 3171 Jun 22 17 23 18 2007 Released 2ac34970 pid 216 file shelffru c line 3196 Jun 22 17 23 18 2007 Acquiring 2ac34970 pid 216 file shelffru c line 3171 Jun 22 17 23 18 2007 Succeeded 2ac34970 pid 216 file shelffru c line 3171 Jun 22 17 23 18 2007 Released 2ac34970 pid 216 file shelffru c line 3196 Total of 32 locks 2ac34 2ac34 2acc1f80 2ac344a0 2acbe5ec 2acc1de0 2ac34950 2ac36230 2ac34b90 2Zacbe620 2ac34
116. 53 option subnet mask 255 255 255 0 0x84 0 0 host board2_fru0_ch0O option dhcp client identifier 58 58 58 31 0 0 fixed address 192 168 1 203 option routers 192 168 1 253 option subnet mask 255 255 255 0 host board2_fru0_chl option dhcp client identifier 58 58 58 31 0 1 fixed address 192 168 1 203 option routers 192 168 1 253 option subnet mask 255 255 255 0 host board2_fru0_ch2 option dhcp client identifier 58 58 58 31 0 2 fixed address 192 168 1 203 option routers 192 168 1 253 option subnet mask 255 255 255 0 host board2_fru0O_ch3 option dhcp client identifier 58 58 58 31 0 3 fixed address 192 168 1 203 option routers 192 168 1 253 option subnet mask 255 255 255 0 host board2_fru0_ch4 option dhcp client identifier 58 58 58 31 0 4 fixed address 192 168 1 203 option routers 192 168 1 253 option subnet mask 255 255 255 0 4 10 3 Dispatching LAN Configuration Parameters to Boards For each board or module the Shelf Manager detects whether it supports LAN channels and expects the configuration parameters for them to be assigned by the Shelf Manager If this is the Release 3 2 0 163 July 31 2013 Pigeon Point Shelf Manager User Guide case for a particular board or module the Shelf Manager sends it the configuration parameters from the next parameter set for the corresponding physical slot Shelf Manager does this for all boards or modules in
117. 60 I2C 1 0x7a IPMI ATC Req Get Device ID seq 0x02 246 6 765873 I2C 2 0x7a IPMI ATC Req Get Device ID seq 0x02 247 6 766420 I2C 2 0x7b IPMI ATC Req Get Device ID seq 0x03 248 6 766834 I2C 1 0x7b IPMI ATC Req Get Device ID seq 0x03 249 6 767380 I2C 1 Ox7c IPMI ATC Req Get Device ID seq 0x04 250 6 767725 12C 2 Ox7c IPMI ATC Req Get Device ID seq 0x04 251 6 768293 I2C 2 0x7d IPMI ATC Req Get Device ID seq 0x05 l 252 6 768703 I2C 1 0x7d IPMI ATC Req Get Device ID seq 0x05 253 6 769264 I2C 1 0x7e IPMI ATC Req Get Device ID seq 0x06 254 6 769684 I2C 2 0x7e IPMI ATC Req Get Device ID seq 0x06 255 6 770229 I2C 2 0x7f IPMI ATC Req Get Device ID seq 0x07 b Frame 244 7 bytes on wire 7 bytes captured E b Inter Integrated Circuit Data b Intelligent Platform Management Interface 0000 f2 18 f6 20 04 O01 db aa aa TCP 192 168 1 65 lt live capture in pro Packets 298 Displayed 298 Ma gt Profile Default y After a message has been selected its raw contents are shown in the bottom frame of the main window both in hex and ASCII interpretations The middle frame of the main window provides protocol interpretations of the message data By clicking on the right arrow down arrow or GUI elements you can choose a particular protocol layer you wish to analyze Release 3 2 0 188 July 31 2013 Pigeon Point Shelf Manager User Guide Figure 15 IPMI Protocol Layers IPMB trace TCP 192 16
118. 8 1 65 Capturing Wireshark 0X File Edit View Go Capture Analyze Statistics Help a R rf D gt z 5 Qi a ea M 4 828 aaan gumma amp Filter Expression Clear Apply No Time Source Destination Protocol Info s seein unas k kae os bo ED 237 6 761618 I2C 1 0x76 IPMI ATC Req Get Device ID seq 0x3e 238 6 762032 I2C 2 0x76 IPMI ATC Req Get Device ID seq 0x3e 239 6 762581 I2C 2 0x77 IPMI ATC Req Get Device ID seq 0x3f 240 6 762993 I2C 1 0x77 IPMI ATC Req Get Device ID seq 0x3f 241 6 763541 I2C 1 0x78 IPMI ATC Req Get Device ID seq 0x00 1 242 6 763952 12C 2 0x78 IPMI ATC Req Get Device ID seq 0x00 243 6 764502 I2C 2 0x79 IPMI ATC Req Get Device ID seq 0x01 244 6 764913 T2C F 0x79 IPMI ATC Req Get Device ID seq 0x01 745 6 765460 T C 1 0x7a TPMT ATC Rea Get Device TN sea AxA b b Frame 244 7 bytes on wire 7 bytes captured v Inter Integrated Circuit Data Bus I2C 1 Target address 0x79 Flags 0x00000000 v Intelligent Platform Management Interface No corresponding response gt Header Get Device ID Request from 0x20 to Oxf2 Data checksum Oxdb correct 0000 EEEE db ES Text item 6 bytes Packets 382 Displayed 382 Mar Profile Default 4 A trace can be filtered using the filter expressions Specifically for IP MI trace filtering the following expression primitives are available Table 18 IPMI Filter Primitives FILTER PRIMITIVE NAME DESCRIPTION ipmi hea
119. 9 using cp protocol 4 invoking scripts step4h etc_copy 4 image s copy OK 5 watchdog started 6 selected provisional flash 7 reboot 9 WDT not fired upgrade in progress 10 provisional flash 1 disabling watchdog 12 upgrade WDT disabled 13 EEPROM updated 13 invoking scripts step1l3h 14 upgrade completed successfully C re PR Release 3 2 0 216 J uly 31 2013 Pigeon Point Shelf Manager User Guide 7 8 3 Example 3 This example shows an unsuccessful reliable upgrade Power is turned off after the boot from the provisional Flash but before the reliable upgrade is finalized After turning the power back on the rollback to the persistent Flash occurs This reliable upgrade is initiated from the serial console All three images are assumed to be already in tmp rupgrade_tool s k sentry kernel r sentry rfs u u boot bin hook etc_copy v rupgrade_tool PLB is 5 rupgrade_tool EEPROM page saved rupgrade_tool persistent flash is 0 rupgrade_tool provisional flash is 1 rupgrade_tool copying image s rupgrade_tool invoking scripts step4v u u boot bin k sentry kernel r sentry rfs hook etc_copy rupgrade_tool copying u boot bin from tmp to dev mtdchar8 using cp protocol rupgrade_tool copying sentry kernel from tmp to dev mtdchar7 using cp protocol rupgrade_tool copying sentry rfs from tmp to dev mtdchar9 using cp proto
120. 92 168 0 2 192 168 0 2 Enterprise Specific Trap 2453248 Uptime 0 00 41 04 PPS PET MIB ipmi trap text STRING RecordID 0004 RecordType Platform Event Time Dec 19 17 11 43 2007 Generator 20 LUNO Chan0O SensorType 25 SensorNumber Cl EventType 6F EventDirection asserted EventData 00 FF FF PET_FORMAT 2 Release 3 2 0 121 J uly 31 2013 Pigeon Point Shelf Manager User Guide Received 296 bytes from 192 168 0 2 192 168 0 2 Enterprise Specific Trap 2453248 Uptime 489 days dee Bal LTES PPS PET MIB ipmi trap record id INTEGER 4 PPS PET MIB ipmi trap record typ INTEGER platformEvent 2 PPS PET MIB ipmi trap timestamp INTEGER 1198084292 PPS PET MIB ipmi trap generator address INTEGER bmc 32 PPS PET MIB ipmi trap generator lun INTEGER lun0 0 PPS PET MIB ipmi trap generator channel INTEGER ipmb 0 PPS PET MIB ipmi trap sensor typ INTEGER entityPresence 37 PPS PET MIB ipmi trap sensor number INTEGER 193 PPS PET MIB ipmi trap event typ INTEGER sensorSpecific 11l PPS PET MIB ipmi trap event direction INTEGER asserted 0 PPS PET MIB ipmi trap event data Hex STRING 00 FF FF r PPS PET MIB ipmi trap entire record Hex STRING 04 00 02 C4 50 69 47 20 00 04 25 Cl 6F 00 FF FF 3 13 1 Parsed Example of SNMP Trap
121. ANCY_NET_ADAPTER s10 In all other aspects configuration of the serial interface sl0 for redundant communication is the Same as in the case of the second Ethernet adapter see 3 4 2 In particular both the active and backup ShMM specify the same IP address for the redundancy interface but software assigns the next logical IP address to the ShMM with an odd hardware address Compression of data transferred over the Software Redundancy Interface applies also to ShMM 1500 Compression may be even more beneficial for performance on ShMM 1500 since the SLIP interface is somewhat slower than the USB based SRI used on the ShMM 500 However as with the ShMM 500 compression is turned off by default 3 4 5 Changing the Default ShMM Network Parameters Configuring a ShMM to work in a specific network environment requires changing the following network parameters Release 3 2 0 82 July 31 2013 Pigeon Point Shelf Manager User Guide e RMCP IP address rmcpaddress e RMCP gateway address gatewayip e RMCP netmask netmask If the second ShMM Ethernet is used for software redundancy communication between a pair of ShMMs the network parameters above would typically not need to be adapted to a different network environment since this is a dedicated private network between the redundant ShMMs However if the second network interface is used for RMCP communication please use steps 10 and 11 below to change the settings U Boot environmen
122. After removing the ShMM 500 carrier inserta second ShMM 700 carrier into its slot It Should come up as a backup Shelf Manager The switchover procedure is now complete NOTE As implied above a switchover from a ShMM 700 back to a ShMM 500 is NOT supported Once a ShMM 700 becomes active the only supported action is to remove the second ShMM 500 carrier Also aShMM 500 carrier should never be inserted into a shelf with a running ShMM 700 WARNING The above ShMM 500 to ShMM 700 switchover procedure cannot be used with a ShMM 700 module that is mounted on a ShMM 500 carrier via the Pigeon Point Systems ShMM 700R SO DIMM adapter board SHMM 700R SA That adapter board which is not intended for production use is not designed to support testing of this switchover scenario Release 3 2 0 253 July 31 2013 Pigeon Point Shelf Manager User Guide Appendix B Customer Support If you are having problems with the Pigeon Point Shelf Manager or ShMM product please contact your supplier for the Pigeon Point products with questions and problem reports If you have any questions about direct purchase of Pigeon Point products one of the Pigeon Point Board Management Reference designs for instance please contact Pigeon Point Systems Pigeon Point Systems 2191 S El Camino Real Suite 209 Oceanside CA 92054 USA Phone 1 760 757 2304 Fax 1 760 757 2302 Release 3 2 0 254 July 31 2013 Pigeon Point Shelf Manager User Guide Append
123. C Image Type ARM Linux RAMDisk Image gzip compressed Data Size 2282975 Bytes 2 2 MB Load Address 00000000 Entry Point 00000000 Verifying Checksum OK Loading Kernel Image OK OK Starting kernel Release 3 2 0 237 July 31 2013 Pigeon Point Shelf Manager User Guide init started BusyBox v1l 16 2 2011 11 03 18 19 59 MSK etc rce Mounting filesystems etc rc Mounted proc etc rce Mounted sys etc rc Mounted dev pts etc rc Boot using 0 image set etc rc Fetching evnironment settings etc rc Extracted following params etc rce rce2 etc rce shmm700 hpdl etc rce ipaddr 192 168 0 22 etc rce ipdevice etho etc rce ipladdr 192 168 1 2 etc rce ipldevice usb0 etc re ip2addr etc rce ip2device etc re carrier etc rc Checking the reliable upgrade watchdog timer lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW activated Here rupgrade is invoked for the first time after the reboot to strobe the reliable upgrade watchdog timer etc re Attaching UBIO user UBI device number 0 total 512 LEBs 33488896 bytes 31 9 MiB available 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB etc re Attach result 0 etc rce Mount ubi0 user etc re Attaching UBI1 boot UBI device number 1 total 484 LEBs 31657472 bytes 30 2 MiB available 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB etc re Attach result 0 etc rce Mount ub
124. Checksum OK Power is turned off here After some time power is turned back on Assignment of provisional Flash has been lost because of the power loss so the system reverts back to the persistent Flash U Boot 1 1 2 Apr 11 2005 15 16 25 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 8000048 DRAM 64 MB Flash 16 MB prs serial Out serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 Booting image at bfb00000 Image Name IPS Linux 2 4 26 Created 2005 04 11 10 35 08 UTC Image Type IPS Linux Kernel Image gzip compressed Data Size 843129 Bytes 823 4 kB Load Address 80100000 Entry Point 802bc040 Verifying Checksum OK Uncompressing Kernel Image OK Loading Ramdisk Image at bfc40000 Image Name sentry RFS Ramdisk Image Created 2005 04 11 18 27 17 UTC Image Type MIPS Linux RAMDisk Image gzip compressed Data Size 2372311 Bytes 2 3 MB Load Address 00000000 Entry Point 00000000 Verifying Checksum OK Starting kernel init started BusyBox v0 60 5 2005 02 07 16 45 0000 multi call binary hub c new USB device AU1550 1 assigned address 2 usb0O speed config 1 Ethernet Gadget Release 3 2 0 218 July 31 2013 Pigeon Point Shelf Manager User Guide usbl register usbnet usb AU1550 1 Linux Device serial 8000048 not found etc rc Mounted proc etc rce Mounting filesystems et
125. Encryption code present ShMM 1500R 250M64F128R NE 250Mhz 64Mb 128Mb Encryption code removed ShMM 700R AA 297Mhz 64Mb 128Mb Release 3 2 0 20 J uly 31 2013 Pigeon Point Shelf Manager User Guide This edition of the User Guide focuses on the SaMM 500R ShMM 1500R and ShMM 700R and uses the shorthand ShMM to refer to all three variants A separate edition of this document covers the Entry Level Shelf Manager that is implemented on the ShMM 500RE The ShMM 1500R can be ordered with encryption code present or encryption code removed Each variant has a different United States Export Control Classification Number ECCN The ShMM 500R the ShMM 1500R with encryption code present and the ShMM 700R have an ECCN of 5A002 The ShMM 1500R with the encryption code removed has an ECCN of 54992 For more information regarding these export classification topics please contact Pigeon Point Systems Release 3 2 0 21 July 31 2013 Pigeon Point Shelf Manager User Guide 3 Configuration The Shelf Manager application runs on a specialized distribution of Linux For the ShMM 500 and ShMM 1500 the distribution is Monterey Linux see htto www pigeonpointcom library html userdocs for a User Guide For the ShMM 700 that distribution is Pigeon Point Linux The lowest layer in both cases is the firmware monitor which is called U Boot on all three ShMM variants 3 1 In This Section This section contains the topics listed below J ust click
126. Filter Jipmi header target 0x20 amp amp ipmi header source Expression Clear Apply 8 aaah sUse No Time Source Destination Protocol Info 1 0 000000 I2C 2 0x10 IPMI ATC Req Platform Event seq 0x00 4 0 101917 I2C 2 0x10 IPMI ATC Rsp Get Device ID seq 0x10 6 0 132643 I2C 1 0x10 IPMI ATC Rsp ATCA Get PICMG Properties seq 0x11 8 0 161710 I2C 2 0x10 IPMI ATC Rsp Set Event Receiver seq 0x12 9 1 237134 I2C 1 0x10 IPMI ATC Req Platform Event seq 0x00 11 1 263932 I2C 2 0x10 IPMI ATC Req Platform Event seq 0x01 14 1 296905 I2C 2 0x10 IPMI ATC Rsp Get Device ID seq 0x13 15 1 305104 I2C 2 0x10 IPMI ATC Rsp ATCA Get Device Locator Record ID seq 0x14 18 1 326371 I2C 1 0x10 IPMI ATC Req Platform Event seq 0x02 21 1 392419 I2C 2 0x10 IPMI ATC Rsp Reserve Device SDR Repository seq 0x15 23 1 408658 I2C 1 0x10 IPMI ATC Rsp Get Device ID seq 0x16 24 1 418148 I2C 1 0x10 IPMI ATC Rsp Get Device SDR seq 0x17 26 1 439321 12C 2 0x10 IPMI ATC Rsp Get Device SDR seq 0x18 28 1 540049 2C 1 0x10 IPMI ATC Rsp Get Device SDR seq 0x19 32 1 581004 I2C 1 0x10 IPMI ATC Rsp Get Device ID seq Oxld 33 1 590353 I2C 1 0x10 IPMI ATC Rsp Get Device SDR Info seq Oxle 25 1 692271 rap AvIn TOMT sayr gon Dlatfarm Cunnt can AvA bo gt Frame 21 10 bytes on wire 10 bytes captured b Inter Integrated Circuit Data v Intelligent Platform Management Interface Response to 17 Responded in 0 081195000 seconds b Header Reserve Device SDR R
127. Guide TYPE IP address DEFAULT None DESCRIPTION The default IP address used for shelf external RMCP based communication on the second network interface it is switched over between the redundant instances of the Shelf Manager This IP address is used only if the corresponding parameter is setto 0 0 0 0 inthe IPMI LAN Configuration P arameters for channel 2 If a non zero IP address is provided in the LAN Configuration Parameters the value provided in the Shelf Manager configuration file is ignored CAN BE OBTAINED FROM SHELF FRU INFO No DEFAULT_RMCP_N ETMASK IP address Variable The network mask for the network adapter used for RMCP communication This mask is used only if the corresponding parameter is setto0 0 0 0 inthe IPMI LAN Configuration Parameters for channel 1 and in the Shelf Manager IP Connection record in Shelf FRU Information The default value depends on the class of the default IP address used for the gateway for Shelf external RMCP based communication see parameter DEFAULT_RMCP_IP_ADDRESS For example for an IP address of class C this parameter is set to 255 255 255 0 Release 3 2 0 46 J uly 31 2013 DEFAULT_RMCP_N ETMASK2 Pigeon Point Shelf Manager User Guide IP address DEFAULT Variable DESCRIPTION The network mask for the second network adapter used for RMCP communication This mask is used only if the corresponding para
128. HZ BUS 99 MHz EMI 130 MHz GPMI 24 MHz T20 ready DRAM 128 MB SFGEN N250512A detected total size 64 MB A2F SPICOMM protocol v1 6 M3 firmware v0 9 FPGA design v0 44 0 0 A2F A2F firmware version 0 9 A2F Last reset cause HARD A2F Device type A2F060M3E FG256 A2F MSS clock frequency 40 MHz A2F Fabric clock frequency 20 MHz A2F Fast delay calibration 1330 cycles per 100uS A2F eNVM 128 KB 00000000 00020000 A2F eSRAM 16 KB 20000000 20004000 A2F Extram start 200022D0 RUPG booting from image 0 confirmed After a power cycle U Boot reports that there is no reliable upgrade in progress and the confirmed image set is being used The confirmed image set index is 0 Tn serial Out serial Release 3 2 0 246 July 31 2013 Pigeon Point Shelf Manager User Guide Err serial Net FECO 00 18 49 01 8f 26 FEC1 00 18 49 01 8f 27 FECO FECL Hit any key to stop autoboot 0 Creating 1 MTD partitions on spiO 0x0000001c0000 0x000002000000 mtd 5 UBI attaching mtdl to ubi0 UBI MTD device 1 is write protected attach in read only mode UBI physical eraseblock size 65536 bytes 64 KiB UBI logical eraseblock size 65408 bytes UBI smallest flash I O unit 1 UBI VID header offset 64 aligned 64 UBI data offset 128 UBI attached mtdl to ubi0 UBI MTD device name mtd 5 UBI MTD device size 30 MiB UBI number of good PEBs 484 UBI num
129. I ATCA Rsp Get Channel Authentication Capabilities seq 0x01 71 996590 80 240 102 57 gt 192 168 1 199 RMCP payload type RMCP Open Session Reques 71 998471 192 168 1 199 gt 80 240 102 57 RMCP payload type RMCP Open Session Respons Release 3 2 0 186 J uly 31 2013 Pigeon Point Shelf Manager User Guide gt gt gt gt gt gt gt 72 000184 80 240 102 57 essage 1 72 005487 192 168 1 199 essage 72 008539 80 240 102 57 essage 3 72 019149 192 168 1 199 Message 4 72 022416 80 240 102 57 gt Privileg evel seq 0x02 72033355 192 168 1 199 Privileg evel seq 0x02 72 057400 80 240 102 57 seq 0x03 72 059134 192 168 1 199 seq 0x03 72 059568 80 240 102 57 seq 0x04 72 062264 192 168 1 199 seq 0x04 72 062680 80 240 102 57 seq 0x05 72 066428 192 168 1 199 seq 0x05 72 068235 192 168 1 199 seq 0x05 72 101030 80 240 102 57 seq 0x06 72 102654 192 168 1 199 seq 0x06 192 168 1 199 80 240 102 57 192 168 1 199 80 240 102 57 192 168 1 199 80 240 102 57 192 168 1 199 80 240 102 57 192 168 1 199 80 240 102 57 192 168 1 199 80 240 102 57 80 240 102 57 192 168 1 199 80 240 102 57 RMCP RMCP RMCP RMCP IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA IPMI ATCA payl payl payl payl Req
130. M but with a small twist both the active and backup ShMM specify the same IP address for the redundancy interface but software assigns the next logical IP address to the ShMM with an odd hardware address For instance the default setting for the redundancy Ethernet port is 192 168 1 2 The odd addressed ShMM assumes the address 192 168 1 3 This way the active and backup ShMM can be identically configured but still assume unique IP addresses for the redundancy Ethernet link When the ShMMs use a non Ethernet network interface for communication between the redundant Shelf Managers see section 3 4 3 the second network interface can potentially be used for other purposes such as for support in the Shelf Manager for ShMC cross connects in accordance with the PICMG ECN 3 0 2 0 001 In that case the second network interface connects the Shelf Manager with one of the ATCA network hub boards To configure the Shelf Manager to support cross connects it is necessary to define the configuration parameter RMCP_NET_ADAPTER2 In that case if the backplane and hub boards also support cross connects the Shelf Manager uses the two network adapters RMCP_NET ADAPTER and RMCP_NET ADAPTERZ2 for RMCP communication Three usage models are available for this feature e active standby e active active and e bonded 3 4 2 1 Active Standby Usage of the Two Network Interfaces The two network interfaces can be used in an active standby way In this mode at
131. NFO gt backend c 764 2283039 bytes was written from file tmp _tmp_rfs to dev 4 1 lt INFO gt shmm700_hal c 1285 Write storage state available gt not available ALLOW lt INFO gt backend c 730 Burning Release 3 2 0 244 July 31 2013 Pigeon Point Shelf Manager User Guide lt INFO gt backend c 764 2110743 bytes was written from file tmp _tmp_app to dev 8 1 lt INFO gt backend c 808 Erasing user data on candidate bank lt INFO gt backend c 821 Copy ssh keys lt INFO gt shmm700_hal c 1189 Write confirmed 0 gt 0 ALLOW lt INFO gt shmm700_hal c 1194 Write candidate 1 gt 1 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 0 gt 1 ALLOW The reliable upgrade procedure resets the ShMM here and starts U Boot from the candidate Flash PowerPrep start initialize power Battery Voltage 1 46V No battery or bad battery detected Disabling battery voltage measurements Mar 21 201223 23 40 FRAC 0x92926152 memory type is DDR2 ait for ddr ready 1 power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory access ddr2 0x40000000 finish simple test U Boot 2009 08 Mar 21 2012 23 22 30 Freescale i MX28 family CPUs 297 MHz BUS 99 MHz EMI 130 MHz GPMI 24 MHz T2C ready DRAM 128 MB SFGEN N25Q512A detected total size 64 MB A2F SPICOMM protocol v1 6 M3 firmware v0 9 FPGA
132. On ShMM 700 you should see the following power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory accress ddr2 0x40000000 finish simple test U Boot 2009 08 Feb 24 2012 PowerPrep start initialize power Battery Voltage 1 44V No battery or bad battery detected Disabling battery voltage measurements Feb 24 201207 45 29 FRAC 0x92926152 memory type is DDR2 ait for ddr ready 1 07 44 12 M3 firmware v0 8 FPGA design v0 42 0 0 0 8 20 MHz 1330 cycles per 100uS 00020000 20004000 FECO 00 18 49 01 8f 26 FEC1 00 18 49 01 8f 27 0 Freescale i MX28 family CPU 297 MHz BUS 99 MHz EMI 130 MHz GPMI 24 MHz T20 ready DRAM 128 MB SFGEN N250512A detected total size 64 MB A2F SPICOMM protocol v1 6 A2F A2F firmware version A2F Last reset cause HARD A2F Device type A2F060M3E FG256 A2F MSS clock frequency 40 MHz A2F Fabric clock frequency A2F Fast delay calibration A2F eNVM 128 KB 00000000 A2F eSRAM 16 KB 20000000 A2F Extram start 200022D0 RUPG booting from image 0 confirmed In serial Out serial Err serial Net FECO FEC1 Hit any key to stop autoboot shmm700 Release 3 2 0 84 July 31 2013 Pigeon Point Shelf Manager User Guide 3 Echo current network settings shmmxx00 printenv rmcpaddr netmask gatewayip rmcpaddr 192 168 0 44 netmask 255 255
133. PE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO FAN_FULL_SPEED Number 0 The delay in seconds after Shelf No DELAY Manager startup or after a switchover during which the cooling algorithm does not check the number of present fan trays giving the existing fan trays enough time to activate This applies to the carriers where cooling management raises the fan speed to maximum if the actual number of fan trays in the shelf is fewer than what is specified in the Shelf Address Table FAN_LEVEL_STEP Number 1 The number of fan steps by which the Yes TOWN fan speed is decreased during operation of the cooling algorithm in the Normal state This parameter may be overridden by a ShMM carrier Specific cooling algorithm FAN_LEVEL_STEP Number 1 The number of fan steps by which the Yes UP fan speed is increased during operation of the cooling algorithm in the Minor Alert state This parameter may be overridden by a carrier specific cooling algorithm HPDL Boolean FALSE Turns on HPDL supportin the Shelf No Manager The carrier and chassis HPDL data and SDRs are taken from the FRU Information or from the files and are used to define the behavior of the platform plus the number and types of managed FRUs and sensors HPDL_ON_SUBSID Boolean FALSE Turns on support of HPDL information Yes TARY ERUS stored on subsidiary FRUS If TRUE the Shelf Manager looks for HP DL data and SDRs in the FRU I
134. PMB 0 SHELF_FRU_IN_EEPROM does not matter MIN_SHELF_FRUS minimum number of IPM controllers on IPMB 0 providing the Shelf FRU Information usually 2 Optionally SHELF_FRU_IPMB_SOURCE1 of the first designated source SHELF_FRU_IPMB_SOURCE2 IPMB address of the second designated source IPMB address SEEPROMS accessed locally by the Shelf LOCAL_SHELF_FRU TRUE Manager SHELF _FRU_IN_EEPROM TRUE MIN_SHELF_FRUS the number of SEEPROMSs providing the Shelf FRU Information usually 2 Flash file LOCAL _SHELF_FRU TRUE SHELF_FRU_IN_EEPROM FALSE MIN_SHELF_FRUS 1 Release 3 2 0 96 July 31 2013 Pigeon Point Shelf Manager User Guide 3 5 2 Setting up the Shelf FRU Information Since the contents of the Shelf FRU Information is crucial for successful management of the shelf itis necessary to set up the Shelf FRU Information on a fresh shelf before starting the Shelf Manager on it This procedure consists of the following steps e Creating a description of the shelf in a formalized text format INF format e Compiling the text description using the FRU Information Compiler e Placing the binary image of the FRU Information into the appropriate storage The first two steps are documented separately in the user manual for the FRU Information Compiler The last step is documented here and depends on where th
135. PROM However if the Carrier FRU Information is stored in an EEPROM the location and type of this EEPROM needs to be communicated to the Shelf Manager Carrier HPDL data cannot be used for that because of a chicken and egg problem in order to read carrier HPDL data the Shelf Manager has to already know the location or the Carrier FRU Information So the location of the Carrier FRU Information if it is stored in an EEPROM is specified as a carrier option CARRIER_FRU_LOCATION for the HPDL carrier part of the string value of the Shelf Manager configuration parameter CARRIER_OPTIONS If the Carrier FRU Information is located in a file on the ShMM file system the carrier option CARRIER_FRU_LOCATION must not be present Release 3 2 0 103 July 31 2013 Pigeon Point Shelf Manager User Guide The following syntax governs this carrier option CARRIER_FRU_LOCATION lt device type gt lt bus gt lt address gt lt size gt where e lt device type gt is the EEPROM device type See Section 3 of the Pigeon Point HPDL Reference for the list of supported EEPROM devices Typical values are AT24C16 AT24LC256 ADM1026 e lt bus gt is the number of the master only 12C bus where the EEPROM is located typically 0 e lt address gt is the 7 bit 2C address of the EEPROM device on that bus in hexadecimal e lt size gt is an optional argument that specifies the size of the occupied portion of the EEPROM thatis the number o
136. PU MPC83xx Rev 1 1 at 249 975 MHz Board ShMM 1500R PCI1 32 bit 33 MHz I2C ready DRAM 128 MB FLASH 64 MB PCI Bus Dev VenId DevId Class Int 00 17 1172 0001 00 OO In serial Out serial Err serial FPGA firmware version 1 12 carrier id 0 Net TSECO TSEC1 Hit any key to stop autoboot 0 Booting image at 00c0000 Image Name Linux 2 4 25 Created 2006 06 29 14 51 42 UTC Image Type PowerPC Linux Kernel Image gzip compressed Data Size 786177 Bytes 767 8 kB Load Address 00000000 Entry Point 00000000 Verifying Checksum OK Uncompressing Kernel Image OK Loading RAMDisk Image at 04c0000 Image Name sentry shmm1500 RFS Ramdisk Imag Created 2008 02 15 16 35 56 UTC Image Type PowerPC Linux RAMDisk Image gzip compressed Data Size 3826312 Bytes 3 6 MB Load Address 00000000 Entry Point 00000000 Verifying Checksum OK Loading Ramdisk to O7bfc0O00 end 07fa2288 OK shmm1500 login root On ShMM 700 Release 3 2 0 86 July 31 2013 Pigeon Point Shelf Manager User Guide shmm700 reset resetting 1 44V Wait for ddr ready 1 power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory accress ddr2 0x40000000 finish simple test U Boot 2009 08 Freescale i MX28 family CPU 297 MHz BUS 99 MHz EMI 130 MHz GPMI 24 MHz I2C ready DRAM 128 MB SFGEN N250512A detected total siz
137. Point Shelf Manager This document describes how to use the Shelf External Interface Reference Manager command line interface web interface Simple Network Management Protocol SNMP interface and Remote Management Control Protocol RMCP interface 1 1 1 Conventions Used in this Document This table describes the textual conventions used in this document Table 2 Conventions Used in this Document setenv This 10 point bold Courier font is used for text entered at keyboard in example dialogues which typically occur as one or more separate lines U Boot 1 0 2 Apr 18 2006 This 10 point normal Courier font is used for Ta abod ShMM output in example dialogues addmisc This 12 point bold Courier font is used for special text within normal paragraphs The types of such special text include command names file names configuration parameters and command parameters plus other text that could be entered by or displayed to a Shelf Manager user This fontis also used for command syntax definitions Get Device ID IPM commands defined by the IP MI Specification or as PICMG extensions are shown in the normal font surrounded by double quotes This matches the corresponding convention used in PICMG specifications Release 3 2 0 9 July 31 2013 Pigeon Point Shelf Manager User Guide 1 2 Additional Resources For more information about Pigeon Point products go to the Pigeon Point Web Site http www pi
138. Point Shelf Manager User Guide lt I gt 14 35 24 311 213 Controller 20 FRU 2 ATCA state set to M4 prev M3 cause 0 locked 0 lt i gt 14735724 315 214 Hot Swap event SA 0x20 FRU 2 sensor_number 0x03 M3 gt M4 cause 0 lt I gt 14 39 24 3226 238 SA 0x20 FRU 2 is OPERATIONAL lt I gt 14 35 24 393 217 IPMC Cooling Management cooling state switched from Unknown to Normal The Shelf Manager software on the ShMM 1500 is optionally delivered without encryption related code for reasons having to do with export regulations If present the encryption related code is located in the shared library Lib libpps_encryption so If the encryption library is absent the Shelf Manager does not allow opening R MCP sessions with encryption support When ordering ShMM 1500s you must specify which version is needed ShMM 1500R 250M32F64R ShMM 1500R 250M64F128R Encryption enabled ShMM 1500R 250M32F64R NE ShMM 1500R 250M64F 128R NE Encryption removed The following line in the log above lt I gt 14 35 18 818 199 shm_crypto_init initialization succeeded indicates that the encryption library has been found and loaded successfully Otherwise the following messages appear lt W gt 14 34 04 480 242 shm_crypto_init failed to load libpps_encryption so library err libpps_encryption so cannot open shared object file No such file or directory lt I gt 14 34 04 482 242 shm_crypto_init the Shelf Manager will not
139. RMCP packets being dropped during transfer If the difference in the sequence numbers of a received packet and the previous packet exceeds the window size the Shelf Manager closes the RMCP Session as a corrupted one SHELF_FRU_IN_E Boolean TRUE If TRUE the Shelf FRU Information is No ETEO retrieved from EEPROMs on the backplane in a carrier specific way if FALSE the Shelf FRU is obtained from a file on the Flash file system SHELF_FRU_IPMB Number 0 If defined non zero specifies the No SOURCE1 IPMB address of the first designated source of Shelf FRU Information in the shelf Shelf FRU is located at FRU 1 If this value is defined the search for the Shelf FRU on the IPMB is limited to the designated sources only SHELF_FRU_IPMB Number 0 If defined non zero specifies the No SOURCE2 IPMB address of the second designated source of Shelf FRU Information in the shelf Shelf FRU is located at FRU 1 If this value is defined the search for the Shelf FRU on the IPMB is limited to the designated sources only SHELF_FRU_TIME Number 5 The time interval in seconds during No OUT initialization that the Shelf Manager waits for Shelf FRU Information devices to be detected SENSOR_POLL_IN Number TERVAL Release 3 2 0 64 J uly 31 2013 SHELF_MANAGER _ CONFIGURATION __ IN_SHELF_FRU_I NFO Pigeon Point Shelf Manager User Guide Boolean DEFAULT FALSE DESCRIPTION
140. SE_DHCP TRUE By default the Shelf Manager uses the first DHCP server that answers the DHCPDISCOVER request to assign the IP addresses If more than one DHCP server is present in the network the configuration parameter DHCP_SERVER_ADDRESS can be used in the Shelf Manager Release 3 2 0 92 July 31 2013 Pigeon Point Shelf Manager User Guide configuration file etc shelfman conf to specify the IP address of the DHCP server to be used as follows DHCP_SERVER_ADDRESS 192 168 1 50 In that case only the specified DHCP server is used In addition if the configuration parameter DHCP_FOR_RMCP_ONLY is set to TRUE only the RMCP accessible Logical Shelf Manager IP addresses are assigned via DHCP In this case the private ShMM IP addresses if any are left untouched The DHCP server should be configured to provide a unique IP address for each Client Identifier To avoid IP address expiration the lease time of each address must be setas infinite time value OXF FFFFFFF If the configuration of the DHCP server changes over time it may be necessary to restart the procedure of IP address assignment on the DHCP client for the Shelf Manager The CLI command dhcp restart can be used for this purpose Another CLI command dhep status can be used to find out the current DHCP client status on the Shelf Manager which IP addresses have already been retrieved and assigned Please refer to the Shelf Manager External Inte
141. SNMP traps LAN alerts generated by the Shelf Manager as a result of PEF processing As allowed by the IP MI Specification the Shelf Manager does not support this extension and always uses the main VLAN ID that is assigned to the corresponding channel for such LAN alerts Each LAN configuration parameter that defines a VLAN ID for a LAN alert destination is read only in the Shelf Manager and identical to the VLAN ID for the corresponding LAN channel The default VLAN ID for a channel is usually 0 VLANs disabled but can be configured via the configuration parameters DEFAULT _VLAN_1ID for channel 1 and DEFAULT_VLAN_ID2 for channel 2 These defaults apply in the absence of the LAN configuration parameters on a Release 3 2 0 91 July 31 2013 Pigeon Point Shelf Manager User Guide ShMM ona fresh ShMM or after a Shelf Manager upgrade from a pre 2 6 version to the version 2 6 Once the LAN configuration parameters are defined and stored the Set LAN Configuration Parameters mechanism should be used to change the VLAN ID VLAN support is available on all ShMM variants 3 4 7 Assigning IP Addresses to the Shelf Manager via DHCP DHCP Dynamic Host Configuration Protocol and DHCP servers can be used to assign IP addresses to the Shelf Manager The following types of IP addresses can be assigned via DHCP e RMCP accessible addresses for one or both network interfaces e Private Shelf Manager addresses for one or both network interfaces fo
142. Section 3 adds ShMM 700 coverage Section 3 3 clarifies the description of the configuration parameter ATCA_TESTER_COMPATIBILITY Section 3 3 describes new configuration variables AXIE_TIMING ASYMMETRIC MATCH ENABLE _LOCKS_LOGGING ENABLE _RTC_TRICKLE_CHARGER ISOLATE_MUX_BUS Section 3 3 describes new IntegralHP I related configuration variables INTEGRALHPI_SHELF_ENTITY_LOCATION INTEGRALHPI_DEFAULT_AUTO_INSERT_TIMEOUT INTEGRALHPI_DEFAULT_AUTO_INSERT_TIMEOUT_MSEC INTEGRALHPI_DEFAULT_AUTO_EXTRACT_TIMEOUT and INTEGRALHPI_DEFAULT_AUTO_EXTRACT_TIMEOUT_MSEC Section 3 4 5 describes the new utility hwri that replaces ep1d utility on the ShMM 700 Section 3 5 2 adds ShMM 700 coverage Section 3 10 1 clarifies the behavior of the system clock before the time is successfully obtained from the time server Section 4 3 corrects the example of the Shelf Manager startup log Section 4 4 adds ShMM 700 coverage Section 4 8 adds ShMM 700 coverage Section 4 9 adds ShMM 700 coverage Section 4 9 adds coverage of the new configuration parameter ISOLATE _MUX_BUS Section 4 12 clarifies current support on the ShMM 500 1500 Release 3 2 0 261 July 31 2013 Pigeon Point Shelf Manager User Guide Section 5 notes current absence of IP MB trace facility on ShMM 700 Section 6 adds ShMM 700 coverage Section 7 notes ShMM 500 1500 focus for reliable upgrade coverage New section 8 covers ShMM 700 specific reliable upgrades New s
143. Separated from the IPMB address by a dot Site type and site number are also each separated by a dot An additional FRU device ID or site number for this IPMB address can be designated separated by semicolon A site number can be specified as a wildcard which makes the descriptor apply to all FRUs with the specified site type on a given IPMC An IPMB address is hexadecimal by default and FRU device ID site type and site number are decimal by default For example the value 82 0 86 8A 7 2 5 90 1 3 A0 7 applies synchronous assignment to the IPMC 82h FRU 0 all FRUs associated with the IPMC 86h AMC 2 and 5 associated with the IPMC 8Ah FRU 1 and FRU 3 associated with the IPMC 90h and all AMC modules associated with the IPMC AOh 4 11 HPISystem Event Sensor This discrete sensor is associated with the logical Shelf Manager IPMB address 20h has number 0 on LUN 3 the name HP I Sys Event and OEM event reading type code DBh Its purpose is to enhance interaction between the Shelf Manager and Pigeon Point HP implementations IntegralHP and Pigeon Point OpenHPI This sensor sends IPMI events in a special format to Signal HP I implementations that changes have occurred within the Shelf Manager for example some fields within a Multi record in the Shelf FRU Information have been updated The HPI implementation is expected to process such an event and perform corresponding actions to react to the new conditions for example to reread
144. TER SIPDEVICE Itis possible to use an alias interface of bondo if itis necessary to have a permanent IP address on the ShMM while the RMCP address floats between active and standby ShMMs i e RMCP_NET_ADAPTER bond0 1 The parameter RMCP_NET_ADAPTER2 must not be set while using the bonding interface as an RMCP adapter The bonding mode assumes single RMCP network adapter usage from the Shelf Manager perspective so the parameter USE_SECOND_CHANNEL must be set to FALSE 3 4 3 Using the ShMM 500 and ShMM 700 Alternate Software Redundancy Interface In SHMM 500s and ShMM 700s configured for ShMC Cross connect operation two additional network interfaces are implemented over the two USB connections In this configuration they always connect the two redundant Shelf Managers These interfaces are named usb0 and usbl The interface usb0 always exists while the interface usb1 exists only if the interface usb0 is active on the peer Shelf Manager which means that the peer Shelf Manager is physically installed and Release 3 2 0 79 July 31 2013 Pigeon Point Shelf Manager User Guide running Also the interfaces are cross connected usb0 on the first Shelf Manager is connected to usb1 on the second Shelf Manager and vice versa The Shelf Manager supports usage of the USB network interfaces for communication between the redundant Shelf Managers To use this feature it is necessary to define two redundancy network adap
145. Trap Format v1 0 specification In this format event data is packed as binary in a single variable with OID 1 3 6 1 4 1 3183 1 1 1 T0 facilitate parsing this format Pigeon Point makes available a sample application snmptc lf PET_FORMAT is Set to 1 the Shelf Manager generates trap messages in plain text format packing event into single variable with OID 1 3 6 1 4 1 3183 1 1 2 as an ASCII String This format is convenient for visual monitoring without specialized tools lf PET_FORMAT is Set to 2 Shelf Manager generates trap messages in multi variable format packing each field of the event event number timestamp generator etc as a separate variable with OIDs in the 1 3 6 1 4 1 3183 1 1 3 1 13 range Processing traps in this format requires the MIB file Itis possible to change the Platform Event Trap format dynamically using the clia setpefconfig pet_format lt N gt command with lt N gt 0 2 The command clia getpefconfig pet_format shows the currently used PET format Here is some sample output of the standard Linux snmpt rapd tool for the supported formats PET FORMAT 0 Received 111 bytes from 192 168 0 2 192 168 0 2 Enterprise Specific Trap 2453248 Uptime 191 days 4 23 00 02 PPS PET MIB ipmi trap data Hex STRING BF CE 7A 60 AE 50 11 DC 00 80 00 50 C2 3F CD 24 00 01 12 BE 74 62 00 00 20 20 02 20 Cl 00 00 00 FF FF 00 00 00 00 00 19 OA 40 00 00 00 00 Cl PET_FORMAT 1 Received 251 bytes from 1
146. U 0 op_state M0 power_cycle 0 lt I gt 14 35 23 743 214 Hot Swap event SA 0x20 FRU 1 sensor_number 0x02 MO gt M1 cause 0 lt I gt 247352 23 745 214 State M1 sa 20 FRU 1 op_state M0 power_cycle 0 lt I gt 14 35 23 755 214 Hot Swap event SA 0x20 FRU 0 sensor_number 0x00 M1 gt M2 cause 2 lt I gt 14 35 23 763 217 ipmc_cooling_scan_sensors Reread sensors on the SA 0x20 lt I gt 14 35 23 767 228 SA 0x20 FRU 0 is ACTIVATING lt I gt 14 35 23 779 214 Hot Swap event SA 0xfc FRU 0 sensor_number 0x00 M1 gt M2 cause 2 lt I gt 14 35 23 1787 217 ipmc_cooling_scan_sensors Reread sensors on the SA 0xFC lt I gt 14 35 23 801 228 Tasklet ACTIVATE 20 0 is_local_address 1 lt I gt 14 35 23 811 221 Initial indicator for SA 0x20 27 state new lt I gt 14 35 23 814 221 Initial indicator for SA Oxfc 33 state new lt I gt 14 35 23 817 221 sdrrep fetched new SDRs time 6 lt I gt 14 35 23 821 213 Controller 20 FRU 0 ATCA state set to M3 prev M2 cause 1 locked 0 lt I gt 14 35 23 824 213 Controller 20 FRU 0 ATCA state set to M4 prev M3 cause 0 locked 0 lt I gt 14 35 23 828 214 Hot Swap event SA 0x20 FRU 0 sensor_number 0x00 M2 gt M3 cause 1 lt I gt 14 35 23 835 217 ipmc_cooling_scan_sensors Reread sensors on the SA 0x20 lt I gt 14 35 23 842 217 ipmc_cooling_scan_sensors Reread sensors on the SA 0x20 lt I gt 14 35 23 847 214 Hot Swap event SA 0x20 FRU 0 sensor_n
147. U INFO Yes CARRIER String 16 PPS The name of the specific carrier board No on which the ShMM is installed CARRIER_OPTION String 256 Variable The carrier specific options defined No S separately for each supported carrier By default this parameter is set from the environment variable CARRIER_OPTIONS CONSOLE_LOGGIN Boolean FALSE Output log messages to the console No C_ENABLED on which the Shelf Manager was started COOLING_FAN_DE Number 0 The minimum timeout between Yes CREASE_TIMEOUT Successive decrements of the fan Speed during operation of the cooling algorithm in Normal state Should be a multiple of COOLING_POLL_TIMEOUT if not itis rounded up to the next multiple If the parameter is omitted or Set to 0 this timeout is equal to COOLING_POLL_TIMEOUT Release 3 2 0 July 31 2013 COOLING_FAN_IN CREASE_TIMEOUT Pigeon Point Shelf Manager User Guide Number DEFAULT DESCRIPTION The minimum timeout between Successive increments of the fan Speed during operation of the cooling algorithm in Minor Alert state Should be a multiple of COOLING_POLL_TIMEOUT if not itis rounded up to the next multiple If the parameter is omitted or Set to 0 this timeout is equal to COOLING_POLL_TIMEOUT CAN BE OBTAINED FROM SHELF FRU INFO Yes COOLING_IGNORE _LOCAL_CONTROL Boolean FALSE Do not use local control capabilities on fan devices Sh
148. Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1200 Write confirmed 1 gt 0 ALLOW lt INFO gt shmm700_hal c 1217 Write upgrade_state in progress 2 gt confirmed 4 ALLOW lt INFO gt shmm700_hal c 1280 Write upgrade watchdog 1 gt 0 ALLOW lt INFO gt frontend c 46 Last upgrade stages log O initial PASSED 1 before storage mount PASSED 2 after storage mount PASSED 3 after marking image set invalid PASSED 4 after u boot burn PASSED 5 after kernel burn PASSED 6 after rootfs burn PASSED 7 after storage umount PASSED 8 after cleaning previous apps set PASSED 9 after writting new apps set PASSED 10 after sync PASSED 11 after processing var and etc PASSED 12 after boot select PASSED 13 after reboot PASSED 14 after confirm PASSED 8 8 2 Example 2 This example shows a reliable upgrade of the application package image only copying ete and var nvdata non volatile directories from confirmed to candidate locations The application package image is taken from an FTP server at the IP address 192 168 1 253 The path to the application package image on the
149. a Shelf Manager command line invocation is defined as follows shelfman lt options gt amp or Release 3 2 0 128 July 31 2013 Pigeon Point Shelf Manager User Guide daemon f shelfman lt options gt The following options are recognized h lt address gt v lt verbosity gt c lt path gt cs w wt 1 c s g lt ip_address gt sf port lt port gt ph lt IPMB addr1 gt lt IPMB addr2 gt i lt command_line gt A detailed description of these options is given below h lt address gt This option overrides the hardware address of the FRU site where the Shelf Manager resides This hardware address is used as the IP MB address for the Shelf Manager the address that is based on its hardware address separate from 20h and is treated as a hexadecimal number This option can be used if no hardware address is automatically available for the Shelf Manager If this option is not used the Shelf Manager obtains the hardware address in a carrier specific way the IP MB address is the hardware address multiplied by 2 v lt verbosity gt Set the initial debug verbosity mask see section 3 3 2 both for the system log and for the console If this option is not present the values of the parameters VERBOSITY and VERBOSITY_CONSOLE in the file etc shelfman conf determine the initial debug verbosity mask c lt path gt Path to non volatile configuration files The default path is var nvdata
150. able upgrade of ShMM firmware but it is Substantially different in implementation details and therefore is described in a separate section of this document see Section 8 7 1 In This Section This section contains the topics listed below J ust click on a topic to go to it Firmware Reliable Upgrade Procedure Overview Flash Partitioning The var upgrade File System Reliable Upgrade Procedure Status File Reliable Upgrade Utility Reliable Upgrade Utility Use Scenarios Reliable Upgrade Examples 7 2 Firmware Reliable Upgrade Procedure Overview Monterey Linux provides a reliable upgrade procedure for the firmware images on a running and functioning ShMM The procedure supports upgrade of the U Boot firmware the Linux kernel and the Linux root file system or an arbitrary combination of these three images If a software upgrade attempt fails for instance due to installation of a faulty U Boot firmware image that is not capable of booting the ShMM ora Shelf Manager that can t start the reliable upgrade procedure automatically falls back to the previous version of the firmware in persistent Flash Flash storage on either the ShMM 500 or the ShMM 1500 is divided into two areas When a stable set of firmware is established in one of these areas itis designated the persistent area When new firmware is installed it goes in the other area which is initially designated provisional Once a new Set of firmware in the provisional area is vali
151. ables first copy 1 25 0 25 dev mtd6 Not mounted U Boot environment variables second copy 1 5 0 25 dev mtd8 Not mounted UBIFS cache used for boot acceleration Release 3 2 0 222 July 31 2013 Pigeon Point Shelf Manager User Guide OFFSETINFLASH SIZE IN Device NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 1 75 30 25 dev mtd10 Not mounted Boot partition UBI temporarily with one UBIFS mounted by U Boot volume named to load images and boot contains Linux by the reliable kernel and RFS upgrade utility to images write images 32 32 dev mtd12 Nar User partition UBI etc with one UBIFS volume named user contains application package images Shelf Manager files other application files and user data The following table lists names of the files and directories on UBIFS volumes that contain other firmware images The file names for A2F 060 upgrade files are also listed the A2F060 upgrade process is described in section 8 6 Table 26 File Names for Firmware Images Stored as UBIFS Files FILE NAME FLASH PARTITION DESCRIPTION ulmage 0 Boot Linux kernel image 0 ulmage 1 Boot Linux kernel image 1 rfs 0 Boot Root file system RFS image 0 rfs 1 Boot Root file system RFS image 1 a2f upgrade dat Boot A2F060 upgrade image present only during an A2F060 upgrade a2f auto rollback dat Boot Rollback A2F 060 upgrade imag
152. addr rootpath rootfs ipdevice eth0 ipldevice usb0 ip2device ethl ipaddr 192 168 0 22 ipladdr 192 168 1 2 gatewayip 192 168 0 1 rmcpaddr 192 168 0 2 netmask 255 255 255 0 hostname shmm700 flash_reset n password_reset n logging ram timezone UTC rc_ifconfig n start_rc2_daemons y rce2 etc rce shmm700 hpdl set_mtdparts set mtdparts mtdparts SPI 0 S mtdparts1 bootargs_common set bootargs console S console mtdparts m25p80 1 S mtdpartsl S quiet run mac_override bootargs_initrd run bootargs_common set bootargs bootargs root dev ram bootargs_nfs run bootargs_common set bootargs S bootargs root dev nfs ip S ipaddr ethO off nfsroot serverip S rootpath v3 tcp mount_ubifs run set_mtdparts load_ubifs ubifsload loadaddr uImage S image_sel ubifsload S rfsaddr rfs image_sel load_ubifs_alt ubifsload loadaddr uImage S alt_image ubifsload S rfsaddr rfs alt_image ubifs run bootargs_initrd run load_ubifs bootm loadaddr S rfsaddr echo ERROR booting active image failed trying alternate image run load_ubifs_alt bootm S loadaddr S rfsaddr echo ERROR bad image s aborting nfs tftp S loadaddr S bootfile run bootargs_nfs bootm loadaddr net run bootargs_initrd tftp S loadaddr S bootfile tftp S rfsaddr S ramdisk bootm S loadaddr rfsaddr Several of these environment variables need to be reconfigured with values that are appropriate
153. adial IPMB 0 1 0cccese 145 4 5 3 Operation in SAMM 700 based Shelves with Radial IPMB 0 ccccscccceeee 148 4 6 AUTOMATIC SEL TRUNCATION c cccccccccscsesseeececcceeseseeseaeceeeeseseseuaeeeseeececsesseaeseeeeeens 150 4 7 COOLING STATE SENSORS cccccceeseceeeceeeneeeeeaeceeneeeeeeeeaeeeeaaeseaaeseeeecnaeesesaesseeeeeeeees 150 4 8 DEADLOCK DETECTION ss iccsiscsccncccestcnevaxadcnwvccndccceneaddicvccedeneeseadanctacedendeacs dectecs facewasdddaetes 152 4 9 MASTER ONLY C BUS FAULT ISOLATION ccccccccccccececeeeeeseeeeeeeeeeeeeseeeeeeeeeeeeeeeeeseeeeenenes 156 4 9 1 Master only PC Bus Fault Isolation Original Approach Single Multiplexer or Switch Not HPLD DA S Q c ccccccssssceceesnececeenececssnseeecseasesecsensecesseaseseseesaesscsesaeseesesseesenaas 157 4 9 2 PC Bus Fault Isolation HPDL based Approach Multiple Multiplexers or Switches 157 4 10 ASSIGNMENT OF LAN CONFIGURATION PARAMETERS TO BOARDS AND MODULES 159 4 10 1 Structure and Composition of the Supported Parameters ccccccccecseeeees 159 4 10 2 Obtaining LAN Configuration Parameters on the Shelf Manager Level 159 4 10 3 Dispatching LAN Configuration Parameters to BoardS cccccccsccessteteseseetees 163 4 10 4 Synchronous Assignment of LAN Configuration Parameters to Boards 164 4 11 HPI SYSTEM EVENT SENSOR c cccsesceceeceeeeseeeesaeeeeecesaeeesaeeeeaaeseeaesseeeeaaeseeeeeeeeeaeen
154. ais obi wa ese siege Se i edie ae ve Sete E nis ge edges ac sabe pees 243 9 HPI BASED SHELF MANAGER UPGRADE cccssecsseeeeeeeeeeseeeeseeeeneeeeeeeeeeseesesenenseees 251 APPENDIX A CONVERTING TO SHMM 700 BASED SHELF MANAGERS IN SHMM 500 MANAGED SHELVES a ar a aa aa a aaraa a er Aaa aAa aaraa aa Aa Sarahani Rain 253 APPENDIX B CUSTOMER SUPPORT nsanssnnsennsennnennnennnennnnnnnnnnnnnnnnnnnennnnnnn nunn nnnn annn nnnn nnana 254 APPENDIX C REVISION HISTORY sex inciastcccteect ceccetseteeseced tececedbcupveads eHris aa aserite niusa 255 C 1 RELEASE 2 NO eriaren eir EEEE EEEE EE ET exaveell osbene EEE 255 C 2 RELEASE 252 0 occ chictevecnsiectitensts cxcnnaccten bigde E A ON R 255 C 3 RELEASE 2 310 reip arena RE EEEE AEEA E EEA ERER 255 C 4 RELEASE 2 A oee aree E N EOG EE A TEEN EEG 256 C 5 RELEASE 2A Treppie a RA N EEEE EE A AE EEA A ERER 256 C 6 RELEASE 2 A 2 oop aree E EA N EEE E T EA ER 256 C 7 RELEASE 2244 ois ccc veiien slothvnsdt eesaedcddenbaadeeyan sd EAE EA SE E REE AEREE E 257 C 8 RELEASE 2 59 O ccscssuvesssstt tence ceessied We biadenpaveiel tients A E EE SEEE EEE 257 C 9 PREEEASE 2202 co feaescoicescceet reas lecesescttnn py eae ceueastecenyceceoeueeshuot weeta ocumeasbeeveessaqesesitoeweex sey 257 GAO RELEASE 2 5 3 seis lt sdceuestcneyies sigan cnsnsan dues cb danepelceensey ecdancs scteguetapdaney sabecvesdesguneaessectesddenses 258 Git RELEASE 2 6 0 nasini a E diet R E A ia 258 G12 RELEASE 2 6 oisi erneer anaa sie a a
155. aken from the local tmp which implies that they have already been copied there in some unspecified way The U Boot image is taken from tmp u boot bin the kernel image is taken from tmp sentry kernel the RFS image is taken from tmp sentry rfs The upgrade procedure is started from the serial console Comments are interspersed in the console log to provide additional background on the steps of the upgrade procedure First the rupgrade_tooL1 is started from the command prompt The parameters show that all three Flash images are to be updated with the Shelf Manager non volatile data and configuration file preserved as well rupgrade_tool s k sentry kernel r sentry rfs u u boot bin hook etc_copy v rupgrade_tool PLB is 5 rupgrade_tool EEPROM page saved rupgrade_tool persistent flash is 0 rupgrade_tool provisional flash is 1 rupgrade_tool copying image s The upgrade utility attempts to invoke a validation script to check the images in tmp currently Supplied If any of the specified file designators is not found in tmp the utility stops and a message like the following is produced rupgrade_tool cannot open tmp u boot bin for reading rupgrade_tool failed to copy images to flash The utility proceeds to copy the images to the specified destinations in provisional Flash Release 3 2 0 211 July 31 2013 Pigeon Point Shelf Manager User Guide rupgrade_tool invoking scripts step4v
156. alue in the multiplexer switch register corresponds to the faulty bus because it monitors all write accesses to the multiplexer switch registers and remembers the last value before the occurrence of the fault e When the ignore count is exhausted an attempt is made to access the faulty bus and check whether the fault is still present If the attempt fails the ignore count is reset to its original value If the attempt succeeds then the recorded fault condition for that bus is removed and operation continues normally The default logical 12C bus number on which the 12C multiplexer resides can be changed with the configuration parameter LSOLATE_MUX_BUS 4 9 2 RC Bus Fault Isolation HPDL based Approach Multiple Multiplexers or Switches In this mode the isolation algorithm is controlled by the HPDL description of the carrier and the shelf The original isolation approach in this case must be turned off that is the configuration variable ISOLATE MUX _ON_GPIO8 must be set to FALSE in the configuration files The typical HPDL description of an 12C multiplexer supporting reset looks as follows DEVICE Multiplexor PCA9545 5 0x70 SIGNALS Channels Channels 6 7 8 9 R eset 8 Mux solated IsolatedC hannelMask MuxisiCh1 Channell_lsolated MuxislCh2 Channel2_lIsolated Release 3 2 0 157 J uly 31 2013 Pigeon Point Shelf Manager User Guide MuxisiCh3 Channel3_ Isolated MuxisiCh4 Channel4_Isolated
157. ameters to boards or modules in a shelf that implement LAN channels in the local management controller Section 7 6 adds a note about reliable upgrade from an encryption present RFS to an encryption absent RFS C 14 Release 2 6 4 2 Section 3 4 7 adds a description of the DHCP client restart and status commands describes provisions for storing in a flash file a TFTP server name and bootfile name received from the DHCP server C 15 Release 2 6 4 4 Section 3 3 adds coverage of a new configuration parameter SWITCHOVER_ON_BROKEN_LINK_BACKUP_DELAY Section 3 4 2 1 provides user guidance for a special situation when a hub board is inserted into a shelf with cross connect Shelf Manager links where no hub boards were present before C 16 Release 2 7 0 Section 3 3 adds new configuration parameters BOARD_LAN PARAMETERS CHANNEL SYNCHRONOUS BOARD_LAN PARAMETERS USE_DHCP COOLING_KEEP_POWERED_OFF_FRUS_IN_M1 The increased length of the configuration parameter string BOARD_ LAN PARAMETERS CHANNEL LIST is noted Section 3 4 2 1 adds a description of active standby management of the network interface on the backup Shelf Manager when RMCP address propagation is in effect Section 3 7 1 adds a description of the effects of the new configuration variable COOLING_KEEP_POWERED_OFF_FRUS_IN_M1 anda description of the handling cold sensitive FRUs in the default cooling algorithm Release 3 2 0 259 July 31 2013 Pigeon Point Shelf Ma
158. an z timeout value for all exposed HPI resources INTEGRALHPI_DE Number 0 This variable defines the seconds Yes ape gern a oa portion of the default auto insertion timeout value for this IntegralHP domain A negative value causes the timeout to be set to SAHPI_TIMEOUT_BLOCK INTEGRALHPI_DE Number 1 This variable defines the milliseconds Yes FAULT AUTO INS portion of the default auto extraction ERT_TIMEOUT_MS FC timeout value for this IntegralHP domain INTEGRALHPT_DO Number 1 The HPI domain ID of the current No BEEN shelf used to differentiate on the HPI client level among multiple shelves whose Shelf Managers support HPI Release 3 2 0 54 July 31 2013 Pigeon Point Shelf Manager User Guide TYPE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO INTEGRALHPI_SH Number The entity location for the Yes a AdvancedTCA Shelf HPI Resource used to differentiate on the HPI client level among multiple shelves whose Shelf Managers support HPI IPMB_ADDRESS Number 0 The IPMB address of the Shelf No Manager overriding the hardware address The value of O causes the Shelf Manager to read the hardware address from hardware and set IP MB address to hardware address 2 IPMB_LINK_ISOL Number 1 In radial shelves if an IP MB link is Yes ALIONCTIMEOUT disabled due to the isolation algorithm the link is automatically enabled after this time interval in seconds 1 the default stands for forever
159. and their names Default value is 512K uboot0 512K uboot1 256K envvars0 256 K envvars1 256K ubifs_cache 30976K boot 3 2M user See the ShMM 700R Hardware Architecture Specification for more information net This variable can be used as a replacement for boot cmd as a means of booting a kernel and RFS image from TFTP Use run net netmask Network netmask default value is 255 255 255 0 nfs This variable can be used as a replacement for bootcmd as a means of booting and running with an NFS mounted root filesystem The Monterey Linux User Guide sample NFS project provides details for ShMM 500 and ShMM 1500 users Note the kernel image is fetched from a TFTP server by default password_rese t Instructs Linux to restore factory default password for user root which is the empty password Default is n post_normal Determines the list of POST tests that are executed on each boot up If not set compile time default settings are used The test names listed in a value of this variable are separated by space characters post_poweron Determines the list of POST tests that are executed after power on reset only vs on each boot up If not set compile time default settings are used The test names listed in a value of this variable are separated by Space characters serial This variable contains the ShMM serial number assigned during manufacturing The variable is set automa
160. approaches to event processing in the external event handler e Read the standard input line by line and process incoming data In that case one instance of the external event handler is spawned in the beginning and handles all subsequent events e Read one line from the standard input process the data on that line and exit In that case the Shelf Manager spawns a new instance of the external event handler for the next event Here is an example of a simple external event handler script bin sh while read line do raw eval line if z Sraw then do event processing here echo event Sraw fi Release 3 2 0 120 July 31 2013 Pigeon Point Shelf Manager User Guide 3 13 Configuring the Platform Event Trap Format The Shelf Manager sends SNMP Traps as a result of event processing when an Alert action is initiated or as a result of the Alertimmediate command which is used to test Alert destinations The Platform Event Trap formatis controlled by the PET_FORMAT configuration variable Currently there are three supported formats All formats use the same enterprise OID iso 1 org 3 dod 6 internet 1 private 4 enterprises 1 wired_for_man agement 3183 PET 1 version 1 but have different OIDs for the embedded variable s Pigeon Point supplies a separate MIB file for supported SNMP traps By default or with PET_FORMAT Set to O the Shelf Manager generates trap messages according to IP MI Platform Event
161. ariable is undefined by default the RFC 868 rdate protocol is used This variable should be set to ntp to enable the NTP protocol The usage of the other variables is identical their usage with RFC 868 rdate and is described below To enable obtaining the network time via the RFC 868 protocol rdate over TCP itis necessary to define the U Boot variable time_server and optionally the additional variable timezone The variable time_proto should be left undefined or set to rdate The variable time_server contains the IP address of the time server that the Shelf Manager queries for the system time after startup This server should support RFC 868 over TCP as required by the rdate utility or support NTP as required by the ntpdate utility This variable is propagated to the Linux level as the environment variable TIMESERVER If this variable is set the startup script etc netconfig Starts the script etc timesync as a daemon which runs in an endless loop and queries the time server with a default interval of 300 seconds To change this interval edit the script etc timesync and change the value of the variable INTERVAL If a query fails the script starts polling the server more frequently with the default interval of 30 seconds this new interval is the value of the script Release 3 2 0 115 J uly 31 2013 Pigeon Point Shelf Manager User Guide variable INTERVAL2 When a subsequent query succeeds the polling rate reverts back to INTERVAL
162. ask of the sensor can be retrieved in many ways e g with the CLI command sensordata The sensor also generates events when the state mask changes The format of the event message is given in the following table BYTE DATA FIELD 1 Event Message Rev 04h IPMI 1 5 2 Sensor Type DEh Redundancy and CPLD state 3 Sensor Number 80h 4 Event Direction 7 0b Assertion Event Type 6 0 6Fh General Availability 5 Event Data 1 7 4 7h previous state and severity in Event Data 2 OEM code in Event Data 3 3 0 The highest new offset 00h The current Shelf Manager is Active with no Backup Olh The current Shelf Manager is Active with a Backup 02h The current Shelf Manager is a Backup 04h The Shelf Manager is a Backup but the remote presence bit is not set 05h The Shelf Manager is a Backup but the remote switchover request bit is not set 06h The Shelf Manager is a Backup but the CPLD Active bit is set 07h The Shelf Manager is Active with a Backup but the remote presence bit is not set 08h The Shelf Manager is Active with a Backup but the remote healthy bit is not set 09h The Shelf Manager is Active with a Backup but the CPLD Active bitis not set OAh The local presence bitis not set for the current Shelf Manager OBh The Shelf Manager is Active with no Backup but the remote healthy bit is set OCh The Shelf Manager is Active with no Backup but the remote switchover request bi
163. at in order to facilitate upgrades from non HPDL to HPDL based configurations the HPDL based backup Shelf Manager can coexist with a non HPDL active Shelf Manager but only when a reliable upgrade operation is in progress 4 4 2 Redundancy and CPLD State Sensor Both active and standby Shelf Managers expose a sensor that indicates the high level redundancy State of the ShMM along with the state of the low level redundancy bits exposed by the CPLD and redundancy related exceptional conditions in the CPLD if any This discrete sensor is associated with the physical Shelf Manager has number 128 name CPLD State and OEM event reading type code DEh On the ShMM 700 where there is no CPLD and the programmatic interface to HRI is different this sensor has the name HWRI state though it is fully state compatible with its counterpart on the ShMM 500 1500 On the ShMM 500 1500 the sensor reads the second byte of the CPLD CSRO register bits 8 15 which contains various redundancy related data such as the Remote Presence Remote Healthy Remote Switchover Request Local Presence and Active bits For a detailed description of the CPLD register values please see the ShMM Hardware Architecture specification On the ShMM 700 the sensor uses the ShMM 700 HRI programmatic interface to obtain the values of the HRI bits The state mask reflects the high level redundancy state normal state of the ShMM as well as various low level redundancy re
164. at the var upgrade J FFS2 partition is composed of two non contiguous Flash blocks 0 5 MBytes each residing in both the lower and upper halves of the Flash device Monterey Linux takes advantage of the ability of the Linux MTD and J FFS2 layers to Support a file system in non contiguous Flash sectors in order to implement var upgrade this way Another feature of the FFS2 file system that makes var upgrade work for purposes of the reliable upgrade procedure is that the J FFS2 internal structures do not create any dependencies such as linked lists based on Flash sector numbers or absolute offsets in Flash Instead when mounting a file system on a partition the J FFS2 scans all the Flash sectors comprising the partition and recreates the logical content of a file system in an internal in RAM representation This feature ensures that regardless of which half of the Flash the ShMM 500 has booted from Linux is able to mount var upgrade as a J FFS2 file system and make use of the previous content of the file system 7 5 Reliable Upgrade Procedure Status File The software reliable upgrade procedure maintains the status of the most recent upgrade procedure session in the file var upgrade status residing in a dedicated file system var upgrade which is mounted by Linux regardless of which Flash the ShMM has booted from If the file exists it contains the status of an upgrade procedure session that either is in progress presently or has recentl
165. ata and the Shelf Manager configuration file are preserved Release 3 2 0 207 July 31 2013 Pigeon Point Shelf Manager User Guide e copy the scripterases both the provisional etc and provisional vax directories then copies the full contents of the etc and var directories onto the provisional partition In this case not only the configuration but also the executable files placed to var bin is copied and overrides executable files with the same name from the RFS image This mode of operation is useful if the directory var bin contains some special executables e g a special version of the Shelf Manager or other utilities that must be preserved across the upgrade e flip the script overrides all other hook options causing just the termination of the ShMM flipping of the Flash banks and a reboot of the ShMM with the upgrade WDT active Note When upgrading ShMM software from an encryption present RFS to an encryption removed RFS always use the erase option to repopulate the directory etc from the RFS This is because some files in the etc directory in an encryption present RFS are encrypted and are not readable in a non encrypted RFS during the reliable upgrade they should be replaced from the new encryption removed RFS where they are not encrypted The script returns 0 on success and non zero for failure If a non zero value is returned the upgrade procedure is terminated The utility starts the upgrade WDT with a 12 8
166. atile data The utility then starts the upgrade WDT and reboots rupgrade_tool invoking scripts step4h etc_copy etc upgrade step4hshm Stopping Shelf Manager Done etc upgrade step4hshm Cleaning new var partition Done etc upgrade step4hshm Cleaning new etc partition Done etc upgrade step4hshm Copying etc contents Done etc upgrade step4hshm Copying var nvdata contents Done etc upgrade step4hshm Upgrade complete rupgrade_tool image s copy OK rupgrade_tool watchdog started rupgrade_tool selected provisional flash rupgrade_tool reboot Restarting system At this point the telnet session is closed after a certain inactivity period after several seconds itis possible to reconnect to the target again and check the status of the reliable upgrade by invoking rupgrade_tool w telnet 192 168 1 174 Trying 192 168 1 174 Connected to 192 168 1 174 Escape character is BusyBox on shmm 174 login root BusyBox v0 60 5 2005 05 07 17 27 0000 Built in shell msh rupgrade_tool w Recent upgrade status PLB is 9 EEPROM page saved persistent flash is 0 provisional flash is 1 copying image s 4 copying sentry shmm1500 rfs from 192 168 1 253 tftpboot ru ppc to tmp using ftp protocol 4 invoking scripts step4v r sentry shmm1500 rfs proto ftp 192 168 1 253 tftpboot ru ppc admin hook etc_copy 4 copying sentry shmm1500 rfs from tmp to dev mtdchar
167. ave different IP MI sequence numbers AUTO_SEND_MESS Boolean TRUE Automatically convertan RMCP No AGE request sent to a non Shelf Manager IPMB address into a Send Message request directed to that address AXIE_NOT_READY Number 1 The Axle startup delay time in No _STARTUP_DELAY milliseconds AXIE_POWER_MAN Boolean FALSE Enables overall support of AXle No AGEMENT power management functionality in the Shelf Manager must be turned on explicitly Release 3 2 0 39 July 31 2013 AXIE_SEQUENCIN G_ENUM_READY T IMEOUT Pigeon Point Shelf Manager User Guide Number DEFAULT DESCRIPTION Timeout for the AXle Enumeration Not Ready signal to get cleared This timeout value is measured in milliseconds The value 0 means to not wait and the value 1 means to wait indefinitely Any other value Specifies the number of milliseconds that the Shelf Manager waits for the Axle Enumeration Not Ready signal to get cleared If the timeout expires and the signal does not get cleared the Shelf Manager terminates the wait issues an error message and proceeds to the next state of power sequencing CAN BE OBTAINED FROM SHELF FRU INFO No AXIE_SEQUENCIN G_M4_TIMEOUT Number Timeout for all AXle modules to reach State M4 This timeout value is measured in milliseconds The value O means to not wait and the value 1 means to wait indefinitely Any other value specifie
168. ber of bad PEBs 0 UBI max allowed volumes 128 UBI wear leveling threshold 4096 UBI number of internal volumes 1 UBI number of user volumes 1 UBI available PEBs 0 UBI total number of reserved PEBs 484 UBI number of PEBs reserved for bad PEB handling 0 UBI max mean erase counter 12 10 UBIFS read only UBI device UBIFS mounted UBI device 0 volume 0 name boot UBIFS mounted read only UBIFS file system size 30807168 bytes 30085 KiB 29 MiB 471 LEBs UBIFS journal size 1569792 bytes 1533 KiB 1 MiB 24 LEBs UBIFS media format w4 r0 latest is w4 r0 UBIFS default compressor LZO UBIFS reserved for root 1522516 bytes 1486 KiB RUPG upgrade file is not present skipping coading file ulImage 0 to addr 0x42000000 with size 1604076 0x001879ec Done coading file rfs 0 to addr 0x46000000 with size 2283039 Ox0022d61f Done Booting kernel from Legacy Image at 42000000 Image Name Linux 2 6 34 8 shmm700 Created 2012 03 26 17 40 37 UTC Image Type ARM Linux Kernel Image uncompressed Data Size 1402284 Bytes 1 3 MB Load Address 40008000 Entry Point 40008000 Verifying Checksum OK Loading init Ramdisk from Legacy Image at 46000000 Image Name ShMM700 RFS 3 2 0 Created 2013 05 17 18 21 50 UTC Image Type ARM Linux RAMDisk Image gzip compressed Data Size 2282975 Bytes 2 2 MB Load Address 00000000 Entry Point 0000
169. bit value of 1b in the bit position corresponding to a cipher suite ID 0 to 14 indicates that the corresponding cipher suite can be used There must be at least one cipher suite enabled so a value 0 of this parameter is treated as if all cipher suites are allowed just as with the default value Release 3 2 0 38 July 31 2013 Pigeon Point Shelf Manager User Guide TYPE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO ALTERNATE_CONT Boolean TRUE Use alternate controller on the Shelf No PORR Manager with the address that is equal to the ShMM hardware address If this variable is set to TRUE the active Shelf Manager exposes two IPMB addresses 20h and a second address based on its hardware address the backup Shelf Manager exposes only the IP MB address based on its hardware address After a switchover the address 20h is exposed by the former backup Shelf Manager which now exposes two IPMB addresses If this variable is set to FALSE the Shelf Manager exposes only the logical address 20h this is allowed only for a non redundant Shelf Manager In the redundant configuration this variable must be set to TRUE ATCA_TESTER_CO Boolean FALSE This variable if set turns off event No MEATIBIULEY handling optimizations in the Shelf Manager so thatthe Shelf Manager behavior is compatible with the Polaris ATCA Tester in particular duplicate hot swap events are handled as separate events if they h
170. bled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 8 LUN 0 Sensor 20 IPMB LINK 8 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 9 LUN 0 Sensor 21 IPMB LINK 9 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 10 LUN 0 Sensor 22 IPMB LINK 10 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 11 LUN 0 Sensor 23 IPMB LINK 11 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure Release 3 2 0 144 July 31 2013 20 Li Bu IP IP 20 Li Bu IP IP 20 Li Bu IP IP 20 Li Bu IP IP nk 12 s Status B A State B B State LUN nk 13 LUN s Status B A State B B State nk 14 s Status B A State B B State LUN nk 15 s Status B A State B B State LUN 0x8 0x8 0x8 0x8 Pigeon Point Shelf Manager User Guide 0
171. c rce Mounted dev pts etc rce Mounted dev mtdblock0O to var etc rce Mounted dev mtdblock10 to var upgrade The next step in the re scriptis to call rupgrade_tool c to check whether a reliable upgrade is in progress The check determines that an attempted reliable upgrade failed The message restoring ADM1060 EEPROM to RAM refers to the ShMM system supervisory device an ADM1060 which supervises the ShMM boot up process and implements some of the hardware aspects of the reliable upgrade support This message indicates that key variables affecting the boot process are being reverted to their state before the reliable upgrade was attempted etc rc Checking the reliable upgrade watchdog timer rupgrade_tool Watchdog not active rupgrade_tool restoring ADM1060 EEPROM to RAM rupgrade_tool upgrade failed etc rce Rupgrade c Ret 255 etc re Mounted ram disk to var log etc re Started syslogd and klogd etc rce Mounted ram disk to var tmp etc rce Setting hostname shmm 193 etc rc Mounted dev mtdblockl to etc etc re Calling etc rc carrier3 Board Hardware Address OxFE etc netconfig etc hosts has valid 192 168 1 193 entry etc netconfig Updating etc profile sentry with IP settings etc netconfig Starting inetd etc rce carrier3 Starting up IPMBs etc rce carrier3 Updating etc profile sentry with specific settings etc rce carrier3 RC2 daemons not started by request
172. c re Calling etc rc carrier3 Board Hardware Address OXxFE etc netconfig etc hosts has valid 192 168 1 193 entry etc netconfig Updating etc profile sentry with IP settings etc netconfig ifconfig ethO 192 168 1 193 etc netconfig ifconfig ethl 192 168 0 193 etc netconfig route add default gw 192 168 1 253 etc netconfig Starting inetd etc rce carrier3 Starting up IPMBs etc rce carrier3 Updating etc profile sentry with specific settings etc rce carrier3 Starting snmpd etc rce carrier3 Starting httpd etc rce carrier3 Starting Shelf Manager lt I gt 02 48 08 463 171 Pigeon Point Shelf Manager ver 3 2 0 Built on May 17 2013 14 48 57 lt gt 02 48 08 469 171 Limits code 400000 5076f0 end_data 10062000 start_stack 7fff7e30 sp 7fff78a0 eip 2ab0d2e4 The Shelf Manager starts and finalizes the reliable upgrade calling rupgrade_tool f ethO link up ethl link up ethl going to full duplex shmm 193 login root BusyBox v0 60 5 2005 05 12 22 46 0000 Built in shell msh Release 3 2 0 213 July 31 2013 Pigeon Point Shelf Manager User Guide Finally the user checks the status of the reliable upgrade by calling rupgrade_tool w rupgrade_tool w Recent upgrade status 1 PLB is 5 1 EEPROM page saved 2 persistent flash is 0 3 provisional flash is 1 4 copying image s 4 invoking scripts step4v u u boot bin k sentry kernel r sentry rfs hook etc_
173. can yield a report like this Release 3 2 0 143 July 31 2013 Pigeon Point Shelf Manager User Guide Pigeon Point Shelf Manager Command Line Interpreter 20 Link 1 LUN 0 Sensor 10 IPMB LINK 1 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 2 LUN 0 Sensor 11 IPMB LINK 2 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 3 LUN 0 Sensor 15 IPMB LINK 3 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 4 LUN 0 Sensor 16 IPMB LINK 4 Bus Status 0x4 IPMB A Disabled IPMB B Enabled IPMB A State 0x07 Isolated Undiagnosed communication failure IPMB B State 0x08 LocalControl No failure 20 Link 5 LUN 0 Sensor 17 IPMB LINK 5 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 6 LUN 0 Sensor 18 IPMB LINK 6 Bus Status 0x8 IPMB A Enabled IPMB B Enabled IPMB A State 0x08 LocalControl No failure IPMB B State 0x08 LocalControl No failure 20 Link 7 LUN 0 Sensor 19 IPMB LINK 7 Bus Status 0x8 IPMB A Ena
174. cccceccceseeceeececeseeseneeceeneecsenessnaeseeseeceneeesaes 124 3 14 3 IntegralHPI Client Configuration 0 21 cccececeeececeeeeeenseceeneecenecessaneeensececneetsnneetaaes 124 3 15 CONFIGURING SYSTEM SERVICES ccccsceceeeeeceeeeeeeaeeeeeeeceeeeecaaeeeeaaeeeeeeeseaeeesaeeeeaaesennees 124 3 16 CONFIGURING THE SPEED OF I C BUSES SHMM 700 ONLY c sessecsesessesesseeteesseesesen 126 4 USING THE SHELF MANAGER ccccccsseeceseeeeeeeeeeneeeeseeeeneeeeseaeseseeeenseeeeeeesescaeseanenenseaes 128 4 1 IN THIS SECTION eironi taaier esvipes ieee av anttie nadie te eda eae ate eels 128 4 2 SHMM LOGIN aisre nnanet nasaia aaaea aa aaea aaea aaae aaaea Ea tage selec vviuea pret adak 128 4 3 STARTING THE SHELF MANAGER cccesccesessceesceceeececeeeeesnaeeneaceseeeeeseaeaesnaeeceaeeseeeaeneeeee 128 4 4 REDUNDANT OPERATION saitessrsechevesseicieverecoesstdaad Ave aaan aaea iaaah TASG 136 4 4 1 Initialization of the Redundant Shelf MAnaQers 2 ccccccccccecsectsseeceeseeteeeeetsaes 139 4 4 2 Redundancy and CPLD State S NSO0 ccccccseccesseceeneeceeesesseneseneeseeeeessnneetiaes 140 4 4 3 Reboot Reason Sensor u cccccccceceeceeenseceeseeceeneeceaessenaeceeeeeceeeeessaesseeeseeeeecsnneetiaes 142 4 5 OPERATION IN SHELVES WITH RADIAL IPMB O c cccccccececsessseceseeseeessteaeeeeeeeeeeees 142 4 5 1 Operation in SAMM 500 based Shelves with Radial IPMB 0 ccccscccceee 143 4 5 2 Operation in SAMM 1500 based Shelves with R
175. cceeeeeeeeeeeeeeeseeeeeteaeeeteaeeeeneeee 73 Table 10 Shelf FRU Information and Configuration Variable Settings ccceseceseeeeseeeeeees 96 Table 11 Network ServiCes c ccccceeeeseceeeceeeaeeeenaeeeeeeeceaeeeeaaesaaeeseaeeeceaeseeaaesgeeeeseeeeseaaeseeneeeeaes 125 Table 12 IPMB Topology RECOM 0 000 eeeeeeeeeeeeee scene ee a A 147 Table 13 Cooling State SCNSOMS ceccceeeececeeeeeeeeeeeeeeeceaeeeeeaesseneeseeeesaeeesaaeseeeeeenaeeeseaeeneaeeeeaes 151 Table 14 ShMM Tests Implemented in the Diagnostic Infrastructure ec ceeeeeeeeseeeeeneeteees 166 Table 15 Options for the IPMB Analyzer Trace Collection Daemon 0 cccccceeeteeeetteeeeneeteees 171 Table 16 IPMB Trace Collection Packet Field ccccecececeeeeeeeeeceeeeeeeeeeseaeeseaeeseeeesaeessaeeeeeees 172 Table 17 IPMB Message List Basic Information c c ccceceeeeeeeeeeeaeceeeeeeeeeeaeaeeeeeeeeesensnnaees 174 Table 18 IPMI Filter Primitives cceccceescceeeeeeeeeeeeeseeeeeeeseeeeeeaeeeseeaeeeeeeaeeeeeeneeaeeeeneeeeeeennnees 189 Table 19 Flash Partitioning for 16MB Flash SAMM 5006 cccceesseceeseneeeeeeneeessesaeeeeseaaes 197 Table 20 Flash Partitioning for 32MB Flash SAMM 500S cceeeceeceeeeeeeeeeeeeeeseeeeeseaeeeeneeeeaes 198 Table 21 Flash Partitioning for 64MB Flash SAMM 500S c ccceececeeeeeeeeeeeeeeeeeeeeeeeaeeeeaeeeeaes 199 Table 22 Flash Partitioning for 32MB Flash SAMM 1500S
176. ce in response to temperature exceptions the Shelf Manager can raise the fan levels or if that step is not sufficient even start powering down FRUs to reduce the heat load in the shelf 2 4 1 Pigeon Point Shelf Manager Features The Pigeon Point Shelf Manager features are listed below e Executes on the ShMM a compact mezzanine module installed on a suitable carrier board for the shelf e Conforms to the AdvancedTCA specification e Monitors activities within the shelf via the ATCA specified dual redundant Intelligent P latform Management Bus IP MB Release 3 2 0 15 J uly 31 2013 Pigeon Point Shelf Manager User Guide e Accepts and logs events posted by any intelligent FRU in the shelf reflecting exceptions in temperatures voltages etc posts alerts outside the shelf based on configurable IP MI Platform Event Filters e Supports hot swapping of Field Replaceable Units FRUs while maintaining full management visibility e Interfaces to standard Telco Alarm infrastructures via ShMM carrier implemented dry contact relays e Supports redundant Shelf Manager instances for high availability e Integrates a watchdog timer which resets the ShMM if not periodically strobed such resets automatically trigger a switchover to the backup ShMM if configured e Includes battery backed real time clock for time stamping events e Implements rich set of shelf external interfaces accessible over Ethernet including Remote Management
177. ce reference or first frame Frame Number 1 Frame Length 8 bytes Capture Length 8 bytes Frame is marked False Protocols in frame i2c ipmi Inter Integrated Circuit Data B F Intell us I2C 1 arget address lags 0x00000000 Header Target Address 0x00 0x41 No corresponding response Target LUN teei sa 0O 0001 00 Release 3 2 0 Get Sensor Reading 0x82 NetFN Target LUN Request Sensor 0x00 igent Platform Management Interface NetFn Sensor 179 Event Request 2008 20 58 03 238567000 0 000000000 seconds 0 000000000 seconds Event Request from 0x20 to 0x82 0 000000000 seconds 0x04 0x04 July 31 2013 Pigeon Point Shelf Manager User Guide Header checksum Ox6e correct Source Address 0x20 Source LUN 0x00 SeqNo 0x28 2 00 Source LUN 0x00 1010 00 Sequence Number 0x28 Command Get Sensor Reading 0x2d Data Sensor Number 2 Data checksum 0x11 correct 0000 82 10 6e 20 a0 2d 02 11 ape eee Frame 2 11 bytes on wire 11 bytes captured Arrival Time Jun 24 2008 20 58 03 318074000 Time delta from previous captured frame 0 079507000 seconds Time delta from previous displayed frame 0 079507000 seconds Time since reference or first frame 0 079507000 seconds Frame Number 2 Frame Length 11 bytes Capture Length 11 bytes Frame is marked False Protocols in frame i2c ipmi Inter Integrated Circ
178. cessible to the Shelf Manager via the master only 12C bus In that case the binary file should be copied onto the ShMM and then written into the SEEPROMSs The utility eepromw can be used for that purpose Release 3 2 0 97 July 31 2013 Pigeon Point Shelf Manager User Guide The exact location of SEEPROMs on the master I C bus is carrier specific but typically they are located at address 0xA4 on channels 1 and 2 of the 12C multiplexer and the multiplexer itself resides at address 0xEQ In that case the following commands issued on the ShMM can be used to download the file var nvdata shelf_fru_info to the SEEPROMs Itis assumed that the file was compiled by the FRU Information Compiler and downloaded onto the ShMM as in the previous example eepromw c 1 A4 var nvdata shelf_fru_info eepromw c 2 A4 var nvdata shelf_fru_info The general syntax for the eepromw utility is as follows eepromw b lt multiplexer gt c lt channel gt lt eeprom address gt lt file gt lt count gt where lt multiplexer gt is the address of the 12C multiplexer on the 12C bus default 0xE 0 lt channel1 gt is the channel on the multiplexer to use default 0 lt eeprom address gt s the address of the target SEEPROM lt file gt is the path to the file to write onto the target SEEPROM lt count gt is how many bytes to write The optional parameter lt count gt specifies how many bytes to write If this parameter is not Specifi
179. ck frequency 40 MHz A2F Fabric clock frequency 20 MHz A2F Fast delay calibration 1330 cycles per 100uS A2F eNVM 128 KB 00000000 00020000 A2F eSRAM 16 KB 20000000 20004000 A2F Extram start 200022D0 RUPG booting from image 0 candidate Here U Boot reports that reliable upgrade is in progress and the candidate image set is being used The candidate image set index is 0 and the confirmed image set is 1 In serial Out serial Err serial Net FECO 00 18 49 01 8f 26 FECI 00 18 49 01 8f 27 FECO FECL Hit any key to stop autoboot 0 RUPG reliable upgrade is in progress skipping Here U Boot reports that a reliable upgrade is in progress and therefore A2F060 upgrade stage is Skipped it doesn t matter if the A2F 060 upgrade file is present or not Creating 1 MTD partitions on spi0 0x0000001c0000 0x000002000000 mtd 5 Below U Boot attaches the UBI partition and mounts the UBIFS volume to load the new kernel and RFS images This stage is skipped and the messages listed below do not appear during regular boots because a UBIFS cache is used Release 3 2 0 236 July 31 2013 Pigeon Point Shelf Manager User Guide UBI attaching mtdl to ubi0 UBI MTD device 1 is write protected attach in read only mode UBI physical eraseblock size 65536 bytes 64 KiB UBI logical eraseblock size 65408 bytes UBI smallest flash I O unit 1 UBI VID head
180. cl sctecteancactcdesanccexecasteetessaceteaneeezen 9 1 1 SHELF MANAGER DOCUMENTATION cccceesecsceseeececseeaececeeseseseaaeeeceeececseseaaeseeeesesesessnaeess 9 1 1 1 Conventions Used in this Document 2 ccccccceceeeeeenseceeseeceeeteseaeeeeeeseeseeceeneetaaes 9 1 2 ADDITIONAL RESOURCE Smicridea a a E e E EEEE 10 2 INTRODUCTION aeiiae aeree aaaea cece aaa enced ee penny aE E aAA EE SARENA ERRERA 11 2 1 IN THIS SECTION E 11 2 2 INTELLIGENT PLATFORM MANAGEMENT AN ATCA OVERVIEW 0sccccceceeeceesenseeeeeeeees 11 2 3 PIGEON POINT BOARD MANAGEMENT REFERENCE HARDWARE AND FIRMWARE 14 2 4 PIGEON POINT SHELF MANAGER AND SHMDM cccccccecsesscsecesceeceessaeeeceseesesessaeeeeeeeens 15 2 4 1 Pigeon Point Shelf Manager Features cccccccceccesseceeneeceneeecsseetsneeeseeeessnneeees 15 2 4 2 Support for Dual Redundant Operation 2 cccccccccccecseeceeneeceeeeecseeeesnseesseeetsnnenees 16 2 4 3 System Manager Interface 22 ccccccccceeceessecseeeceneeeseaeeseneecseesessaneeeneeeeeneessnnseeas 18 2 4 4 Pigeon Point SAMM Shelf Management Mezzanines cccccceesceesseeeeteeteeeeeees 19 3 CONFIGURATION ooo ccc osu ocak cock cen alee ce ech ak ese cede aa Aa E Laaa aa aaa aaant 22 3 1 IN THIS SECTION steveccehenes cotta secthesl taaan leeds Avaccthesnaitlids neds anita eather 22 3 2 SETTING UP U BOOT sisirin ennnen eaa aat ane aaaea eaka aai na Sa aN aaaea Nia AE Naa Aa ANENE 22 3 2 1 U Boot Envir
181. col rupgrade_tool invoking scripts step4h etc_copy Stopping Shelf Manager Pigeon Point Shelf Manager Command Line Interpreter Terminating the Shelf Manager Erasing var and etc copying var nvdata Operation copy etc and var nvdata Copying completed rupgrade_tool image s copy OK rupgrade_tool watchdog started rupgrade_tool selected provisional flash rupgrade_tool reboot Restarting system The reliable upgrade procedure resets the ShMM here and starts U Boot from the provisional Flash x Resetting Integrated Peripherals U Boot 1 1 2 Apr 11 2005 15 16 25 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 8000044 DRAM 64 MB Flash 16 MB In serial Out serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 Release 3 2 0 217 J uly 31 2013 Pigeon Point Shelf Manager User Guide Booting image at bfb00000 Image Name MIPS Linux 2 4 26 Created 2005 04 11 10 35 08 UTC Image Type MIPS Linux Kernel Image gzip compressed Data Size 843129 Bytes 823 4 kB Load Address 80100000 Entry Point 802bc040 Verifying Checksum OK Uncompressing Kernel Image OK Loading Ramdisk Image at bfc40000 Image Name sentry RFS Ramdisk Image Created 2005 04 22 9 10 41 UTC Image Type MIPS Linux RAMDisk Image gzip compressed Data Size 2400736 Bytes 2 3 MB Load Address 00000000 Entry Point 00000000 Verifying
182. common configuration file This mechanism allows redefinition of common settings on a carrier specific basis Typically only a few critical configuration variables are defined in the carrier specific file For instance the appropriate value for the MIN_FAN_ LEVEL parameter may well be determined by the shelf architecture and the fan facilities that it implements This mechanism allows such shelf specific constraints to be enforced One result of this mechanism is that to change the effective value of a configuration parameter that is specified in the carrier specific configuration file the change must be made in that configuration file A change for such a variable in the common configuration file will not have any effect 3 3 2 Obtaining Configuration Variables from Shelf FRU Information Besides carrier specific redefinition of configuration variables there is another mechanism for retrieving and applying configuration variables from the Shelf FRU Information This mechanism allows redefining certain configuration variables for a specific instance or specific type of shelf Configuration variables are stored in the FRU Information as a sequence of OEM type multirecords similar to the multirecords defined in the xTCA specifications These multirecords have the Pigeon Point manufacturer ID 00040Ah and the manufacturer specific record type is 10 The concatenated contents of these multirecords represent the configuration variables in the same t
183. copy 4 copying u boot bin from tmp to dev mtdchar8 using cp protocol 4 copying sentry kernel from tmp to dev mtdchar7 using cp protocol 4 copying sentry rfs from tmp to dev mtdchar9 using cp protocol 4 invoking scripts step4h etc_copy 4 image s copy OK 5 watchdog started 6 selected provisional flash 7 reboot 9 WDT not fired upgrade in progress 1l provisional flash 1 updating EEPROM 12 EEPROM updated 13 upgrade WDT disabled 13 invoking scripts stepl3h 14 upgrade completed successfully 7 8 2 Example 2 This example shows a reliable upgrade of the RFS image only copying etc and var nvdata non volatile directories to provisional Flash The RFS image is taken from an FTP server atthe IP address 192 168 1 253 The path to the RFS image on the FTP server is tftpboot ru ppc sentry sentry1500 rfs The upgrade procedure is started from a telnet session Note Since only the RFS image is explicitly updated the U Boot and kernel images are automatically copied from the persistent Flash partition to the provisional Flash partition The local system must be able to access the FTP server over the network thatis its network adapter must be up and configured and a route must exist from the ShMM to the FTP server In the example below the ShMM is configured with the network address 192 168 1 174 which is in the same network with the FTP server telnet 192 168 1 17
184. coverable Threshold 0x50 Full Sensor Record Owner Id 0x10 Sensor Number 3 Sensor Initialization THRESHOLDS Lower Non Critical Threshold 0xb0 Lower Critical Threshold 0xc0 Lower Non Recoverable Threshold 0xd0 Upper Non Critical Threshold 0x40 Upper Critical Threshold 0x48 Upper Non Recoverable Threshold 0x50 Full Sensor Record Owner Id 0x12 Sensor Number 2 Sensor Initialization THRESHOLDS Lower Non Critical Threshold 0xb0 Lower Critical Threshold 0xc0 Lower Non Recoverable Threshold 0xd0 Upper Non Critical Threshold 0x40 Upper Critical Threshold 0x48 Upper Non Recoverable Threshold 0x50 Full Sensor Record Owner Id 0x12 Sensor Number 3 Sensor Initialization THRESHOLDS Lower Non Critical Threshold 0xb0 Lower Critical Threshold 0xc0 Lower Non Recoverable Threshold 0xd0 Upper Non Critical Threshold 0x40 Release 3 2 0 113 July 31 2013 Pigeon Point Shelf Manager User Guide Upper Critical Threshold 0x48 Upper Non Recoverable Threshold 0x50 3 9 Setting the Auxiliary Firmware Revision The Auxiliary Firmware Revision can be set when the Shelf Manager is started The Auxiliary Firmware Revision is reported by the Get Device Id command targeted to a physical ShMC at the hardware specified IPMB 0 address versus the logical Shelf Manager at IP MB 0 address 0x20 and is stored in a single Flash file var nvdata aux fw revision If the file var nvdata aux
185. customize the workings of the reliable upgrade process must become thoroughly familiar with the Monterey Linux User s Guide s coverage of this topic The reliable upgrade procedure implemented in Monterey Linux is neutral to the embedded application running on the ShMM The procedure provides a sufficient set of hooks allowing a Specific application running on the ShMM to ensure that custom actions are carried out at appropriate points of the reliable upgrade The remainder of this section focuses on provisions for reliable upgrade of the Pigeon Point Shelf Manager firmware that have been implemented using these hooks 7 3 Flash Partitioning Both the ShMM 500 and the ShMM 1500 provide a hardware mechanism that allows swapping of the lower and upper halves of the Flash in the system memory map under control of software running on the CPU This capability is implemented in support of the reliable upgrade procedure for Software images in Flash The reliable software upgrade procedure assumes that the Flash device contains two copies of the software located in the lower and upper halves of Flash All ShMMs are Shipped with this partitioning in which the Flash device is divided into two equal parts each dedicated to holding one copy of the ShMM software This section assumes that the ShMM is configured to support reliable upgrade including the two Separate Flash regions The tables below provide summaries of the Flash partitions maintained on
186. d HPM 1 the reliable upgrade utility performs the following actions e obtains the images to copy locally or via the network e copies the images to the candidate locations e copies non volatile data to the candidate locations or erases the corresponding candidate locations etc var depending on the presence of command line parameters copy config copy etc copy var copy all1 and erase all e resets the ShMM and instructs it to boot the candidate U Boot which in turn loads images from the candidate image set Because of the last stage an invocation of rupgrade in upgrade mode typically does not return and instead resets the ShMM If rupgrade in upgrade mode does return it indicates that the reliable upgrade procedure has failed and was terminated before proceeding to reset the ShMM in order to boot the candidate images Release 3 2 0 228 July 31 2013 Pigeon Point Shelf Manager User Guide Not all images need to be specified to rupgrade in upgrade mode or in the HPM 1 upgrade mode the HPM 1 image may not include all possible components If any components are not specified during the reliable upgrade the missing components are copied from current confirmed component set This allows reliable upgrades to be done when only certain firmware components e g the application package change between releases while other components are left intact in that case upgrade images can be made smaller and the upgrade i
187. d Line Interface to Place LAN Configuration Parameters in the Shelf FRU Information An extended version of the CLI command rudataw can be used to dynamically update Shelf FRU Information with LAN configuration information The corresponding configuration file optionally compressed should first be downloaded to the ShMM The corresponding command line has the following syntax clia frudataw 1 lt ipmb address gt lt fru id gt lt in file gt clia frudataw 1 c lt ipmb address gt lt fru id gt Release 3 2 0 161 July 31 2013 Pigeon Point Shelf Manager User Guide The parameters lt ipmb address gt and lt fru id gt specify the IPMB address and FRU device ID of the FRU Information to update For the Shelf FRU Information use IPMB address 20h and FRU device ID 254 The parameter lt in file gt indicates the LAN configuration parameters file which is possibly compressed to be stored in the target FRU Information Option c removes clears the LAN configuration data from the target FRU Information For example the following command can be used to update the Shelf FRU Information with a new version of the LAN configuration parameters file clia frudataw 1 20 254 subsidiary lan param gz 4 10 2 3 Configuring a Linux DHCP server for obtaining LAN Configuration Parameters The etc dhepd conf file on the DHCP server should have a record with a unique DHCP client identifier string for each channel of each FRU for whic
188. d in EEPROM Flash and reset or power cycle the ShMM To erase the EEPROM ona ShMM 500 you need to enter the following command from the U Boot prompt shmm500 eeprom write 80400000 0 1800 EEPROM 0x50 write addr 80400000 off 0000 count 6144 done shmm500 To erase the EEPROM on a ShMM 1500 you need to enter the following command from the U Boot prompt shmm1500 eeprom write 80400000 0 1000 EEPROM 0x50 write addr 80400000 off 0000 count 4096 done shmm1500 OnaShMM 700 the U Boot environment variables are stored in two redundant dedicated Flash partitions To erase the U Boot environment variables on a ShMM 700 you need to enter the following commands from the U Boot prompt shmm700 sf probe 2 0 65536 KiB N25Q512A at 2 0 is now current device shmm700 sf erase 100000 80000 Erasing Flash done shmm700 Then you need to reset the ShMM using the reset command as shown below with resulting console output for the ShMM 500 corresponding output for the ShMM 1500 and ShMM 700 is Slightly different shmm500 reset Release 3 2 0 193 July 31 2013 Pigeon Point Shelf Manager User Guide U Boot 1 1 2 Apr 27 2005 19 17 09 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 8000041 DRAM 64 MB Flash 16 MB xxx Warning bad CRC using default environment In serial Out serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 sh
189. d message or event The middle frame provides the protocol break down and data interpretation for a message selected in the message list provided in the upper frame In an IPMB trace analyzer usage this frame provides interpretation of the layered contents of the message Three protocol layers are supported e Layer 1 provides low level information about a message such as the frame size arrival time relative to the start of the trace and the frame protocol i2c ipmi for IPMI messages e Layer 2 includes physical IPMB 0 and I2C information such as the bus the message was captured on IPMB A or IPMB B and the destination 12C address of the frame e Layer 3 is available only for IPMI ATCA messages and provides message header information and information about the request response data Specifically the message is identified as either a request or a reply and the request reply header fields are interpreted The data fields are interpreted as defined by IP MI and PICMG specs for a particular request or response The bottom frame provides interpreted hexadecimal and textual representations of a message selected in the message list provided in the upper frame The GUI provides a number of other GUI elements dialogs menu windows etc that are used to exercise various features of the IP MI Analyzer as well as customize its behavior For instance Release 3 2 0 174 July 31 2013 Pigeon Point Shelf Manager User Guide the
190. d on Client Identifiers noton MAC addresses In the default implementation the Client Identifier is based on the Shelf Address string that is stored in the Shelf FRU Information plus three endpoint R equest Identifier Request ID bytes in the following format e Identifier byte 1 indicates the board physical slot address the identifier byte value is slot address 2Fh e Identifier byte 2 is the AMC site number on the board for boards this byte has value 0 e Identifier byte 3 provides a parameter set number within the specific slot For boards modules that require the configuration of more than one LAN channel the value of this byte is 0 for the first parameter set 1 for the second parameter set and so on For boards modules that require configuration of a single LAN channel this byte is 0 Parameters for different slots are retrieved in parallel The same sequence of retransmission timeouts 1 2 4 8 16 32 64 64 seconds applies if the DHCP server does not respond as in the case of retrieval of the Shelf Manager parameters The lease time is infinite The Shelf Manager maintains internally a record of assigned parameter sets After a specific parameter set is no longer needed e g when the board or the corresponding module is extracted the Shelf Manager removes the corresponding parameter set from its records and sends a DHCP Release request for the corresponding client ID to the DHCP server This is done to prevent exhaustion
191. date can be accessed via the serial console date Thu Jan 1 03 16 30 UTC 1970 In order to change the date you should type in the correct date using the date application The format for the date command is MMDDhhmm CC YY ss where MM Month DD Day hh Hour using 24 hour notation Release 3 2 0 114 J uly 31 2013 Pigeon Point Shelf Manager User Guide mm Minute SS Second YY Y ear 2 digit CC Century For example date 042916282006 Sat Apr 29 16 28 00 UTC 2006 To make the date persistent you need to store it using the hwclock application hwclock w In some cases you might get the error message mktime cannot convert RTC time to UNIX time This error can be ignored It is due to the original date being in an uninitialized state 3 10 1 Obtaining Date and Time from a Time Server Itis possible to obtain the system date and time from a time server during system startup and synchronize it periodically thereafter This facility is critical if the ShMM carrier does not have an RTC battery There are two network time protocols that can be used for that purpose NTP and RFC 868 rdate The specific protocol to be used is selected when configuring the ShMM To enable obtaining the network time via the NTP protocol it is necessary to define the U Boot variables time_proto time_server and optionally the additional variable timezone The variable time_proto determines the adjust time protocol if this v
192. dated that area is designated the persistent area and continues in use until a future upgrade cycle starts the process over The reliable upgrade hardware mechanisms ensure that no matter what is installed to the provisional Flash the ShMM always manages to boot from a software copy that is either fully functional or sufficiently sane to determine that that there has been a failure in the upgrade session and consequently take appropriate corrective actions to revert to the safe software copy in persistent Flash Ata higher level the reliable upgrade hardware mechanisms are assisted by a software protocol based on logging of the status of the upgrade session to a non volatile file in var upgrade status The software protocol ensures that the reliable upgrade does not finish until all the required actions including those defined by custom hook scripts that may be needed for a specific application have all completed successfully Release 3 2 0 196 July 31 2013 Pigeon Point Shelf Manager User Guide The Monterey Linux reliable upgrade procedure is described in full detail in Chapter 7 of the Monterey Linux User s Guide Au1550 Edition for the ShMM 500 or MPC83xx Edition for the ShMM 1500 Users of the reliable upgrade procedure should review that material as background That material explains the underlying mechanisms which provide substantial opportunities for customization to particular requirements Any users who wish to extend or
193. de threads fails to strobe the internal watchdog timer on timely basis the Shelf Manager does the following e issues a warning message to the console and or to the syslog file e collects information about current usage of locks in the Shelf Manager e writes this information to a Flash file on the ShMM var nvdata dumplog bin where it survives reboots and can be subsequently analyzed by the analysis tool dumplog e if one of the monitored threads fails to do a timely strobe of the internal watchdog timer three times in a row stops strobing the hardware watchdog timer which causes a reset of the ShMM and a switchover to the backup Shelf Manager In this scenario output similar to the following appears on the console lt W gt 17 25 49 368 218 ipmf_watchdog_thread_proc 239 facility has failed don t confirm the watchdog lt I gt 17 25 49 760 218 Write 132232 bytes to the var nvdata dumplog bin 0 seconds lock offset 131848 lt W gt 17 25 50 068 218 ipmf_watchdog_thread_proc 239 facility has failed don t confirm the watchdog lt W gt 17 25 50 378 218 ipmf_watchdog_thread_proc 239 facility has failed don t confirm the watchdog lt W gt 17 25 50 688 218 ipmf_watchdog_thread_proc 239 facility has failed don t confirm the watchdog lt W gt 17 25 50 998 218 ipmf_watchdog_thread_proc 239 facility has failed don t confirm the watchdog The direct deadlock detection mechanis
194. der broadcast Broadcast destination address only for Broadcast Get Device ID Request messages ipmi header target Destination device responder or requester slave address ipmi header netfn Network Function NetFn of the message ipmi header trg_lun Destination responder or requester Logical Unit Number LUN ipmi header cre Header checksum ipmi header source Source device requester or responder slave address ipmi header sequence Request sequence ipmi header src_lun Source requester or responder LUN ipmi header command Command to execute ipmi data completion Completion code only for response messages ipmi data cre Data checksum The filtering expression primitives above are applicable for both IPMB traffic and RMCP RMCP IP MI traffic over the network In the latter case several RMCP filtering expression primitives are also available but these are not widely used To specify a filter use the Filter text entry box in the upper left corner of the main window Release 3 2 0 189 J uly 31 2013 Pigeon Point Shelf Manager User Guide For instance to select the messages with the destination address 0x20 and the source address 0x82 type the string ipmi header target 0x20 amp amp ipmi header source 0x82 in the Filter text entry box Figure 16 Filtered Trace TCP 192 168 1 199 Capturing Wireshark ae Fog File Edit View Go Capture Analyze Statistics Help a a ea Me e 2431 E
195. df0 2ac34970 1004bd60 2ac36350 530 ado Release 3 2 0 OW OW OW OW OW OW OW OW OW OW OW OW OW OW ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 ner 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 waiting 0 155 July 31 2013 Pigeon Point Shelf Manager User Guide 10051070 owner 0 waiting 0 2ac34450 owner 0 waiting 0 2accle28 owner 0 waiting 0 2accle40 owner 0 waiting 0 10090890 owner 0 waiting 0 2ac344f0 owner 0 waiting 0 2acbdb98 owner 0 waiting 0 10091fa8 owner 0 waiting 0 10091c10 owner 0 waiting 0 100923e8 owner 0 waiting 0 10092840 owner 0 waiting 0 1008d7a4 owner 0 waiting 0 Jefffa80 owner 0 waiting 0 2ac34430 owner 0 waiting 0 2ac36310 owner 0 waiting 0 2ac33d98 owner 0 waiting 0 1008d81c owner 0 waiting 0 10094c08 owner 0 waiting 0 For backwards compatibility with older versions of the Shelf Manager deadlock detection is turned off if the value of the configuration variable DETECT_DEADLOCKS is FALSE By default this configuration variable has the value TRUE This variable controls both the watchdog based and the direct deadlock detection mechanisms 4 9 Master Only I2C Bus Fault Isolation On some carriers ShMM GPIOs can be used to reset an 12C multiplexer or 12C switch on the master only 1
196. ding hardware reference design which together implement various types of management controllers e Pigeon Point Shelf Manager software and ShMM mezzanine module which together with an appropriate ShMM carrier board implement an AdvancedTCA compliant Shelf Manager and Shelf Management Controller ShMC 2 3 Pigeon Point Board Management Reference Hardware and Firmware This hardware and firmware level includes the local management of full size 8U AdvancedTCA boards as well as other auxiliary FRUs such as fan trays or power entry modules Based on the interfaces specified by IP MI and extended by AdvancedTCA and AdvancedMC any compliant Shelf Manager can work with any compliant IPM Controller and the FRUs that it represents including AMCs This section focuses on controllers based on Pigeon Point technology as a concrete example The focus here is on controller solutions for AdvancedTCA and AdvancedMC Pigeon Point also provides solutions for MicroTCA management controllers The Pigeon Point BMR reference design can be implemented as part of any board or other FRU and executes the corresponding firmware thereby realizing a compliant IPM Controller The BMR firmware represents one or more FRUs via IP MB 0 to the Shelf Manager including e Providing inventory information identifying each such FRU including its manufacturer and other data e Describing and implementing a set of logical sensors such as for temperature state of IP MB
197. e available gt not available ALLOW lt INFO gt backend c 730 Burning lt INFO gt backend c 764 2110743 bytes was written from file tmp _tmp_app to dev 8 0 lt INFO gt backend c 808 Erasing user data on candidate bank lt INFO gt backend c 818 Copy etc lt INFO gt shmm700_hal c 1189 Write confirmed 1 gt 1 ALLOW lt INFO gt shmm700_hal c 1194 Write candidate 0 gt 0 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 0 gt 1 ALLOW Here the reliable upgrade procedure resets the ShMM This causes U Boot to start from the candidate Flash PowerPrep start initialize power Battery Voltage 1 50V No battery or bad battery detected Disabling battery voltage measurements Mar 21 201223 23 40 FRAC 0x92926152 Release 3 2 0 235 July 31 2013 Pigeon Point Shelf Manager User Guide memory type is DDR2 Wait for ddr ready 1 power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory access ddr2 0x40000000 finish simple test U Boot 2009 08 Mar 21 2012 23 22 30 Freescale i MX28 family CPUS 297 MHz BUS 99 MHz EMI 130 MHz GPMI 24 MHz T20 ready DRAM 128 MB SFGEN N250512A detected total size 64 MB A2F SPICOMM protocol v1 6 M3 firmware v0 9 FPGA design v0 44 0 0 A2F A2F firmware version 0 9 A2F Last reset cause HARD A2F Device type A2F060M3E FG256 A2F MSS clo
198. e for use if an A2F060 upgrade fails for any reason a2f manual rollback dat Boot Previous A2F060 upgrade image for manual rollback O etc User etc directory instance 0 mounted as etc if image set 0 is active l etc User etc directory instance 1 mounted as etc if image set 1 is active O var User ar directory instance 0 mounted as var if image set 0 is active Nar User var directory instance 1 mounted as var if image set 1 is active O var sentry shmm700 app User Application package image 0 1 var sentry shmm700 app User Application package image 1 Release 3 2 0 223 July 31 2013 Pigeon Point Shelf Manager User Guide 8 4 Reliable Upgrade Procedure Status File The special file var upgrade status that is located on the User Flash partition contains the status of the most recent reliable upgrade session that is the session that is currently in progress or most recently completed var upgrade status is a text file and consists of multiple lines each line describes the Status of a particular stage in the upgrade procedure The format of a record line is as follows lt stage gt lt status gt lt description gt where e lt stage gt is an integer in the range 1 to 14 where stage 14 is the final stage of a reliable upgrade session meaning that the procedure is complete e lt status gt is a human readable string that describes the status passed or failed of the corre
199. e 64 MB A2F SPICOMM protocol v1 6 M3 firmware v0 8 FPGA A2F A2F firmware version 0 8 A2F Last reset cause HARD A2F Device type A2F060M3E FG256 A2F MSS clock frequency 40 MHz A2F Fabric clock frequency 20 MHz A2F Fast delay calibration 1330 cycles per 100uS A2F eNVM 128 KB 00000000 00020000 A2F eSRAM 16 KB 20000000 20004000 A2F Extram start 200022D0 RUPG booting from image 0 confirmed Tres serial Out serial Err serial Net FECO 00 18 49 01 8f 26 FEC1 00 18 49 01 8f FECO FEC1 Feb 24 2012 RUPG Hit any key to stop autoboot upgrade file is not present to addr 0x42000000 with size xoading file 0x00155970 Loaded from UBIFS cache xoading file ulmage 1 Tris 1 Ox002acfd5 Loaded from UBIFS cache Release 3 2 0 PowerPrep start initialize power Battery Voltage No battery or bad battery detected Disabling battery voltage measurements Feb 24 201207 45 29 FRAC 0x92926152 memory type is DDR2 07 44 12 0 skipping 87 design v0 42 0 0 27 1399152 to addr 0x46000000 with size 2805717 July 31 2013 Pigeon Point Shelf Manager User Guide Booting kernel from Legacy Image at 42000000 Image Name Linux 2 6 34 8 shmm700 Image Type ARM Linux Kernel Image uncompressed Data Size 1399088 Bytes 1 3 MB Load Address 40008000 Entry Point 40008000 Verifying Checksum OK
200. e IP address and the least Significant non zero bit of the netmask e usb1 192 168 1 13 toggling the least significant bit of the IP address e Starting from Shelf Manager release 2 8 0 data transferred over the Software Redundancy Interface can be compressed A configuration variable REDUNDANCY_COMPRESSION_THRESHOLD specifies the threshold for message size messages that are bigger than this threshold are compressed by the gzip algorithm before sending and uncompressed after receiving Smaller messages are sent in uncompressed form The default value of the threshold is 1 which effectively turns compression off for compatibility with previous Shelf Manager releases Since previous releases of the Shelf Manager do not understand compressed data compression should be turned on only if both redundant Shelf Managers are release 2 8 0 or higher of the Shelf Manager 3 4 4 Using the ShMM 1500 Alternate Software Redundancy Interface For ShMM 1500s that are configured for ShMC Cross connect operation software redundancy communication takes place over an FPGA implemented serial interface with SLIP Serial Line IP Support configured on it thereby enabling TCP IP to be used by the Shelf Manager for this link just as itis on the ShMM 500 The corresponding Linux Serial interface is named sl0 To use this interface it is necessary to define the redundancy network adapter in the Shelf Manager configuration file etc shelfman conf as follows REDUND
201. e Shelf FRU Information is stored The simplest case is if the Shelf FRU Information is stored on a Flash file on the ShMM In that case the binary image file should be downloaded on the ShMM via FTP and copied to the location var nvdata shelf fru_info The following log represents an example of the above process It should be performed on an x86 Linux machine NOT on the ShMM itself mv shelf_fru bin shelf_fru_info ftp 192 168 1 230 Connected to 192 168 1 230 220 shmm 230 FTP server Version wu 2 6 2 1 Sun Dec 15 17 40 37 GMT 2002 ready Name 192 168 1 230 serjio ftp 331 Guest login ok send your complete e mail address as password Password 230 Guest login ok access restrictions apply Remote system type is UNIX Using binary mode to transfer files ftp gt cd var nvdata 250 CWD command successful ftp gt del shelf_fru_info 250 DELE command successful ftp gt put shelf_fru_info local shelf_fru_info remote shelf_fru_info 227 Entering Passive Mode 192 168 1 230 107 162 150 Opening BINARY mode data connection for shelf_fru_info 226 Transfer complete 129 bytes sent in 5 le 04 secs 2 3e 02 Kbytes sec ftp gt quit 221 You have transferred 129 bytes in 1 files 221 Total traffic for this session was 640 bytes in 1 transfers 221 Thank you for using the FTP service on shmm 230 221 Goodbye Another case is when the Shelf FRU Information is stored in SEEPROMs that are ac
202. e Shelf Manager including operation in a redundant active standby pair 4 1 In This Section This section contains the topics listed below J ust click on a topic to go to it ShMM Login Starting the Shelf Manager Redundant Operation Operation in Shelves with Radial IPMB 0 Automatic SEL Truncation Cooling State Sensors Deadlock Detection Master Only I2C Bus Fault Isolation Assignment of LAN Configuration Parameters to Boards and Modules HPI System Event Sensor ShMM Tests Available via the Diagnostic Infrastructure 4 2 ShMM Login Once the ShMM has been fully booted you are prompted to login You can log in as user root With the factory defaults no password is requested We highly recommend that you change passwords during the configuration of the ShMM The password can be reset to factory defaults if the password is forgotten Here is a typical log in session for ShMM 500 which would look similar on ShMM 1500 and ShMM 700 shmm500 login root Password ls bin dev etc lib mnt proc sbin tmp usr var 4 3 Starting the Shelf Manager The Shelf Manager software is implemented in the executable file she1 man in the directory bin During normal use of the Shelf Manager it is invoked automatically by startup scripts Typical users never need to invoke it manually as described here Nevertheless for the unusual circumstances in which manual invocation may be necessary the details are described here The syntax of
203. e amount of time to wait depends on the size of the upgrade images and the copy protocol used to pull the images to the ShMM as well as actions performed by the image validation script e Onthe ShMM the startup script ete re unconditionally makes a call to rupgrade_tool c Ifthe call returns a value of 1 indicating that there is no upgrade in progress or an error code value indicating that the upgrade session has failed the startup scripts proceed with the normal mode boot up sequence If however a value of 0 is returned indicating that there is an upgrade session in progress the startup scripts proceed with validation of the sanity of the newly installed software calling rupgrade_tool s inthe middle of operation to strobe the upgrade WDT in case the validation takes longer than the upgrade WDT timeout period and finally start the Shelf Manager to perform final validation The watchdog timer interval is set to 12 8 seconds so the processing time in the etc re script between the call to rupgrade_tool c and strobing the WDT and between Strobing the WDT and starting the Shelf Manager must not exceed 12 8 seconds each e During initialization the Shelf Manager strobes the upgrade WDT once again before trying to establishing a network connection with the peer Shelf Manager Establishing a network connection may take up to 6 seconds After that and after successfully finishing the initialization which indicates validity of the new configuration
204. e backup Shelf Manager The HP paradigm for FUMIs is defined in the HPI specification and consists of a series of function calls to select the upgrade image download the image to the target resource and activate the new image Also the FUMI API includes optional support for manual and automatic rollbacks to the previous state Shelf Manager FUMIs implement all functions defined for the logical bank model except for SaHpiF umiAutoR ollbackDisableS et the explicit bank model is not supported for these FUMIs An image used for HP based Shelf Manager upgrades must be in HPM 1 format This format allows creating a single image that consists of multiple components with a special header An HPM 1 Shelf Manager image includes as HPM 1 components the normal U Boot kernel and RFS images used for reliable upgrades On ShMM 500 1500 the HPM 1 image consists only of these three components On ShMM 700 the HPM 1 image also includes the application package and therefore consists of four components Starting with release 2 8 0 Shelf Manager images in HPM 1 format are available to ShMM based shelf developers from a secure partner page in addition to the traditional U Boot kernel and RFS images The implementation of these Shelf Manager FUMIs is done on top of the reliable upgrade facility Once the image is downloaded to a physical ShMM a call to activate the image invokes the reliable upgrade utility On ShMM 500 1500 the reliable u
205. e data in the unattended trace mode By default the file name is var ipmb_traced ipmb_traced log delay lt usec gt c Run daemon in the controlled trace mode controlled e Trace IPMB state change events Applies to local events trace mode only d lt usec gt Specify a delay in microseconds for polling the IPMB A IPMB B interfaces This option has an effect only if the e events option is enabled If there are IPMB state change events they are extracted and collected at intervals that are no longer than the defined delay but always whenever an IPMB message is received The delay value is Specified in a range of 0 to 1000000 microseconds 1 second If the d option is omitted the delay value defaults to 100000 microseconds 0 1 second Applies to local trace mode only p lt port gt port lt port gt In controlled trace mode specify the TCP port number to listen to the client application connects to this port Default port 19000 q lt size gt quota lt size gt In unattended trace mode the maximum trace file size in KB After this size is reached the existing file is closed and renamed with the extension old and a new trace file is created w wireshark Produce output in the Wireshark format otherwise ipmb_traced produces output in the Ethereal format On ShMM 700 only Wireshark output format is Supported so this option is ignored on ShMM 700
206. e following query dumps the values of net functions and command codes for IPMB messages sent from the Source address 0x84 to the Shelf Manager destination address 0x20 tshark r ipmb_traced log e ipmi header netfn e ipmi header command ipmi header target 0x84 amp amp ipmi header source 0x20 T fields 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 0x2d 0x04 0x2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x06 0x01 0x04 0x2d 0x04 0x20 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d 0x04 Ox2d Release 3 2 0 191 July 31 2013 Pigeon Point Shelf Manager User Guide 0x04 Ox2d Full information about t shark options and syntax can be found in the tshark manual page man tshark Release 3 2 0 192 July 31 2013 Pigeon Point Shelf Manager User Guide 6 Re initializing the ShMM 6 1 In This Section This section applies to SHMM 500 ShMM 1500 and ShMM 700 and contains the topics listed below J ust click on a topic to go to it e Re initializing the U Boot Environment e Re initializing the File System e Resetting the Login Password 6 2 Re initializing the U Boot Environment The U Boot environment variables are stored in the ShMM EEPROM on the ShMM 500 and ShMM 1500 and in two redundant dedicated Flash partitions on the ShMM 700 If you would like to restore the factory defaults for the U Boot environment variables you must first erase the environment variables store
207. e procedure If the utility detects a failure the reliable upgrade procedure is terminated by adding to var upgrade status a record indicating an unsuccessful completion of the upgrade procedure and exiting with an appropriate error code If the utility is called with the c capital C option it copies the images to Flash partitions according to other command line parameters but does not perform the final upgrade step of Switching active partitions and rebooting This final step can be done later using the command rupgrade_tool s R The utility prints any informational messages to stdout Providing a v flag to any option that Supports it increases the verbosity of the informational messages The utility prints any error messages to stderr The utility has the following synopsis rupgrade_tool s lt dst gt lt src gt proto lt protocol gt d hook lt args gt v a R rupgrade_tool c v rupgrade_tool f hook lt args gt v rupgrade_tool w f rupgrade_tool S v rupgrade_tool h rupgrade_tool C lt dst gt lt src gt proto lt protocol gt d v a where the parameters are defined as follows s lt dst gt lt sre gt proto lt protocol gt hook lt args gt v a R Release 3 2 0 203 July 31 2013 Pigeon Point Shelf Manager User Guide Initiate the reliable upgrade procedure As delivered with Shelf Manager support this step incl
208. e set to M3 prev M2 cause 1 locked 0 lt I gt 14 35 23 984 230 sdrrep_full_ipmc_sdrs_update_thread for 20 attempt 0 lt I gt 14 35 23 992 230 Registered potential fan 0x100eb2b8 SA 0x20 FRU 2 lt i gt 14735723 997 230 sdrrep finished reading SDRs for SA 20 added 15 preserved 0 deleted 0 time 7 seconds lt I gt 14 35 24 003 214 Hot Swap event SA O0xfc FRU 0 sensor_number 0x00 M2 gt M3 cause 1 lt I gt 14 35 24 021 231 sdrrep_full_ipmc_sdrs_update_thread for FC lt I gt 14 35 24 024 235 SA 0x20 FRU 2 is ACTIVATING lt I gt 14 35 24 031 217 Move fan 0x1l00eb2b8 cnt 2 0 gt 1 lt I gt 14 35 24 031 217 ipmc_cooling_scan_sensors Reread sensors on the SA OxFC lt I gt 14 35 24 023 231 sdrrep_full_ipmc_sdrs_update_thread for FC attempt 0 lt I gt 14 35 24 071 218 Adjust power for FC 00 from 0 to 20000 lt I gt 14 35 24 073 218 Sending sync point lt I gt 14 35 24 077 213 Controller FC FRU 0 ATCA state set to M4 prev M3 cause 0 locked 0 lt I gt 14 35 24 081 214 Hot Swap event SA 0xfc FRU 0 sensor_number 0x00 M3 gt M4 cause 0 lt I gt 14 35 24 097 236 Set Port State disable ipmc FC chan 1 it 0 ports 1 lt 1 ext 0 group 0 lt I gt 14 35 24 106 222 IPMC 20 1 is in the list already Exiting lt I gt 14 35 24 109 234 Tasklet ACTIVATE 20 1 is_local_address 1 lt W gt 14 35 24 111 234 Local FRU 1 at SA 0x20 is not listed in Shelf FRU Info still activate lt I gt
209. e with the IPMI 2 0 specification e The reading returned by that sensor is equal to the percentage of the SEL that is occupied 0 to 100 e The sensor assumes the state SEL Almost Full 5h when the percentage of free entries in the SEL falls below the high watermark the value of the configuration parameter SEL_HIGH_WATERMARK An event is generated in that case and placed in the SEL Event Data Byte 3 of the event contains the percentage of the SEL that is occupied No event if generated for this state until the occupancy of the SEL falls below the low watermark this is done to prevent multiple events if SEL occupancy oscillates around the high watermark e A custom state 6h is defined for the state when the SEL occupancy is below the low watermark the percentage of occupied entries falls below the value of the configuration parameter SEL_LOW_WATERMARK There is no event associated with this state the System Manager can detect this state by polling the sensor value and its asserted states Using this information the System Manager performs truncation of the SEL presumably using the IPMI command Delete SEL Entry sending it over the RMCP session 4 7 Cooling State Sensors The cooling management strategy in the Shelf Manager is configurable and also depends on the currently chosen carrier Cooling management strategy configuration is described in section 3 7 Irrespective of the current cooling management strategy however severa
210. eate the IP MI traffic on IPMB 0 to be analyzed For example if insertion of a board generates the sequence of events and commands that need to be analyzed do such a board insertion 3 The GUI will update the main window showing the collected trace in real time Release 3 2 0 182 July 31 2013 Pigeon Point Shelf Manager User Guide Figure 11 Collected Trace Controlled mode File Edit View Go Capture Analyze Statistics Help a E amp menrvts Bl Filter Expression Clear Apply Q a No Time Source 237 6 761618 T2c 1 238 6 762032 I2C 2 239 6 762581 I2C 2 240 6 762993 I2C 1 241 6 763541 I2C 1 242 6 763952 I2C 2 243 6 764502 I2C 2 244 6 764913 I2C 1 245 6 765460 I2C 1 246 6 765873 I2C 2 247 6 766420 I2C 2 248 6 766834 I2C 1 249 6 767380 12c 1 250 6 767725 I2C 2 251 6 768293 I2C 2 252 6 768703 I2C 1 253 6 769264 I2C 1 254 6 769684 I2C 2 255 6 770229 I2C 2 TCP 192 168 1 65 Capturing Wireshark Destination 0x76 0x76 0x77 0x77 0x78 0x78 0x79 0x79 0x7a 0x7a 0x7b 0x7b Ox7c Ox7c Ox7d Ox7d Ox7e Ox7e Ox7f b Frame 1 7 bytes on wire 7 bytes captured b Inter Integrated Circuit Data b Intelligent Platform Management Interface 0000 12 18 d6 20 24 01 bb Protocol Info J IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Re
211. eboot the ShMM A user who has access to the ShMM serial console can see the messages produced there by U Boot firmware and the Linux kernel and therefore can easily monitor the status of the reboot However in the case of a remote connection to the ShMM the status of the reboot can be more difficult to determine For example a telnet or ssh based virtual terminal connection usually times out on the reboot of the ShMM The user can either assume that the upgrade was successful or wait for some time reconnect to the ShMM and remotely invoke rupgrade status in order to find out the current upgrade session status The amount of time to wait is determined by the Release 3 2 0 232 July 31 2013 Pigeon Point Shelf Manager User Guide initialization time of the ShMM the time needed to boot the Linux kernel and start the proper network daemons and by the time needed to validate the functionality of the newly installed firmware image set e Onthe ShMM the startup script etc re unconditionally makes a call to rupgrade status The returned value indicates whether there is an upgrade session in progress a value 0 is returned no upgrade session is in progress a positive value is returned or the current upgrade session has failed a negative error code is returned If there is NO upgrade session in progress the returned value is non zero the startup script continues with the normal mode boot up sequence If however there is an upgrade sess
212. ection 9 covers HP based firmware upgrades for all three ShMM types Renumbered sections 8 Appendix C updated section numbers reflect new sections 8 and 9 New Appendix A describes how to convert a shelf from ShMM 500 based to ShMM 700 based Shelf Managers Appendix B and Appendix C retitled content from previous sections 10 and 11 C 25 Release 3 1 1 Section 5 adds coverage of HPM 2 IPMI Trace Payloads in board trace mode Section 5 removes the limitation for ShMM 700 of not supporting controlled mode Section 8 6 adds a note about encryption of A2F060 upgrade images C 26 Release 3 2 0 Section 3 3 describes new configuration variables ALLOWED_CIPHER_SUITES AXIE_SEQUENCING_ENUM_READY_ TIMEOUT AXIE_SEQUENCING_M4 TIMEOUT AXIE_SEQUENCING WAIT_DELAY AXIE_SYSTEM_MODULE_IPMB_ ADDRESS AXIE_NOT_READY_STARTUP_DELAY TURBO_MODE_MIN MISSING_FAN_TRAYS WATCHDOG_TIMEOUT New section 3 16 describes how to configure the maximum speed of I2C buses on the ShMM 700 Section 4 5 3 introduces the ripmb_prog utility for programming of ShMM 700 routing element FPGA s Section 4 9 adds a description of the HPDL based method of C bus fault isolation Appendix A notes that ShMM 500 to ShMM 700 switchovers involving a ShMM 700 module mounted on a ShMM 500 carrier via the Pigeon Point ShMM 700R SO DIMM adapter board SHMM 700R SA are not supported Release 3 2 0 262 July 31 2013
213. ed all the contents of the file are written into the target SEEPROM The total number of bytes written to SEEPROM does not exceed the size of the SEEPROM The reciprocal eepromr utility allows the user to read the contents of an SEEPROM into a file on the Flash and has the following parameters eepromr b lt multiplexer gt c lt channel gt lt eeprom address gt lt file gt lt count gt where lt file gt s the path to the file where data are written from SEEPROM lt count gt is how many bytes to read The remaining parameters are the same as for eepromw There is one caveat on using the utility eepromw The Shelf Manager must not be running when the SEEPROM is updated because the Shelf Manager can change the 12C multiplexer settings trying to access an 12C device behind the multiplexer while the update is in progress This will cause the update to fail with high probability There are several ways to ensure that the Shelf Manager is not running on the ShMM One is to turn off automatic start of the Shelf Manager by setting the U Boot variable Release 3 2 0 98 July 31 2013 Pigeon Point Shelf Manager User Guide start_rc2_daemons to n The other way is to terminate the Shelf Manager using the command clia terminate On some shelves access to Shelf FRU SEEPROMs is also guarded by the ACTIVE signal on the ShMM In that case make sure before running eepromw that the Hardware Redundancy Interface flags Local Healt
214. ed that copies the U Boot and kernel images from the persistent Flash to the provisional Flash After that the upgrade utility proceeds to copy the RFS image to its designated position in provisional Flash rupgrade_tool invoking proto ftp 192 168 1 253 etc upgrade step4vshm etc upgrade step4vshm dev mtdchar7 Done etc upgrade step4vshm etc upgrade step4vshm dev mtdchar8s Done rupgrade_tool copying using cp protocol rupgrade_tool invoking etc upgrade step4hshm etc upgrade step4hshm etc upgrade step4hshm etc upgrade step4hshm etc upgrade step4hshm etc upgrade step4hshm rupgrade_tool image s rupgrade_tool watchdog Release 3 2 0 scripts step4v r sentry shmm1500 rfs tftpboot ru ppc admin assword hook etc_copy Erasing dev mtdchar7 Done Copying Kernel from dev mtdchar2 to Erasing dev mtdchar8 Done Copying U Boot from dev mtdchar3 to sentry shmm1500 rfs from tmp to dev mtdchar9 scripts step4h etc_copy Stopping Shelf Manager Done Cleaning new var partition Done Cleaning new etc partition Done Copying etc contents Done Copying var nvdata contents Done Upgrade complete copy OK started 215 J uly 31 2013 Pigeon Point Shelf Manager User Guide rupgrade_tool selected provisional flash rupgrade_tool reboot Restarting system The step4hshm hook script is invoked which stops the Shelf Manager and preserves the non vol
215. ed by Pigeon Point to describe the structure and device population of specific AdvancedTCA shelves ShMM carriers and non intelligent subsidiary FRUs managed by the Shelf Manager such as Fan trays PEMs alarm panels This language is called HPDL which is short for Pigeon P oint Hardware P latform Description Language A full description of the HPDL is available to shelf developers in a separate document the Pigeon Point HPDL Reference HPDL descriptions are organized in units called HPDL modules Each module corresponds to a single text file One module can describe one or more carriers and or shelves The HPDL modules are compiled by a special compiler hpd1c and stored as binary data The encoding of the binary data conforms to the ASN 1 Basic Encoding standard ISO 8825 The Shelf Manager reads and interprets this data during initialization and obtains the following information from it e Population of the master only 12C bus on the carrier and shelf e Types and attributes of non intelligent FRUs managed by the Shelf Manager e Information about IPMI sensors exposed by the Shelf Manager e Associations between IPMI sensors and physical signals exposed by master only 2C devices Based on this information the Shelf Manager configures master only I2C devices creates FRU descriptors for non intelligent FRUs creates IPMI sensors and associates them with the corresponding FRUs In addition to HP DL modules collections of SDRs are u
216. ed in bootargs generation and by default configures ethaddr to be passed as a configuration parameter Default value is set bootargs bootargs fec_mac ethaddr If a different MAC address is specified and itis a valid MAC address the FECO controller is configured with it while FEC1 controller is configured with an incremented value If no value is specified i e mac_override is empty or doesn t set fec_mac value the Linux kernel driver selects MAC addresses by its built in algorithm The MAC addresses generated by the kernel in this latter case may be the same as what the U Boot reports via ethaddr and ethladdr variables but this is not guaranteed mount_ubifs ShMM 700 only U Boot command that invokes MTD configuration U Boot scripts This command is executed by U Boot during bootstrap Default value is run set_mtdparts mtdids ShMM 700 only This variable contains SPI interfaces configuration and is used privately by the U Boot to configure the SPI subsystem to access Flash drives Default value is spiO SPI O Release 3 2 0 28 July 31 2013 ENVIRONMENT VARIABLE mtdpartl Pigeon Point Shelf Manager User Guide DESCRIPTION ShMM 700 only This variable contains the Flash drive partition map The variable is used by the U Boot and its value is also passed to the Linux kernel as part of bootargs String The variable specifies how the Flash drive is split into partitions and consists of pairs of partition sizes
217. edChannelMask are modified accordingly The multiplexer switch is re enabled via the corresponding GPIO the next time some part of the Shelf Manager tries to access it This allows the Shelf Manager to access devices on the Subordinate buses that are still functional When an attempt is made to access the faulty subordinate bus by writing the channel number or mask for the faulty bus into the multiplexer switch this operation is not permitted for a certain configurable period of time after the failure During this period the faulty bus is not enabled an error is instead returned for the corresponding write operation This time period is specified by the configuration parameter LSOLATE_MUX_IGNORE_TIME The algorithm knows which value in the multiplexer switch register corresponds to the faulty bus because it monitors all write accesses to the multiplexer switch registers and remembers the last value before the occurrence of the fault When the ignore time expires an attempt is made to access the faulty bus and check whether the fault is still present If the attempt fails the ignore time is reset to its original value If the attempt succeeds then the recorded fault condition for that bus is removed and operation continues normally Release 3 2 0 158 July 31 2013 Pigeon Point Shelf Manager User Guide 4 10 Assignment of LAN Configuration Parameters to Boards and Modules In shelves where some of the boards and intelligent
218. edundant operation of IPMB 0 e Establishes two distinct and unconnected fault zones for the two Shelf Managers e Defined by the ATCA architecture only after the amendments made by ECN 002 for PICMG 3 0 R2 0 Figure 4 Dual Star Radial IPMB 0 Topology Shelf Manager Shelf Manager With the redundant dual star topology each ShMM carrier has two IPMB 0 hubs with each hub connecting one of the ShMM IPMB ports A or B to corresponding radial links on the backplane This topology has the following properties e Defined in PICMG 3 0 R1 0 and supported as the only IPMB 0 radial topology until the ECN 002 amendments to PICMG 3 0 R2 0 e Supports dual redundant operation of IP MB 0 even with the only one Shelf Manager installed e Does not need or include IPMB 0 cross links between the two ShMM carriers Release 3 2 0 146 J uly 31 2013 Pigeon Point Shelf Manager User Guide Figure 5 Redundant Dual Star Radial IPMB 0 Topology Shelf Manager Shelf Manager The IPMB topology record has the following format which follows the regular FRU Info multirecord format Table 12 IPMB Topology Record OFFSET LENGTH DEFINITION 0 1 Record Type ID The value of COh OEM is used 1 1 End of List Version 7 End of List Set to one for the last record 6 4 Reserved write as Oh 3 0 Record format version 2h for this definition 2 1 Record length 3 1 Record Checksum Holds the zero checksum of the record 4
219. ee frames Figure 6 Main Window IPMB trace analysis File Edit View Go Capture Analyze Statistics Help Qi a S meoarvFtt EE Q2QQFR BUS g Filter Expression Clear Apply No Time Source Destination Protocol Info s os 4 a ae VAUS IME ATC Wey UEC vevite 10 sey vaza 256 4 359893 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x2a 257 19 369289 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x00 258 19 369289 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x00 259 19 398524 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x00 260 19 399152 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x00 261 19 399886 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x00 262 19 400538 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x00 263 34 408530 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x03 264 34 408530 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x03 265 34 457179 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x03 266 34 457680 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x03 267 34 458276 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x03 268 34 458709 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x03 269 49 438130 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x19 270 49 438130 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x19 271 49 439784 I2C 2 0x09 IPMI ATC Req Get Device ID seq 0x19 272 49 440248 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x19 273 49 440783 I2C 1 0x09 IPMI ATC Req Get Device ID seq 0x19 1 274 49 441199 12C 2 0x09 IPMI ATC Req Get Device ID seq
220. eeees 186 Figure 14 Message 244 is Selected ceccceceeeceeeeeeeeeeeceaeeeeeeeseeeeeceaeeesaaeseeaeeseeeeeseaeeeeaeeeeeeeee 188 Figure 15 IPMI Protocol Layers IPMB traCe cccceeeesseeeeeeneeeeeeeaeeeeeeaaeeeseeaaeeeseeaeeeeeeaeeeenenaes 189 Figure T6 Filtered Trac acs sei cae tes eaten kt ecaa sees ania shaastat saci anscaudiysanasdeansccuassbacaausassgrenen 190 Release 3 2 0 7 July 31 2013 Pigeon Point Shelf Manager User Guide Tables Table 1 Shelf Manager DOCUMENTATION cece cee eeeeee cece ee ceeeeeeaeeeeaeeseaeeeseaeeeeaaeseneeeseaeeesaeeteneeeeaes 9 Table 2 Conventions Used in this Document eeeceeeeeeeeeeeeeeeee tennessee eeeaaeeseaeeseeeeetaeeetaeeeeeeeeea 9 Table 3 SAMM 500R ShMM 1500R and ShMM 700R Features and Variants cceeeeee 19 Table 4 SAMM Models ccsceceeeeeceneeeeeeeeeneeceeeeeeaeeeeaaesneecaeeeceaeseeaaesgeneeseeeeeeaesteaaeeeeeeeseeeees 20 Table 5 U Boot Environment Variables 2 0 0 eececeeeeeeeeeeeeeeeeeeeeeeeeeeeseaeeeeeseeeeeeseeeeeeseneaeeeeeeeaeees 24 Table 6 Configuration Parameter Types and Descriptions c ccceeeceeeeeeeeeeeteeeeeteeeeeaeeeeneeees 34 Table 7 Shelf Manager Configuration Parameters cccccccecceeseeceeeeeceaeeeeaeeeeeeeceeeeseaeeeeaeeeenees 36 Table 8 U Boot Environment Variables and Descriptions eccceeeeeeeeeeeeeeeeeeeeeeeeeeteeeeeeeteenaees 71 Table 9 Verbosity Configuration Parameters and LeVelS ccc
221. eesceseseeeeaaeeeeeaesaneseaaeeeaaesaaeaesaeeeeaaesaaaeseeeeeaaesseaaeseeeaeeaes 13 Figure 2 Pigeon Point Shelf Manager Architecture and Conte xt ccccesceceeeeeeeeeeeeeeeeeeeeeaes 17 Figure 3 Implementation Options for SMC Cross COnnects eeceeeeeeeeeeree eee tnaeenaeeeeeeeees 78 Figure 4 Dual Star Radial IPMB 0 Topology ccccccceseeeeeeeeeeeeeeeseeeeeeaaeseeaeeseneeesaaeeesaeeeeeeeees 146 Figure 5 Redundant Dual Star Radial IPMB 0 Topology ccceeceeeeeeeeeeeeeeeeeeseeeeseaeeneaeeeeaes 147 Figure 6 Main Window IPMB trace analysis cceceeeeeeeeeeeeeeeeeeeeenaeeeeeeaaeeeseeaeeeeneaeeeeeenaes 173 Figure 7 Main Window IPMI over network trace analysis cccceeeeeseeeeeneeeeseeneeeeeenaeeeeeenaes 175 Figure 8 Open Capture File Window cccccccceceeeeeceeeeeceaeeeeeee scenes caeeeeaaeseeaeeseeeeeseaeeesaeeeeeeeee 178 Figure 9 GUI Main Window Unattended mode cceccceceeeeeeeeeeeeeeeeeaeeeeaaeeeeeeeseaeeesaeeeeaeeseaes 178 Figure 10 Capture Options Window cccccceseceeseeceeeeecaeeeeeeeseeeeesaeeesaaeseeaeeseeeeeseaeeeeaeeseeeeens 182 Figure 11 Collected Trace Controlled mode cccccceceeeceeeeeceeeeeceeeeeeaeeeeneeseeeeeseaeeesaeeeeeeeees 183 Figure 12 Capture Options Window analyzing networked IPMI traffic ccceeeeeeeeeeeees 185 Figure 13 Collected Trace networked IPMI traffic ceceeeeseeeeeeeeceeeeeeaeeeeeeeseeeeeseaeeesaaeeeee
222. efault cooling management strategy is provided It is not necessary to configure the cooling management strategy if the default is satisfactory for a given carrier shelf which is likely the case for many carriers and shelves The default cooling management has the following features zoned cooling is supported the Shelf Fan Geography Record from the Shelf FRU Information is used to define cooling zones in the normal cooling state that is when no thermal thresholds on any sensors are crossed the Shelf Manager attempts to minimize the fan level but at the same time prevent thermal alerts Itdoes that by adaptively choosing the lowest possible fan level that allows the shelf to avoid thermal alerts for each fan taking cooling zones into account In the minor alert cooling state non critical thermal thresholds are crossed for one or more sensors the Shelf Manager periodically increases the fan level for the fans that serve the cooling zone s where those thresholds have been crossed until the fan level reaches its maximum or the thermal condition goes away In the major alert cooling state critical thermal thresholds are crossed for one or more sensors the Shelf Manager sets the fan level to the maximum for the fans that serve the cooling zone s where those thresholds have been crossed In addition if the thermal condition is caused by a specific FRU and the FRU supports power levels lower than the current one the Shelf Manager reduces
223. eir normal state but before a completion record has been added to the status file This feature is not currently used during reliable upgrades of the Shelf Manager As the last step the utility updates var upgrade status with a record indicating a Successful completion of the upgrade procedure and exits with a value of 0 w f Print the current status of the most recent upgrade procedure Essentially this option dumps the content of the var upgrade status file to stdout rupgrade_tool w returns a value of 0 ifthe upgrade procedure has completed successfully 1 if the upgrade procedure has completed unsuccessfully an appropriate error code if there isno var upgrade status file to be found If the specifier is supplied rupgrade_tool w removes the var upgrade status file before exiting S v Strobe the upgrade WDT rupgrade_tool S is intended as a shell level interface to strobe the upgrade WDT for use by newly installed software that is validating its sanity rupgrade_tool S returns a value of 0 u This option was previously used to undo a successful upgrade session As of the 2 7 3 release this option has been removed and is no longer supported To undo an upgrade use the options sR hook flip 0r SR Release 3 2 0 209 J uly 31 2013 Pigeon Point Shelf Manager User Guide C lt dst gt lt sre gt proto lt protocol gt v d a Copy images to the Flash but do not execute the final up
224. either in the HPDL carrier definition or in an IPMB Topology record in the carrier FRU Information with IPMB Topology Type 2h or 3h e the Shelf FRU Information contains an IPMB radial mapping record If the above prerequisites are satisfied the logical Shelf Manager creates and hosts multiple IP MB link sensors each of which reports the state of the corresponding radial segment As in the case of ShMM 500 the CLI command getipmbstate Can be used to show the state of all segments or of a specific IPMB segment the CLI command setipmbstate Can be used to selectively enable or disable a specific IPMB segment 4 5 3 Operation in ShMM 700 based Shelves with Radial IPMB 0 On ShMM 700 based shelves radial IPMB 0 support is implemented very similar to the ShMM 1500 case As on the ShMM 1500 there is a separate logical IPMB interface for each of the two portions IPMB A and IPMB B of each radial IPMB segment and the supported topologies include dual star and redundant dual star Since all ShMM 700 based shelves are HPDL based IPMB topology in those systems is always Specified in HPDL and no IPMB Topology record in the Carrier FRU Information is needed or Supported The IPMB topology clause has the following syntax which is backwards compatible with ShMM 1500 but with the extension that the position of the IPMB cross link can be specified lt ipmb_topology_clause gt IPMB_ TOPOLOGY lt bus_type gt lt cross_link_num gt
225. elf Manager explicitly manages the fan level No COOLING_KEEP_P OWERED_OFF_FRU S_IN_M1 Boolean FALSE If set the FRUs that are powered off by the Shelf Manager due to a critical temperature stay in state M1 and are not automatically activated and powered on when the temperature condition goes away manual intervention is needed to reactivate each such FRU Yes COOLING_NO_POW ER_DOWN_IN_CRI TICAL_ALERT Boolean FALSE Do not power down FRUs that experience critical thermal alerts This mode is not compliant with the PICMG 3 0 Advanced TCA Specification and should be used with extreme caution Yes COOLING_MANAGE MENT String 64 un undefined If specified the name of the shared library that implements cooling management The actual name of the library is libcooling_ lt xxx gt so where lt xxx gt is the value of this configuration parameter This library is dynamically loaded by the Shelf Manager and must be located in var bin of 1ib No COOLING_POLL_T IMEOUT Number 30 The maximum time in seconds between successive invocations of the cooling monitoring and management thread Yes Release 3 2 0 43 July 31 2013 Pigeon Point Shelf Manager User Guide DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO CPLD_ACTIVE WO Boolean This flag when TRUE enables a eee special workaround to detect the loss of the Active si
226. elf Manager is present the active Shelf Manager initiates a reboot of the ShMM in this way initiating a switchover to the backup Shelf Manager This allows third party applications on the ShMM to participate in monitoring health of the whole ShMM and initiate a switchover if necessary Also this behavior allows the former active ShMM to reinitialize itself as a backup in the case of a forced switchover from the backup Shelf Manager However the behavior of a standalone Shelf Manager one without a backup in the case of the loss of either Local Healthy or Active bit is different In the case of the Local Healthy bit the behavior is configurable and depends on the value of the configuration variable EXIT_IF_HEALTHY LOST_IN_STANDALONE_MODE By default the value is FALSE and in that case the Shelf Manager restores the Local Healthy bit and continues the operation However if the value is TRUE the standalone Shelf Manager reinitializes itself by rebooting the ShMM On ShMM 500 when the Active bit changes to 0 in the absence of a backup Shelf Manager itis likely the result of a known hardware issue that exists on some ShMM carriers In that case the Shelf Manager attempts to restore the Active bit by requesting the active state again via setting the Switchover R equest bit in the CPLD If the recovery of the Active bit fails the standalone Shelf Manager reinitializes itself by rebooting the ShMM No check of the Active bit takes place if the conf
227. elp Jay SBExeS Meortt BS aaan BUSH g e o y pression Clear apply No Time Source Destination Protocol Info 27 12 711118 I12C 1 0x41 IPMI ATC Req Get Device SDR seq 0x09 29 12 730525 I2C 2 0x41 IPMI ATC Req Get Device SDR seq 0x0a 31 12 752292 I2C 1 0x41 W gt IPMI ATC Req Get Device SDR seq OxOb 33 12 789674 I2C 1 0x41 IPMI ATC Req Get Device ID seq Ox0f 34 12 789674 I2C 2 LESS IPMI ATC Req Get Device SDR Info seq 0x10 35 12 791296 I2C 1 0x41 IPMI ATC Req ATCA Set FRU Activation seq 0x11 37 12 862226 I2C 1 Oxl10 PMIE ATC Rsp Get Device SDR Info seq 0x10 38 12 870679 I2C 2 0x41 IPMI ATC Req Get Device SDR seq 0x12 39 12 895527 I2C 2 0x10 IPMI ATC Req Platform Event seq 0x09 ZT aa 1a AAMGON TIF a Avar TOMT SATS Dan Gat Nawien CND cnn Av12 sss b Frame 34 7 bytes on wire 7 bytes captured b Inter Integrated Circuit Data v Intelligent Platform Management Interface Response in 37 b Header Get Device SDR Info Request from 0x20 to 0x82 Data checksum 0x80 correct 0000 82 10 6e 20 40 20 80 ne File home serge ipmb_traced log 4590 Packets 132 Displayed 132 Mar Profile Default 4 Release 3 2 0 178 J uly 31 2013 Pigeon Point Shelf Manager User Guide In a command line context at step 5 instead of starting the GUI utility run the t shark utility option x specifies the trace data file name
228. epending on the hardware address of the ShMM The least significant bit of the IP address is set to the least significant bit of the hardware address In the example above the IP address is 192 168 1 240 for the ShMM with an even hardware address and is 192 168 1 241 for the ShMM with an odd hardware address This modification of the IP address can be turned off by removing the file etc readhwaddr 3 4 1 2 RMCP Address Propagation There is an optional feature of the Shelf Manager that allows the backup ShMM also to be exposed on the external network with an IP address that is different from the RMCP IP address only in the least significant bit The netmask and default gateway on the backup ShMM is the same as on the active ShMM For example if the RMCP IP address is 192 168 0 2 the backup ShMM has the corresponding IP address 192 168 0 3 with the same netmask and default gateway To enable this feature itis necessary to define the Shelf Manager configuration parameter PROPAGATE_RMCP_ADDRESS as TRUE in the Shelf Manager configuration file Release 3 2 0 74 July 31 2013 Pigeon Point Shelf Manager User Guide 3 4 2 Usage of the Second Ethernet Interface The second Ethernet interface can be dedicated for use as a private network between redundant ShMMs and used to synchronize state information between the active and backup ShMMs Unlike the RMCP Ethernet port this second Ethernet interface is always enabled on both the active and backup ShM
229. epository Response from 0x82 to 0x20 b Data Data checksum 0x76 correct 0000 RCM aey yee EEE Frame frame 10 bytes Packets 168 Displayed 84 Mark Profile Default 4 Note Since the broadcast destination address is always zero the ipmi header broadcast filter expression should be used by specifying the following filter String in the Filter entry of the main window ipmi header broadcast 0x00 5 4 6 Analyzing a Trace Using Command Line Tools GUI tools are much more powerful for analyzing a trace than command line tools so you should use GUI tools for trace analysis if possible When using command line tools you should first capture the data into a file in unattended mode and then analyze the captured data file using tshark In this usage style tshark can be run multiple times over the same data thus allowing some degree of interactivity in the analysis In controlled mode tshark is run only once over the captured data as a result all applicable options must be specified in the command line at Startup Another limitation of t shark compared to Wireshark is that t shark can only provide back references for the packets the links Response toand Response in that allow the user to navigate across packets quickly This limitation results from the fact that t shark is optimized 190 J uly 31 2013 Release 3 2 0 Pigeon Point Shelf Manager User Guide for speed and thus calls the packet dissector only once per packet T
230. er offset 64 aligned 64 UBI data offset 128 UBI attached mtdl to ubi0 UBI MTD device name mtd 5 UBI MTD device size 30 MiB UBI number of good PEBs 484 UBI number of bad PEBs 0 UBI max allowed volumes 128 UBI wear leveling threshold 4096 UBI number of internal volumes 1 UBI number of user volumes 1 UBI available PEBs 0 UBI total number of reserved PEBs 484 UBI number of PEBs reserved for bad PEB handling 0 UBI max mean erase counter 11 9 UBIFS read only UBI device UBIFS mounted UBI device 0 volume 0 name boot UBIFS mounted read only UBIFS file system size 30807168 bytes 30085 KiB 29 MiB 471 LEBs UBIFS journal size 1569792 bytes 1533 KiB 1 MiB 24 LEBs UBIFS media format w4 r0 latest is w4 r0 UBIFS default compressor LZO UBIFS reserved for root 1522516 bytes 1486 KiB coading file uImage 0 to addr 0x42000000 with size 1604076 0x001879ec Done ooading file rfs 0 to addr 0x46000000 with size 2283039 Ox0022d61f Done Booting kernel from Legacy Image at 42000000 Image Name Linux 2 6 34 8 shmm700 Created 2012 03 26 17 40 37 UTC Image Type ARM Linux Kernel Image uncompressed Data Size 1402284 Bytes 1 3 MB Load Address 40008000 Entry Point 40008000 Verifying Checksum OK Loading init Ramdisk from Legacy Image at 46000000 Image Name ShMM700 RFS 3 2 0 Created 2013 05 17 18 21 50 UT
231. erase requested via U Boot variable etc rce erasing etc etc rce erasing var etc rce Selecting carrier from the environment etc re etc rce Extracting shelfman rootfs patch from 1 sentry shmm700 app etc rce Extraction result 0 etc re Placed var tmp to ram disk etc re Setting hostname shmm 121 etc re Started syslogd and klogd etc re Strobing the reliable upgrade WDT etc rce etc appears to be empty restoring from factory etc 6 4 Resetting the Login Password The factory default login for the ShMM is a user ID of root without any password A shelf supplier can change the default password We highly encourage users to change the password when configuring the Shelf Manager In the event that the new password is forgotten the password can be reset to its factory default or shelf supplier default via the password_reset U Boot variable By setting this variable to y and then booting up the system the root password is removed shmmxx00 setenv password_reset y shmmxx00 saveenv shmmxx00 boot The following output is shown on the console during boot up etc rc hostname demo Release 3 2 0 195 July 31 2013 Pigeon Point Shelf Manager User Guide 7 Re programming the ShMM 500 1500 This section describes how to update the firmware on the ShMM 500 or the ShMM 1500 using the reliable upgrade facilities that are built into Monterey Linux On the ShMM 700 a similar facility exists for reli
232. es e g the conversion formula default threshold values or default hysteresis it is sufficient to have only SDRs One problem with using HPDL data in a specific FRU is how to correlate the sensor numbers in the description to the actual sensor numbers The sensor numbers are assigned within an IPM controller not within a specific FRU so the same sensors on different FRUs of the same type have different numbers and it is typically not known in advance in which slot a specific FRU is going to be placed This problem is solved in the following way the sensor numbers in the HPDL SDR data for a specific FRU have relative numbers and these numbers are correlated to the actual numbers in ascending order For example assume that the replacement SDRs describe sensors 1 2 3 and 4 and the sensor numbers for the corresponding FRU in the HPDL chassis definition are 124 208 209 and 210 In that case the replacement SDR for sensor 1 applies to sensor 124 the replacement SDR for sensor 2 applies to sensor 208 and so on 3 7 Configuring the Cooling Management Strategy Different shelf vendors have different requirements for the cooling management strategy of the Shelf Manager The PICMG 3 0 specification contains several requirements related to cooling management that must be satisfied by all vendors e g the specification requires the Shelf Manager to deactivate a FRU that is in a critical thermal condition However some vendors have their own ve
233. esolution after after v Enable transport name resolution 2 Press the Start button to begin collecting the trace 3 Setup the filter mcp to include only RMCP related traffic into the trace Otherwise all network traffic going through the specified adapter will be included into the trace To do this type rmep in the Filter window in the top left corner of the main screen and press the Apply button 4 Execute actions that create the IP MI traffic on the network to be analyzed For example run a system management application that uses RMCP RMCP to talk to the Shelf Manager 5 The GUI will update the main window showing the collected trace in real time Release 3 2 0 185 J uly 31 2013 Pigeon Point Shelf Manager User Guide Figure 13 Collected Trace networked IPMI traffic ethO Capturing Wireshark File Edit View Go Capture Analyze Statistics Help ao o Oe emMeovFt ES QQAQR VUES g ilter rmcp_ Expression Clear Apply Time Source Destination Protocol Info z o 506 116 932515 192 168 1 62 80 240 102 57 IPMI ATC Rsp Get Device ID seq 0x03 E E ee gt b Internet Protocol Src 192 168 1 62 192 168 1 62 Dst 80 240 102 57 80 240 102 57 b User Datagram Protocol Src Port asf rmcp 623 Dst Port 36048 36048 Remote Management Control Protocol Class IPMI Version 0x06 E Sequence Oxff b Tyne Normal RMCP Class TPMT 0000 0
234. esults string in the case of both success and failure is the list of extended IPMB state and error flags reported by the driver Some of the tests are not implemented on ShMM 700 because they don t make sense for that platform this is noted explicitly in the table below However to ensure compatible allocation of test identifiers across all platforms the corresponding tests are still described as present on the ShMM 700 when invoked such tests report failure with the textual result string This test is not implemented Table 14 ShMM Tests Implemented in the Diagnostic Infrastructure TEST NAME TEST PARAMETERS NAME TYPE TEST DESCRIPTION DESCRIPTION UART INT 0 uart INT32 UART ID to The UART RX and TX channels are Internal UART test internally looped back while the test is Loopback Acceptable values are 0 in progress which may result in data Test and 1 loss for external communication on Default value is 0 the affected UART For example if Not UART 0 is the system console implemented some console output may be lost or on ShMM 700 corrupted UARTEXT 1 uart INT32 UART ID to This test requires manual interaction External test with the ShMM carrier to loop back UART Acceptable values are 0 the UART RX and TX pins Loopback and 1 Test Default value is 0 Not implemented on ShMM 700 Release 3 2 0 166 July 31 2013 Pigeon Point Shelf Manager
235. etc and var directories and either erases all files in them or copies the current contents of these directories to the candidate locations This behavior is controlled by several command line parameters all of them mutually exclusive e copy config if specified the SSH keys etc ssh the Shelf Manager configuration files etc shelfman con and the subdirectory var nvdata are copied to the candidate directories etc and var from the current location any other files in the candidate etc and var directories are erased e copy etc if specified the entire etc directory and the subdirectory var nvdata are copied to the candidate directories etc and var from the current location any other files in the candidate var directory are erased e copy var if specified the SSH keys etc ssh and the entire var directory are copied to the candidate directories etc and var from the current location any other files in the candidate etc directory are erased e copy all if specified the entire etc directory and the entire var directory are copied to the candidate directories etc and var from the current location e erase all if specified the candidate directories var and etc are erased only the SSH keys etc ssh are copied from the current location e fnone of the options above is specified default mode then the SSH keys etc ssh and the subdirectory var nvdata are copied to the candidate directories etc and var
236. ev mtdblock2 image 1 75 1 Idev mtdchar1 etc The etc J FFS2 dev mtdblock1 filesystem 2 75 2 dev mtdchar0 Nar The var FFS2 dev mdtblock0 filesystem 4 75 16 dev mtdchar4 Not mounted The Linux root file dev mtdblock4 system rfs image 20 75 11 25 Idev mtdchar11 Not mounted The first half of the dev mtdblock11 app_jffs application specific FFS2 partition 32 0 25 dev mtdchar8 Not mounted The other U Boot firmware image 32 25 0 5 dev mtdcharl0 var upgrade The second half of dev mtdblock10 the var upgrade FFS2 file system 32 75 1 dev mtdchar7 Not mounted The other Linux dev mtdblock7 kernel image Release 3 2 0 201 July 31 2013 Pigeon Point Shelf Manager User Guide OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 33 75 1 dev mtdchar6 Not mounted The other etc dev mtdblock6 FFS2 file system 34 75 2 dev mtdchar5 Not mounted The other var dev mtdblock5 J FFS2 file system 36 75 16 dev mtdchar9 Not mounted The other Linux root dev mtdblock9 file system rfs image 52 75 11 25 dev mtdcharl1 Not mounted The second half of dev mtdblock11 the app_jffs application specific FFS2 partition 7 4 The var upgrade File System Monterey Linux mounts a 1 MByte partition as a J FFS2 file system at var upgrade This file system hosts the reliable upgrade procedure status file It is important to note th
237. ev mtdblock4 system rfs image 12 25 3 75 dev mtdcharl1 Not mounted The first half of the dev mtdblock11 app_jffs application specific J FFS2 partition 16 0 25 dev mtdchar8 Not mounted The other U Boot firmware image Release 3 2 0 200 July 31 2013 Pigeon Point Shelf Manager User Guide OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 16 25 0 5 dev mtdcharl0 var upgrade The second half of dev mtdblock10 the var upgrade J FFS2 file system 16 75 1 dev mtdchar7 Not mounted The other Linux dev mtdblock7 kernel image 17 75 1 dev mtdchar6 Not mounted The other etc dev mtdblock6 J FFS2 file system 18 75 1 75 dev mtdchar5 Not mounted The other var dev mtdblock5 J FFS2 file system 20 5 7 15 dev mtdchar9 Not mounted The other Linux root dev mtdblock9 file system rfs image 28 25 3 75 dev mtdcharl1 Not mounted The second half of dev mtdblock11 the app_jffs application specific FFS2 partition Table 23 Flash Partitioning for 64MB Flash ShMM 1500s OFFSETINFLASH SIZE IN Device NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP dev mtdchar3 Not mounted The U Boot firmware image 0 25 0 5 dev mtdcharl0 var upgrade The first half of the dev mtdblock10 var upgrade J FFS2 filesystem 0 75 1 dev mtdchar2 Not mounted The Linux kernel d
238. ext format as in configuration files but optionally compressed using gzip Some restrictions apply to the configuration variables in the Shelf FRU Information First not all variables can be redefined from the Shelf FRU Information Table 7 indicates for each variable Release 3 2 0 71 July 31 2013 Pigeon Point Shelf Manager User Guide whether it can be redefined from the Shelf FRU Information Second retrieval of configuration variables from the Shelf FRU Information must be explicitly enabled for the Shelf Manager via the configuration variable SHELF_MANAGER_CONFIGURATION_IN_SHELF_FRU_INFO By default this variable is set to FALSE This is done to allow the Shelf Manager to start in the case of a bad configuration stored in the Shelf FRU Information and for compatibility with previous releases of the Shelf Manager 3 3 2 1 Using the FRU Info Compiler to Place Configuration Variables in the Shelf FRU Information The FRU Information Compiler is a Pigeon Point tool that generates a FRU Information image from its textual description The image can then be stored in an appropriate EEPROM device using various techniques The FRU Information Compiler implements special syntax that can be used to incorporate a configuration file into a FRU Information image that is being created For example the following lines add a compressed configuration file shelfman conf acb3 gz to the current FRU Information image PPS Shelf Manager Configuration
239. f 0 indicating that there is an upgrade procedure session in progress Release 3 2 0 208 July 31 2013 Pigeon Point Shelf Manager User Guide If the upgrade WDT is not active but the content of the var upgrade status file indicates that the upgrade procedure is still in progress this means that the ShMM rebooted due to a power cycle at one of the steps of the upgrade procedure In this scenario the utility performs the Same corrective actions as for the situation when the upgrade WDT is active and has fired see above Finally if the upgrade WDT is not active and var upgrade status is either not present or indicates that the upgrade procedure has finished either successfully or unsuccessfully the utility exits with the return value of 1 indicating that there is no upgrade procedure in progress f hook args v Complete the upgrade procedure The invocation of rupgrade_tool f is done from inside the Shelf Manager after the Shelf Manager successfully completes its initialization If the Shelf Manager is not started automatically that invocation is done at the end of the ete re script As soon as invoked rupgrade_tool f strobes the upgrade WDT and proceeds with establishing the new persistent Flash and disabling the upgrade WDT After disabling the upgrade WDT the upgrade procedure can invoke a special script to take any actions necessary after the hardware resources associated with a reliable upgrade have been returned to th
240. f Manager User Guide DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO MICRO_TCA Boolean FALSE If TRUE the Shelf Manager operates No as a MicroTCA Shelf Manager the second RMCP channel is used for interaction with Carrier Managers MIN_FAN_LEVEL Number 1 The minimum fan level the cooling Yes management code does not reduce the fan level of any fan below this value when controlling the fan level automatically MIN_SHELF_FRUS Number 2 The minimum number of ShelfFRUs No in the shelf that the Shelf Manager must detect to start up successfully NORMAL_STABLE_ Number 3600 The time in seconds for which the Yes TIME Shelf Manager preserves the minimum fan level dynamically found in Normal mode that is the minimum fan level that does not cause thermal alerts After this time expires the cooling algorithm decreases the minimum fan level if possible to allow the shelf to decrease the fan level if the thermal load in it has also decreased NO_MO_M1_EVENT Boolean FALSE If TRUE the physical Shelf Manager No eee Sava does not send an M0 gt M1 hot swap event when the Shelf Manager application is restarted on the ShMM the event is sent only the first time after the ShMM is plugged into a Shelf The default value is FALSE for backward compatibility PEF_USE_KEYED_ Boolean FALSE If TRUE the Shelf Manager uses No PERS special methods of managing TELCO alarms raised via PEF as OEM actions Each event condition that
241. f bytes which the Shelf Manager should read to get the entire Carrier FRU Information If this argument is not specified the Shelf Manager reads the entire EEPROM For example the following definition CARRIER_OPTIONS CARRIER_FRU_LOCATION ADM1026 0 2E 2048 specifies that the Carrier FRU Information should be retrieved from the SEEPROM of the ADM1026 device that is located on bus 0 at address 2Eh 7 bit and that only the first 2048 bytes of this SEEPROM which has the capacity of 8192 bytes should be read 3 6 4 2 Using the FRU Info Compiler to Place HPDL and SDR data in FRU Information The FRU Information Compiler is a Pigeon P oint tool that generates a FRU Information image from its textual description The image can then be stored in an appropriate EEPROM device using various techniques The FRU Information Compiler implements special syntax that can be used to incorporate binary HPDL and SDR data into a FRU Information image that is being created For example the following instructions add binary files acb3 hpd1 bin gz and acb3 sdr bin gz which are compressed HP DL and SDR files respectively to the current FRU Information image PPS HPDL Data File acb3 hpdl bin gz PPS HPDL SDR Data File acb3 sdr bin gz Release 3 2 0 104 July 31 2013 Pigeon Point Shelf Manager User Guide 3 6 4 3 Using the Shelf Manager Command Line Interface An extended version of the CLI command rudataw can be used
242. file system 18 5 1 75 dev mtdchar5 Not mounted The other var dev mtdblock5 J FFS2 file system 20 25 7 75 dev mtdchar9 Not mounted The other Linux root dev mtdblock9 file system rfs image 28 0 25 dev mtdchar8 Not mounted The other U Boot firmware image 28 25 3 75 dev mtdcharl1 Not mounted The second half of dev mtdblock11 the app_jffs application specific FFS2 partition Table 21 Flash Partitioning for 64MB Flash ShMM 500s OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED As CONTENT IN MBYTES MBYTES ON STARTUP 0 0 5 dev mtdcharl0 var upgrade The second half of dev mtdblock10 the var upgrade J FFS2 file system 0 5 1 dev mtdchar7 Not mounted The other Linux kernel image 1 5 1 dev mtdchar6 Not mounted The other etc dev mtdblock6 J FFS2 file system 2 5 1 75 dev mtdchar5 Not mounted The other var dev mtdblock5 J FFS2 file system 4 25 15 75 dev mtdchar9 Not mounted The other Linux root file system rfs image 20 8 dev mtdcharl2 Not mounted The second half of dev mtdblock12 the app1_jffs application specific FFS2 partition 28 0 25 dev mtdchar8 Not mounted The other U Boot firmware image 28 25 3 75 dev mtdcharl1 Not mounted The second half of dev mtdblock11 the app_jffs application specific FFS2 partition 32 0 5 dev mtdcharl0 var upgrade The first half of the dev mtdblock10 var upgrade J FFS2 file system 32 5 1 dev mtdchar2 Not mounted The Linux ke
243. fo storage size 2048 data size 529 lt I gt 14 35 22 875 226 1 Equal Shelf FRUs have been detected lt I gt 14354224877 226 Calling shelf FRU notifications count 1 op 1 lt I gt 14 35 22 879 226 Calling shelf FRU notification Ox2abd05f8 lt I gt 14 35722 881 226 Calling shelf FRU notification O0x4eb18c lt I gt 14 35 22 883 226 Lan apply cfg parameters HW addr OxFC cross 0 use_second 0 adapter ethO adapter2 lt I gt 14 35 22 886 226 Lan apply cfg parameters IP C601A8C0O GW FDO1A8CO MASK OOFFFFFF lt W gt 14 35 22 888 226 Shelf FRU Info found failed to get IP Connection record from Shelf FRU err 61 lt I gt 14 35 22 891 226 LAN channel 1 address 192 168 1 198 623 lt I gt 14 35 22 893 226 LAN setting RMCP IP address to 192 168 1 198 lt I gt 14 35 22 914 212 Controller 20 FRU 0 ATCA state set to M2 prev M1 cause 2 locked 0 lt I gt 14 35 22 916 212 Controller FC FRU 0 ATCA state set to M2 prev M1 cause 2 locked 0 lt I gt 14 35 22 935 226 LAN setting RMCP subnet mask to 255 255 255 0 lt I gt 14 35 22 939 226 channel_param_start_channel_on_backup channel 1 propagate 0 lt I gt 14 35 22 941 226 DHCP Client is disabled lt I gt 14 35 22 943 226 Calling shelf FRU notification Ox2ab44fdc lt I gt 14 35 22 945 226 LAN Config Parameters data in the Shelf FRU 0 bytes lt I gt 14 35 22 948 226 Found Power Distribution record instance 0 size 49 lt I gt 14 35 22 950 226 Al
244. from the current location any other files in the candidate etc and var directories are erased The behavior defined by these parameters is compatible with the behavior of the reliable upgrade tool on the ShMM 500 and ShMM 1500 Release 3 2 0 230 July 31 2013 Pigeon Point Shelf Manager User Guide As the final stage of the upgrade the utility starts the reliable upgrade watchdog timer WDT and resets the ShMM using the FPGA capabilities for ShMM reset Refer to ShMM 700 Hardware Architecture Specification for background and details on the reliable upgrade WDT The reliable upgrade WDT is started with a 120 second timeout period This period should be sufficient for the ShMM software to reach after the reset and without strobing the reliable upgrade WDT the point in time where the ShMM software can detect whether a reliable upgrade is in progress by invoking rupgrade status and if so strobe the reliable upgrade WDT via backend strobe The reliable upgrade utility can invoke an optional hook script that enables custom actions required by an application After each stage of the upgrade procedure the hook script is invoked as a sub shell and given a single parameter the stage number Table 27 above describes all Stages of the reliable upgrade that are visible to the hook script The script returns 0 on success and a non zero value in the case of a failure in which case the upgrade procedure is terminated No hook script
245. g UBIFS cache lt INFO gt shmm700_hal c 1913 Clean on defice 1 1 Leaving file dev mtd2 untouched lt INFO gt backend c 730 Burning image lt INFO gt backend c 764 393216 bytes was written from file tmp _tmp_uboot to dev 1 1 lt INFO gt backend c 730 Burning image lt INFO gt backend c 764 1402348 bytes was written from file tmp _tmp_kernel to dev 2 1 lt INFO gt backend c 730 Burning image Release 3 2 0 239 July 31 2013 Pigeon Point Shelf Manager User Guide lt INFO gt backend c 764 2283039 bytes was written from file tmp _tmp_rfs to dev 4 1 lt INFO gt shmm700_hal c 1285 Write storage state available gt not available ALLOW lt INFO gt backend c 730 Burning image lt INFO gt backend c 764 2110743 bytes was written from file tmp _tmp_app to dev 8 1 lt INFO gt backend c 808 Erasing user data on candidate bank lt INFO gt backend c 821 Copy etc lt INFO gt shmm700_hal c 1189 Write confirmed 0 gt 0 ALLOW lt INFO gt shmm700_hal c 1194 Write candidate 1 gt 1 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 0 gt 1 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 0 gt 1 ALLOW lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW lt INFO gt shmm700_hal c 1265
246. geonpoint com products html Release 3 2 0 10 July 31 2013 Pigeon Point Shelf Manager User Guide 2 Introduction This section provides an overview of the Pigeon Point Shelf Manager and Shelf Management Mezzanine or ShMM currently the ShMM 500R ShMM 1500R and ShMM 700R products The Pigeon Point Shelf Manager is a shelf level management solution for AdvancedTCA ATCA products The Pigeon Point ShMM when coupled with a corresponding carrier board provides the necessary hardware to run the Shelf Manager within an ATCA shelf This document focuses on aspects of the Shelf Manager and ShMM that are common to any ShMM carrier used in an AdvancedTCA context Carrier specific and shelf specific details are documented by shelf providers The ShMM 500R complies with the Restriction of Hazardous Substances RoHS directive but is equivalent to its predecessor the ShMM 500 from a software point of view All references to the ShMM 500 in this document apply to the ShMM 500R unless otherwise noted Both the ShMM 1500R and ShMM 700R were designed to be RoHS compliant from the start and there are no ShMM 1500 or ShMM 700 products though references to both product names may be simplified as ShMM 1500 or ShMM 700 respectively The ShMM 1500R is an additional ShMM variant that is based on a PowerPC processor unlike the ShMM 500R which is based on a MIPS 32 processor The ShMM 70OR is the newest ShMM variant itis based on the Freescale i MX28
247. gnal in the CPLD Registers in the CPLD Complex Programmable Logic Device on the ShMM expose the state of the hardware redundancy signals to the software This loss may happen on some platforms when hot inserting a ShMM carrier The workaround can be turned off by setting this value to FALSE if this problem does not exist for a specific platform This parameter is not applicable to the ShMM 700 CTCA_FRU_RESET Number 500 CompactPCI shelves only The time No aie cial in milliseconds during which the Shelf Manager holds the BD_SEL line low in order to reset a CompactPCI board CTCA_HEALTHY_T Number 0 CompactPCI shelves only The time No es in seconds during which the Shelf Manager waits for the HEALTHY Signal to appear when powering ona CompactPCI board If the HEALTHY Signal does not appear within the Specified time the Shelf Manager deactivates the board 0 the default Stands for infinity CTCA_INITIAL_F Number 15 CompactPCI shelves only The initial No AN aaa fan speed in the range 0 15 that Shelf Manager applies to fan trays in CompactPCI shelves 0 corresponds to the slowest and 15 to the fastest possible speed Release 3 2 0 44 July 31 2013 DEFAULT_GATEWA Y_IP_ADDRESS Pigeon Point Shelf Manager User Guide TYPE IP address DEFAULT None DESCRIPTION The default IP address used for the gateway for shelf external RMCP based communication if the
248. gned to fit smoothly into the physical and management architecture of AdvancedTCA Figure 1 includes an AMC carrier with a Carrier IPMC and two installed AMC modules each with a Module Management Controller MMC The figure also shows an instance of an MMC installed on a Rear Transition Module RTM which is then defined as an intelligent RTM In both cases on carrier management communication between a Carrier IPMC and its MMCs occurs over IPMB L L for Local An overall System Manager typically external to the shelf can coordinate the activities of multiple shelves A System Manager typically communicates with each Shelf Manager over Ethernet The Pigeon Point Shelf Manager External Interface Reference document details the interfaces and protocols that are available for such interactions Those interfaces include optional support for the SA Forum s Hardware Platform Interface HPI Pigeon Point offers two implementations of HPI Release 3 2 0 13 J uly 31 2013 Pigeon Point Shelf Manager User Guide one implemented internally and called IntegralHP I and the other based on OpenHPI www openhpi com Both implementations are documented in the Pigeon Point HP User Guide The next two sections address the board and shelf levels of management highlighting the following Pigeon Point products and their capabilities as well as the relevant AdvancedTCA functionality e Pigeon Point Board Management Reference BMR firmware and correspon
249. grade step This usage scenario is complementary to the upgrade without copying usage scenario With the option c rupgrade_tool performs the same operations as described above for the option s and terminates after copying images into the Flash where the rupgrade_tool run with the options s R Starts its operation h Show help to stdout 7 7 Reliable Upgrade Utility Use Scenarios Itis intended that the reliable upgrade utility is used in the following sequence in order to carry out an upgrade of the ShMM e The user makes a call to rupgrade_tool s to initiate the upgrade procedure The call can be made either locally from the ShMM serial console or remotely over the network via telnet rsh ssh or any equivalent e The user waits for the rupgrade_tool s to reboot the ShMM If the user is connected to the serial console locally the status of the reboot is obvious from the messages printed by the U Boot firmware and Linux to the serial console If the connection to the ShMM is remote the status of the reboot is less obvious For instance a telnet connection times out on the reboot of the ShMM The user can either assume that the upgrade procedure has been carried out successfully or wait for a certain amount of time required for the upgrade session to complete and then make a call to rupgrade_tool w again remotely over any of the remote shell tools mentioned above in order to find out the status of the upgrade session Th
250. grity of the physical network link between the Shelf Manager and the System Manager the RMCP link see the description of the configuration parameter SWITCHOVER_TIMEOUT_ON_B ROKEN_LINK A non zero delay can be used to accommodate slow network links that need significant time to initialize after shelf power up Release 3 2 0 53 J uly 31 2013 INNER_SEQUENCE _NUMBER_IN_SEN D_MSG_RESPONSE Pigeon Point Shelf Manager User Guide TYPE Boolean DEFAULT TRUE DESCRIPTION This variable controls which sequence number is used in the response to a Send Message command bridged from LAN to IPMB If TRUE the Sequence number of the command encapsulated in the Send Message requestis used If FALSE the Sequence number of the Send Message request itself is used According to a clarification being proposed for the latest version of the PICMG 3 0 specification the first variant is correct while the Shelf Manager historically used the second variant CAN BE OBTAINED FROM SHELF FRU INFO No INTEGRALHPI_DE Number 0 This variable defines the seconds Yes FAULT AUTO EXT portion of the default auto extraction RACT_TIMEOUT f timeout value for all exposed HPI resources A negative value causes the timeout to be set to SAHPI_TIMEOUT_BLOCK INTEGRALHPI_DE Number 1 This variable defines the milliseconds Yes pg kata er portion of the default auto extraction a
251. h the LAN configuration parameters need to be assigned Unless otherwise documented explicitly for certain specific types of ATCA shelves the format of the client identifier is as follows lt dhcp client identifier gt lt shelfaddress in bytes gt lt 2F board physical number gt lt FRU ID gt lt channel number gt The example below is for a shelf with 2 board slots with IPMB addresses 82h 84h and the shelf address XXX in ASCII HEHHEHHEEHEEHEE Board LAN Configuration Params t HHH HH F 0x82 0 0 host boardl_fruO_cho option dhcp client identifier 58 58 58 30 0 0 fixed address 192 168 1 202 option routers 192 168 1 253 option subnet mask 255 255 255 0 host boardl_fru0_chl option dhcp client identifier 58 58 58 30 0 1 fixed address 192 168 1 202 option routers 192 168 1 253 option subnet mask 255 255 255 0 host boardl_fru0_ch2 option dhcp client identifier 58 58 58 30 0 2 fixed address 192 168 1 202 option routers 192 168 1 253 option subnet mask 255 255 255 0 Release 3 2 0 162 July 31 2013 Pigeon Point Shelf Manager User Guide host boardl_fru0_ch3 option dhcp client identifier 58 58 58 30 0 3 fixed address 192 168 1 202 option routers 192 168 1 253 option subnet mask 255 255 255 0 host boardl_fru0_ch4 option dhcp client identifier 58 58 58 30 0 4 fixed address 192 168 1 202 option routers 192 168 1 2
252. handle partial upgrade sbin backend application tmp _shmm700_sentry shmm700 app copy etc lt WARN gt shmm700_hal c 322 Device ubil boot is mounted to folder boot lt INFO gt shmm700_hal c 1255 Write storage state not available gt available ALLOW lt INFO gt shmm700_hal c 368 Resetting UBIFS cache Erasing 64 Kibyte 30000 75 complete lt INFO gt shmm700_hal c 1863 Clone lt INFO gt shmm700_hal c 1898 File dev mtd0 of device 1 0 cloned to file dev mtd2 of device 1 1 lt INFO gt shmm700_hal c 1642 Cleaning image var upgrade boot uImage 1 m700_hal c 1863 lt INFO gt shmm700_hal c 1898 File var upgrade boot uImage 0 of device 2 0 cloned to file var upgrade boot uImage 1 of device 2 1 lt INFO gt shmm700_hal c 1642 Cleaning image var upgrade boot rfs 1 lt INFO gt shmm700_hal c 1863 Release 3 2 0 241 July 31 2013 Pigeon Point Shelf Manager User Guide lt INFO gt shmm700_hal c 1898 File var upgrade boot rfs 0 of device 4 0 cloned to file var upgrade boot rfs 1 of device 4 1 lt INFO gt shmm700_hal c 1255 Write storage state available gt not available ALLOW lt INFO gt shmm700_hal c 1642 Cleaning image 1 sentry shmm700 app lt INFO gt backend c 772 Burning lt INFO gt backend c 806 2377014 bytes were written from file tmp _shmm700_sentry shmm700 app to dev 8 1 lt INFO gt backend c 850
253. he verbose tshark capture in section 5 4 1 shows an example of this limitation The description of the first packet indicates No corresponding message while the description of the second packet shows Response to 1 To filter messages in tshark both in controlled and unattended mode use the R option followed by the text of the filter expression in quotes For example to select and dump in details the messages with the destination address 0x20 and the source address 0x82 as in the previous section when reading from a trace file use the following command line tshark r ipmb_traced log V R ipmi header target 0x20 amp amp ipmi header source 0x82 The T option can be used to specify the format of the output The following formats are Supported plain text PostScript PDML Packet Details Markup Language based on XML PSML Packet Summary Markup Language based on XML and so called field format For example the following command line can be used to dump message details in the PostScript format tshark r ipmb_traced log V T ps The e option can be used to select specific packet fields to be dumped if this option is used the field format must be also specified T fields Several e options can be used to specify multiple fields In conjunction with the filtering option R this option provides a powerful tool for queries on the message stream similar to relational database queries For example th
254. here it stores the map of the last booted Linux kernel and RFS images The map describes how the images are laid out in terms of physical Flash sectors and using this map allows U Boot during the second and subsequent boots of the same image set to load images using direct Flash read operations and to avoid expensive mounting of UBIFS volumes This information is stored ona dedicated Flash partition It is invalidated explicitly by the reliable upgrade utility before writing a new image set during a reliable upgrade The U Boot instance number and the number of the chosen set of images is calculated by the following formula Number UBOOT_IMAGE_SEL Candidate Where UBOOT_IMAGE_ SEL is the ShMM 700 soft jumper the value of which is the number of the current confirmed U Boot image and other firmware images 0 1 Candidate is a flag that has a value of 1 if a reliable upgrade is currently in progress and the candidate set of images should be loaded and 0 otherwise indicates the Exclusive OR logical operation The following table provides a summary of the Flash partitions maintained by Pigeon P oint Linux for the ShMM 700 Table 25 Flash Partitioning for ShMM 700s 64MB Flash OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 0 0 5 dev mtd0 Not mounted U Boot image 0 0 5 0 5 dev mtd2 Not mounted U Boot image 1 1 0 25 dev mtd4 Not mounted U Boot environment vari
255. here the trace is forwarded to directly to the analyzer as itis collected The following sections provide guidance for each of these modes as well as for collecting and analyzing an IP MI over network trace 5 4 1 Collecting IPMI Traces on IPMB 0 in Unattended Mode In unattended mode a trace is collected into a target local file transferred to the analyzer host and then read and analyzed using the trace analyzer The following stepwise procedure shows how to collect an unattended trace and open it for analysis in the GUI The GUI specific steps are followed by corresponding guidance for opening the trace for analysis in tshark Release 3 2 0 176 July 31 2013 Pigeon Point Shelf Manager User Guide 1 Startipmb_traced onthe ShMM with the option w optionally specifying the file to put the trace to ipmb_traced w u var ipmb_traced log amp If no u option is provided ipmb_t raced puts the trace in the following file var ipmb_traced ipmb_traced 1log In this case the var ipmb_traced directory must be created manually using the mkdir command before ipmb_t raced is Started mkdir var ipmb_traced 2 Execute actions that create the IP MI traffic on IPMB 0 to be analyzed For example if insertion of a board generates the sequence of events and commands that need to be analyzed do such a board insertion 3 When trace collection is complete stop ipmb_traced by sending the SIGINT signal to ipmb_traced e Obtain
256. hy and Active are set for the local ShMM For the ShMM 500 and ShMM 1500 make sure that the CPLD control word has the following bits Set e 0x02 Local Healthy e 0x04 Local Switchover Request e 0x20 Independent IPMB Watchdog To turn all these bits on use the following ep1d command cpld 26 26 For the ShMM 700 use the hwri command to set these flags and the Local Presence flag hwri set presence healthy switchover After that make sure that the output of hwri reports the Active flag set The remaining case where the Shelf FRU Information resides on separate IPM controllers inside the shelf is completely shelf specific and is beyond the scope of this document 3 5 3 Setting up the Shelf FRU Information Using the CLI Because of the limitations and system dependencies of the approaches listed in the previous Sections another method is recommended to update the Shelf FRU Information This method involves using the CLI command frudataw and itis applicable to all of the three cases considered in the previous section This method requires that the Shelf FRU Information file to be placed on the active ShMM After that the following command can be used possibly several times to update the locations that contain the Shelf FRU Information clia frudataw lt ipmc address gt lt fru id gt lt file gt where lt ipmc address gt is the address of the IPM controller that contains the Shelf FRU Informati
257. iguration variable CPLD_ACTIVE_WORKAROUND is set to FALSE the default value is TRUE for all carriers and changing it is not recommended The workaround described in this paragraph does not apply to the ShMM 700 and the configuration variable CPLD_ACTIVE_WORKAROUND has no effect on the ShMM 700 After a switchover the former backup now active Shelf Manager performs additional initialization applies cached state changes and collects any necessary further information from the IPM Controllers on IPMB The new active Shelf Manager then exposes the ShMC device address 20h Release 3 2 0 138 July 31 2013 Pigeon Point Shelf Manager User Guide on IPMB and assumes the IP address that was used for RMCP and other shelf external interactions between the formerly active Shelf Manager and the System Manager Since the RMCP Session information is propagated from the active Shelf Manager to the backup Shelf Manager RMCP sessions survive the switchover For the System Manager using RMCP the switchover is transparent The switchover is also transparent for the Web interface and for the SNMP interface provided that they use the redundancy IP address that is switched over Command line interface sessions since they are initiated locally on the target Shelf Manager do not survive a switchover and need to be re established again on the newly active Shelf Manager The command line interface support on the backup Shelf Manager is limited but it does allow
258. il boot etc rce Selecting carrier from the environment etc re etc rc Extracting shelfman rootfs patch from 0 sentry shmm700 app 32979 1 records in 4122 1 records out 2110464 bytes 2 0MB copied 3 647968 seconds 565 0KB s etc rce Extraction result 0 etc rce Placed var tmp to ram disk etc rce Setting hostname shmm700 etc rce Started syslogd and klogd lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW etc re Strobing the reliable upgrade WDT Here rupgrade is invoked for the second time to strobe the reliable upgrade WDT etc rc Mounted 0 etc to etc etc re Calling etc rce shmm700 hpdl etc readhwaddr Board Hardware Address 0x12 etc netconfig etc hosts updated with shmm700 192 168 0 23 entry etc netconfig Updating etc profile sentry with IP settings etc netconfig Starting bin inetd etc netconfig Starting time synchronization script Release 3 2 0 238 J uly 31 2013 Pigeon Point Shelf Manager User Guide etc rc shmm700 hpdl Updating etc profile sentry with specific settings etc rce shm etc rce shm etc rce shm 700 hpdl Starting snmpd 700 hpdl Starting httpd 700 hpdl Starting Shelf Manager for carrier type PPS700 etc rc shmm700 hpdl Command line shelfman sf etc rce shmm700 hpdl Strobing the reliable upgrade WDT lt INFO gt shmm700_hal c 1265 Write upgrade watchdog 1 gt 2 ALLOW
259. ing is ON lt I gt 14 35 22 285 205 SEL dedicated write thread started lt I gt 14 35 22 291 206 SEL truncation thread pid started successfully lt I gt 14 35 22 294 199 ShM SEL Activation complete lt I gt 14 35 22 352 208 PEF thread starting lt W gt 14 35 22 355 199 External event handler is not defined so PEF external handler thread is not started lt I gt 14 35 22 357 199 PEF activated successfully lt I gt 14 35 22 359 199 PEF initialized successfully System Event Sensor 133 lt I gt 14 35 22 363 199 Chassis facility activated successfully lt I gt 14 35 22 366 199 Chassis facility initialized successfully lt I gt 14035223L 199 Registering shelf FRU notification 0x2abd05f8 lt I gt 14 35 22 416 199 Session session descriptor size 996 bytes lt I gt 14 35 22 490 215 Started incoming message thread hc 1 lt I gt 14 35 22 493 215 Status change on IPMB 0 flags 186 ONL ATT ONL_CH ATT_CH lt I gt 14 35 22 495 215 IPMB O went online lt I gt 14 35 22 498 215 Status change on IPMB 1 flags 186 ONL ATT ONL_CH ATT_CH lt I gt 14 35 22 500 215 IPMB 1 went online lt I gt 14 35 22 524 216 IPMC Stored write thread started lt I gt 14 35 22 663 199 Starting power thread lt I gt 14 35 22 681 199 Controller FC FRU 0 ATCA state set to M1 prev M0 cause 0 locked 0 lt I gt 14 35 22 689 199 Created IPMC for 84 exists 0 lt I gt 14 35 22 691 199 Get Stored Device ID OK fo
260. ion in progress the returned value is 0 the startup script proceeds with sanity validation of the newly installed software In the middle of validation it calls backend strobe to strobe the upgrade WDT in case the validation takes longer than the upgrade WDT timeout period Finally the Startup script starts the Shelf Manager to perform final validation The watchdog timer interval is set to 120 seconds so the processing time in the etc rc Script between the call to rupgrade status and the call to strobe the WDT and between the call to strobe the WDT and the startup of the Shelf Manager must not exceed 120 seconds each e During initialization the Shelf Manager strobes the upgrade WDT once again before trying to establish a network connection with the peer Shelf Manager Establishing a network connection may take up to 6 seconds After that and after successfully finishing the initialization which indicates validity of the new configuration the Shelf Manager makes a call to backend confirm which completes the upgrade procedure e Atthis point the user can invoke rupgrade status in order to find out the status of the current upgrade session This may be useful if a reliable upgrade procedure is started remotely over the network via a virtual terminal telnet rsh or ssh connection because in that case the messages issued to the local serial console are not directly available e After the completion of the reliable upgrade the use
261. ion is based on the value of the Shelf Manager geographic address In PICMG 3 0 systems where full hardware addresses are available hardware addresses assigned to the slots where peer Shelf Managers reside must have different least significant bits In the case of a tie the Shelf Manager with the even hardware address becomes the active Shelf Manager and the Shelf Manager with the odd hardware address becomes the backup In PICMG 2 x systems where the Shelf Manager geographic address is represented by the GEOGRAPHIC_POSITION bit the Shelf Manager with GEOGRAPHIC_POSITION 0 becomes the active Shelf Manager and the other Shelf Manager becomes the backup After the connection is established the active Shelf Manager performs the initial synchronization with the backup Shelf Manager sending it all the appropriate redundant data During the initial synchronization the backup Shelf Manager verifies that the active Shelf Manager uses the same Release 3 2 0 139 July 31 2013 Pigeon Point Shelf Manager User Guide carrier specific module If the carrier specific modules are different further data propagation and collaboration between the active and the backup Shelf Managers is not possible Therefore in that case the backup Shelf Manager does not proceed with initialization but reboots and reestablishes the connection in a loop expecting the active Shelf Manager with the correct carrier specific module to appear The only exception to this is th
262. ironment variables are defined For SAMM 500 addip setenv bootargs S bootargs ip S ipaddr serverip gatewayip netmask hostname ipdevice addmisc setenv bootargs bootargs quiet console S console baudrate bootargs root dev ram rw console ttyS0 115200 bootcmd run ramargs addmisc bootm kernel_start rfs_start bootdelay 3 bootfile sentry kernel baudrate 115200 console ttys0 ethaddr 00 00 1la 18 xx yy ethladdr 00 00 1la 18 xx 2zz netmask 255 255 0 0 hostname shmm500 gatewayip 192 168 0 1 ipdevice eth0 ipladdr 192 168 1 2 ipldevice ethl rc2 etc re carrier3 jpaddr 192 168 0 22 start_rc2_daemons y flash_reset n password_reset n logging ram net tftpboot 80400000 S bootfile tftpboot 81200000 ramdisk run ramargs addmisc bootm 80400000 81200000 nfs tftpboot 80800000 S bootfile run nfsargs addip addmisc bootm nfsargs setenv bootargs root dev nfs rw nfsroot S serverip rootpath quiet quiet Release 3 2 0 31 July 31 2013 Pigeon Point Shelf Manager User Guide rc_ifconfig n ramargs setenv bootargs root dev ram rw ramdisk sentry rfs rootpath rootfs rmcpaddr 192 168 0 2 serverip 192 168 0 7 timezone UTC For ShMM 1500 addip setenv bootargs bootargs ip S ipaddr serverip gatewayip netmask hostname ipdevice addmisc setenv bootargs bootargs quiet console S console baudrate bootargs root dev ram rw console ttyS0 115
263. is parameter specifies the total number of Shelf Manager retries to execute a tasklet activation deactivation getting information etc if it encounters a transient failure when doing that If the Shelf manager fails to perform a tasklet after the TASKLET_RETRIES retries it gives up Increasing this value may improve the robustness of the Shelf Manager operation but degrade its performance CAN BE OBTAINED FROM SHELF FRU INFO Yes TIMEPROTO String 16 un Protocol used to retrieve time from a network time server possible values are ntp and rdate Yes TIMESERVER IP address None IP address of the time server for synchronization at runtime If this variable is not specified time is extracted from the hardware clock at shelf startup Yes TURBO_MODE_MIN _FAN_FAILURES Number This parameter specifies the number of tachometer underspeed faults crossing Major or Critical thresholds cause the remaining fans to go to TURBO mode that is run at full Speed in the default HPDL cooling management algorithm If the shelf implements zoned cooling the specified number of tachometer faults is counted relative to each zone and setting TURBO mode affects only the fan trays that participate in cooling of the corresponding zone Yes Release 3 2 0 68 July 31 2013 TURBO_MODE_MIN _MISSING_FAN_T RAYS Pigeon Point Shelf Manager User Guide Number DEFAULT
264. is set up by default The user can run the reliable upgrade utility in the Add Hook mode to install a hook script The script is copied to the directory etc hook so if the directory etc is copied to the candidate location this installation is permanent however if the command line parameter reset etc is specified all hook scripts are erased after the upgrade 8 6 Reliable Upgrade of A2F060 Firmware To upgrade the A2F 060 firmware run the reliable upgrade utility in the A2F 060 upgrade mode rupgrade upgrade a2f lt URL gt skip checksums Here lt URL gt designates the file that contains the A2F060 upgrade firmware image the option skip checksuns if specified requests the utility not to verify the checksum of the firmware image The image file includes all programmable contents of the A2F device including FPGA fabric FlashROM configuration and Cortex M3 firmware A reliable upgrade of A2F 060 firmware is done separately from a reliable upgrade of any other ShMM 700 firmware components and involves resetting the ShMM Also if the current image set has number 1 this procedure has a side effect that U Boot image 0 is overwritten by the current U Boot image if the current image set has number 0 there are no such side effects The reliable upgrade of the A2F 060 firmware includes the following steps e The upgrade utility copies the upgrade image file onto the Flash Boot partition as a2 upgrade dat ef the c
265. ism that is not based on HPDL No ISOLATE_MUX_BU S Number Platform dependent 5 on ShMM 700 0 on other platforms The number of the logical master only 12C bus on which the multiplexer resides on platforms where SHMM_GPIO8 is used to control access from the Shelf Manager to the multiplexer on the master only 12C bus via the original mechanism that is not based on HPDL No ISOLATE_MUX_IG NORE_COUNT Number 10 This parameter instructs the isolation algorithm to skip this number of accesses to a faulty bus before trying to enable it on the platforms where SHMM_GP108 is used to control access from the Shelf Manager to the multiplexer on the master only 12C bus via the original mechanism that is not based on HPDL No Release 3 2 0 56 July 31 2013 ISOLATE_MUX_IG NORE_ TIME Pigeon Point Shelf Manager User Guide TYPE Number DEFAULT DESCRIPTION This parameter instructs the isolation algorithm to avoid accessing the faulty bus and immediately return a failure for specified number of seconds after a failure detection on the platforms where SHMM GPIOs are used to control access from the Shelf Manager to multiplexers on the master only IC bus via an HPDL based mechanism CAN BE OBTAINED FROM SHELF FRU INFO No ISOLATE_MUX_ON _GPIO8 Boolean FALSE Must be set to TRUE for the platforms where SHMM_GPI08 is used to control acces
266. isolated from its peer segments on a logical basis Radial operation of ShMM 1500 requires all of the following 1 radial support on the shelf level 2 the presence of a radial IPMB description record in the Shelf FRU Information and 3 radial support on the ShMM carrier For non HP DL systems a special Pigeon Point defined record the IP MB topology record must be present in the ShMM carrier FRU information to specify the type of the IPMB topology supported by the carrier In HPDL based systems the IPMB topology type can be Specified in the IPMB_ TOPOLOGY clause in the HPDL Carrier definition in that case the IP MB topology record is not needed Two different topologies of radial IP MB 0 are supported by the ShMM 1500 dual star and redundant dual star The implemented topology is determined by the design of the ShMM carrier With the dual star topology only one IPMB hub exists on each ShMM carrier This hub connects the ShMM to radial links on either IP MB A or IPMB B depending on the physical position of the ShMM carrier in the shelf This topology has the following properties e Provides equivalent connectivity to the dual star topology that ATCA defines for the Gigabit Ethernet Base Interface Includes two cross links that connect an IPMB either A or B porton one ShMM to the radial hub on the other ShMM carrier Release 3 2 0 145 July 31 2013 Pigeon Point Shelf Manager User Guide e Requires two installed Shelf Managers for dual r
267. ix C Revision History This section lists the changes that have been made in each revision of this document starting with the 2 1 0 release C 1Release 2 1 0 e Section 7 3 corrects the boundaries of the Flash partitions maintained on the ShMM 500 by FOSL for 16MB Flash devices e Section 7 3 covers the boundaries of the Flash partitions maintained on the ShMM 500 by FOSL for 32MB Flash and 64MB Flash devices C 2Release 2 2 0 e Section 3 4 3 elaborates on the algorithm for computation of IP address for endpoints of USB network interface e Section 3 3 introduces new configuration parameters ALLOW_ALL_ COMMANDS FROM_IPMB ALLOW_CHANGE_EVENT_RECEIVER ALLOW_RESET_STANDALONE DEFAULT_RMCP_NETMASK IPMB_LINK_ISOLATION_TIMEOUT MAX _INCOMING_IPMB_ REQUESTS MAX OEM FILTERS SENSOR_POLL_INTERVAL SHELF_FRU_IPMB_SOURCE1 SHELF_FRU_IPMB_SOURCE2 SWITCHOVER_ON_HANDLE_OPEN e Section 3 5 1 adds an option to limit the search for potential sources of Shelf FRU Information on IPMB 0 to the two well known IPMB 0 locations defined by the configuration variables SHELF_FRU_IPMB_SOURCE1 SHELF_FRU_IPMB_SOURCE2 e Section 3 8 covers a new option for local sensors to be configured when the Shelf Manager is Started The Sensor Device Records SDRs defining these sensors are read from the file var nvdata user_sdr e Section 4 3 provides the details of command line invocations of the Shelf Manager e Section 4 5 describes Shelf Manager operation
268. l tmp _tmp_kernel rootfs tmp _tmp_rfs application tmp _tmp_app copy etc lt INFO gt shmm700_hal c 1285 Write storage state not available gt available ALLOW UBI device number 1 total 484 LEBs 31657472 bytes 30 2 MiB available 0 LEBs 0 bytes LEB size 65408 bytes 63 9 KiB lt INFO gt shmm700_hal c 369 Resetting UBIFS cache Erasing 64 Kibyte 30000 75 complete lt INFO gt shmm700_hal c 1913 Clean on defice 1 0 Leaving file dev mtd0 untouched lt INFO gt backend c 730 Burning image lt INFO gt backend c 764 393216 bytes was written from file tmp _tmp_uboot to dev 1 0 Q08 7 8200 00 4 Release 3 2 0 234 July 31 2013 Pigeon Point Shelf Manager User Guide Here and below rupgrade reports that a new image has been written to the storage device The number in brackets is the candidate image set number In this example the confirmed image number is 1 and the candidate number is 0 Erasing 64 Kibyte 30000 75 complete 512 0 records in 512 0 records out 262144 bytes 256 0KB copied 1 567781 seconds 163 3KB s lt INFO gt backend c 730 Burning lt INFO gt backend c 764 1402348 bytes was written from file tmp _tmp_kernel to dev 2 0 lt INFO gt backend c 730 Burning lt INFO gt backend c 764 2283039 bytes was written from file tmp _tmp_rfs to dev 4 0 lt INFO gt shmm700_hal c 1285 Write storage stat
269. l sensors are defined that expose the current cooling state of the shelf to the System Manager These sensors belong to the logical Shelf Manager IPMB address 20h and are defined in the following table Release 3 2 0 150 July 31 2013 Pigeon Point Shelf Manager User Guide Table 13 Cooling State Sensors SENSOR SENSOR NAME EVENT READING SENSOR TYPE OFFSETS NUMBER TYPE 135 FT Oper Status Redundancy Management 00h Full Redundancy all fan Bh Subsystem trays defined in the Address Table Health 28h are operational 01h Redundancy Lost some of the fan trays defined in the Address Table are missing or non operational With the HPDL default cooling management Strategy this causes the fan level for all remaining fan trays to be Set to their maximum 136 Cooling State Severity 7 Management 00h transition to OK The Subsystem cooling state is Normal Health 28h 01h transition to Non Critical from OK The cooling state is now Minor Alert the previous cooling State was Normal 02h transition to Critical from less severe The cooling state is now Major Alert the previous cooling state was either Normal or Minor Alert 04h transition to Non Critical from more severe The cooling State is now Minor Alert the previous cooling state was either Major or Critical Alert 05h transition to Critical from Non recoverable The current cooling state is Major Alert the previous cooling state was Critical Alert 06h
270. l4slot sdr bin gz clia frudataw d 20 1 stdl4slot hpdl bin gz clia frudataw s 20 2 stdl4slot sdr bin gz 3 6 4 4 Placing HPDL Data and SDRs in the FRU Information for Specific FRUs This section addresses placing HPDL data and SDRs into the FRU Information for a specific FRU e g a fan tray This information is used by the Shelf Manager to amend the attributes of that FRU from the attributes for that FRU that are specified in the HPDL chassis definition The following information about the FRU is supplied in the chassis definition and cannot be redefined by the HPDL data in the FRU e FRU site type and number e Presence signal definition Release 3 2 0 105 July 31 2013 Pigeon Point Shelf Manager User Guide Device selection definitions can be redefined by the HPDL data in the FRU however a problem arises when several slots exist in the chassis for the same type of the FRU and addresses of devices of the FRU are different between different slots In that case itis not possible to define device location information on the FRU level it is possible to do that only on the chassis level This consideration limits the possibilities of redefining device information on the FRU level Itis not necessary to have both HP DL data and SDRs in a given FRU Information area only HP DL data or only SDRs may be present For example to change the power consumption of the FRU it is Sufficient to have only HPDL data To change certain sensor attribut
271. lated exceptions Several bits of the exception group can be set in the mask along with one of the normal state bits The meanings of the bits in the sensor state mask are given below High level redundancy state bits normal state 0 The current Shelf Manager is Active with no Backup 1 The current Shelf Manager is Active with a Backup 2 The current Shelf Manager is a Backup Low level exception state bits should be treated as low level errors if set 4 The Shelf Manager is a Backup but the remote presence bit is not set 5 The Shelf Manager is a Backup but the remote switchover request bit the remote active biton ShMM 700 is not set 6 The Shelf Manager is a Backup but the HRI Active bit is set 7 The Shelf Manager is Active with a Backup but the remote presence bit is not set 8 The Shelf Manager is Active with a Backup but the remote healthy bit is not set 9 The Shelf Manager is Active with a Backup but the HRI Active bit is not set 10 The local presence bit is not set for the current Shelf Manager 11 The Shelf Manager is Active with no Backup but the remote healthy bit is set 12 The Shelf Manager is Active with no Backup but the remote switchover request bit the remote active biton ShMM 700 is set Release 3 2 0 140 July 31 2013 Pigeon Point Shelf Manager User Guide 13 The Shelf Manager is Active but senses the HRI Active bit set on other ShMM The current reading and state m
272. le This allows a manual rollback procedure to be initiated by invoking the rupgrade utility running in the upgrade a2f mode with the file a2 manual rollback dat as the parameter The UBIFS Boot volume may need to be explicitly mounted first If the upgrade process is interrupted or an error is detected the ShMM 700 will get reset with the Flash remapping logic turned off As a result it will boot from partition 0 and U Boot will detect that Flash remapping logic is turned off In this case it will try to revert to the previous image by repeating the above steps with the auto rollback image file name a2 auto rollback dat NOTE the A2F060 upgrade images provided by Pigeon Point Systems are encrypted with a private AES key This key is pre programmed in the A2F060 device on the ShMM 700R during manufacturing As a consequence A2F060 upgrade images that were not built by PPS cannot be installed onto the A2F060 device on the ShMM 700R Attempting to do so will result in a failed upgrade 8 7 Reliable Upgrade Utility Use Scenarios Itis intended that the reliable upgrade utility is used in the following sequence in order to carry out an upgrade of the ShMM The user makes a call to rupgrade in upgrade mode or in HPM 1 upgrade mode to initiate the upgrade procedure The call can be made either locally from the ShMM serial console or remotely over the network via telnet rsh ssh or any equivalent The user waits for rupgrade to r
273. le via the network The bonding driver in the kernel is configured to use an active backup policy which means that only one slave device in the bond is active A different slave device becomes active only if the active slave device fails The bond s MAC address is externally visible on only one network adapter port to avoid problems with switch hub hardware Link status of the slave devices in the bond is polled every 100 milliseconds which is the recommended value Also in the case of an active device switch the bonding driver sends out a gratuitous ARP packet to notify switch hardware about the port change since bonding slave devices share the same MAC address taken from the first device in the slave device list When a network link loss is detected active device switching is performed only after 500 milliseconds to avoid potential problems with short term link loss The idea is similar to the Active S tandby usage of two network interfaces except that device link state monitoring and switching is performed at low level by the kernel instead of the Shelf Manager The bonding mode is activated by changing the ipdevice variable value in U Boot to bondo This has to be done at the U Boot level so that the system startup scripts perform the device initialization for bondo before the Shelf Manager starts By default this value is passed to the Shelf Manager as the default RMCP network adapter via the setting in etc shelfman conf RMCP_NET_ADAP
274. lowance for FRU Activation Readiness 10 seconds start 23 lt I gt 14 35 22 952 226 Found Power Management record format version 1 lt I gt 14 35 22 954 226 Retrieving initial power budget lt I gt 14 35 22 975 227 RMCP starting server thread for 192 168 1 198 623 lt I gt 14 35 22 977 227 RMCP started server thread for 192 168 1 198 623 lt I gt 14 35 22 979 227 Registering 227 facility to the triggering watchdog lt I gt 14 35 23 585 214 Hot Swap event SA 0x20 FRU 2 sensor_number 0x03 MO gt M1 cause 0 lt I gt 14 35 23 586 214 State M1 sa 20 FRU 2 op_state M0 power_cycle 0 lt I gt 14 35 23 647 214 Hot Swap event SA O0xfc FRU 0 sensor_number 0x00 MO gt M1 cause 0 lt I gt 14 35 23 649 214 State M1 sa FC FRU 0 op_state M0 power_cycle 0 lt I gt 14 35 23 659 214 Hot Swap event SA 0x20 FRU 2 sensor_number 0x03 MO gt M1 cause 0 lt I gt 14 35 23 661 214 State M1 sa 20 FRU 2 op_state M1 power_cycle 0 Release 3 2 0 133 July 31 2013 Pigeon Point Shelf Manager User Guide lt I gt 14 35 23 667 214 Hot Swap event SA 0xfc FRU 0 sensor_number 0x00 MO gt M1 cause 0 lt I gt 14 35 23 669 214 State M1 sa FC FRU 0 op_state M1 power_cycle 0 lt I gt 14 35 23 729 214 Hot Swap event SA 0x20 FRU 0 sensor_number 0x00 MO gt M1 cause 0 lt I gt 14 35 23 731 214 State M1 sa 20 FR
275. m is supported in the Shelf Manager starting with release 2 5 2 This mechanism checks the state of the currently claimed locks every 5 seconds by looking at the internal lock structures It analyzes that state information for circular dependencies that indicate a deadlock If a deadlock is detected by this mechanism the Shelf Manager does the following e issues a warning message to the console and or to the syslog file e collects information about current usage of locks in the Shelf Manager e writes this information to a Flash file on the ShMM var nvdata dumplog bin where it survives reboots and can be subsequently analyzed by the analysis tool dumplog e stops strobing the hardware watchdog timer which causes a reset of the ShMM anda Switchover to the backup Shelf Manager Note This second direct deadlock detection mechanism is not implemented on the ShMM 700 The reason is that with the new version of the Linux kernel 2 6 x used on the ShMM 700 the detection of deadlocks by direct analysis of internal lock mutex structures is not reliable and generates substantially larger performance overhead The first mechanism watchdog based monitoring of the CLI and RMCP processing threads should be sufficient for detection of deadlocks on this platform In this scenario an output similar to the following appears on the console lock Ox2acd3940 owner 208 waiting 209 lock Ox2ac432f0 owner 223 waiting 217 214 220 208 lock Ox2ac43650
276. me is rmep Passwords are not used h Display help text and exit help If no options are provided ipmb_t raced runs in unattended trace mode produces trace data in Ethereal format Wireshark format on the ShMM 700 and puts them in the var ipmb_traced ipmb_traced 1log file but does not trace IPMB state change events or collect any HPM 2 based IPM controller s trace data This scenario assumes that a Separate ShMM application such as the Pigeon Point Shelf Manager assigns IPMB addresses for the ShMM and attaches to the ShMM IPMB interface so that there can be IPMB traffic to trace The trace collection daemon stores the collected data in a file or sends it directly to the analyzer in the form of packets Each packet describes a single IPMB transaction or event and contains the following fields The detailed layout of the fields depends on whether the Ethereal or Wireshark format has been selected for the trace collection daemon Table 16 IPMB Trace Collection Packet Field Seconds Transaction completion time in seconds Micro seconds Transaction completion time in microseconds calculated from the start of the last second Event IPMB interface state change information Flags Special flags returned by the Linux kernel driver Bus IPMB interface IPMB A or IPMB B on which the transaction was captured Address Destination I2C address of the transaction Size Size of the transaction Data Transaction ra
277. meter is setto0 0 0 0 inthe IPMI LAN Configuration Parameters for channel 2 The default value depends on the Class of the default IP address used for the gateway for shelf external RMCP based communication see parameter DEFAULT_RMCP_IP_ADDRESS 2 For example for an IP address of class C this parameter is set to 255 255 255 0 CAN BE OBTAINED FROM SHELF FRU INFO No DEFAULT_VLAN_I D Number The default Virtual LAN ID used for the first LAN channel This value is used only as the default value for the corresponding LAN configuration parameter once LAN configuration parameters are defined and stored on the ShMM the value provided in the Shelf Manager configuration file is ignored Yes DEFAULT_VLAN_I D2 Number The default Virtual LAN ID used for the second LAN channel This value is used only as the default value for the corresponding LAN configuration parameter once LAN configuration parameters are defined and stored on the ShMM the value provided in the Shelf Manager configuration file is ignored Yes Release 3 2 0 47 July 31 2013 DETECT_DEADLOC KS Pigeon Point Shelf Manager User Guide Boolean DEFAULT TRUE DESCRIPTION This variable turns on the deadlock detection in CLI and RMCP server facilities in the Shelf Manager The detection is based on an internal watchdog that must be periodically strobed by the threads serving CLI and RMCP reque
278. meters for channel 1 sf The option s forces the Shelf Manager to use EEPROMs for Shelf FRU Info storage The option s forces the Shelf Manager not to use EEPROMs for Shelf FRU Info storage If neither of these options is present in the command line the value of the parameter SHELF_FRU_IN_EEPROM from the file etc shelfman conf determines Shelf Manager actions in this area p lt port gt This option sets the redundancy communication port If this option is not present the value of the parameter REDUNDANCY_PORT from the file etc shelfman conf determines the Shelf Manager actions in this area ph lt IPMB addr1 gt lt IPMB addr2 gt This option defines IPMB addresses of pseudo hubs by default 0x82 and 0x84 Pseudo hubs are virtual IPM controllers that are created and emulated by the Shelf Manager and behave like IPM controllers for Base Interface hub boards This option can be used for testing and when non intelligent hub boards non compliant boards that do not implement an IPM controller are installed in the shelf i lt command_line gt This option defines the command line for the initialization script that is invoked by the active Shelf Manager during initialization This initialization script can be used for platform specific initialization The CPLD bits Local Healthy and Active are set when the command line is invoked and the Shelf Manager waits for completion of the initialization script before progressing f
279. mm500 Then save these environment settings Use the saveenv command to store the settings shmmxx00 saveenv 6 3 Re initializing the File System The filesystem is stored within the Flash and can be reset to factory defaults quite easily U Boot has an environment variable called lash_reset By setting this variable to y and then booting up the system the file system is re initialized to factory defaults shmmxx00 setenv flash_reset y shmmxx00 saveenv shmmxx00 boot The flash_reset variable is automatically set to n at system startup after re initializing the Flash The boot command the second command above begins booting the Linux kernel It is during this process that the file system is re initialized The following output is shown on the console etc rc Mounted dev pts etc rc Flash erase requested via U BOOT var etc rce erasing mtdcharl gt etc Erased 1024 Kibyte 0 100 complete etc rce erasing mtdchar0 gt var Erased 1536 Kibyte 0 100 complete etc rce Mounted dev mtdblock3 to var etc rc var log mounted as FLASH disk etc re Started syslogd and klogd etc re var tmp mounted as RAM disk etc rc hostname demo etc rce dev mtdblock2 appears to be empty restoring from factory fRtCe s For the ShMM 700 the console output will look differently etc rce Mount ubid0 user Release 3 2 0 194 July 31 2013 Pigeon Point Shelf Manager User Guide etc rc Flash
280. ms of the reliable upgrade procedure by enabling the ShMM s upgrade watchdog timer WDT Refer to ML User s Guide Chapter 7 for background and details on the upgrade WDT The hook script etc upgrade step4hshm is supplied with the Shelf Manager It performs the following actions e terminates the Shelf Manager performs a switchover to the backup ShMM without restarting the shelf and stops the ATCA watchdog timer e mounts the provisional etc and var Flash partitions and erases all files on them e optionally copies the current contents of the etc directory to the provisional ete Flash partition e optionally copies the current non volatile Shelf Manager information from the directory var nvdata to the provisional var file system or optionally copies the whole var directory to the provisional var Flash partition e temporarily just for the next boot sets the boot delay to 0 this is done to minimize the time of the next boot and prevent the reliable upgrade watchdog timer from premature expiration This script is invoked as a sub shell and given a single parameter which is either the string specified by lt args gt or an empty string if no args is supplied The parameter defines the mode of operation of the script with respect to copying non volatile information from the persistent Flash partitions to the provisional Flash partitions and can take the following values each of which triggers a corresponding set of actions e no
281. mutexes in an internal memory area This information can be subsequently recovered and analyzed by the analysis tool dumplog in case of a deadlock or another synchronization violation Setting this variable to FALSE makes post mortem lock analysis impossible but substantially improves the performance of the Shelf Manager See section 4 8 for further background CAN BE OBTAINED FROM SHELF FRU INFO ENABLE_RTC_TRI Boolean FALSE If this variable is TRUE the trickle No ee oe charge feature of the RTC DS1339 device on the ShMM is enabled This Setting is required for ShMM carriers that connect a super cap to the RTC backup power supply input VBAT or VBACKUP_RTC EXIT_IF_HEALTH Boolean FALSE This variable defines what to do ifthe Yes Y LOST IN_STAN Shelf Manager runs without backup a and detects a loss of the local Healthy bit If this variable is TRUE the Shelf Manager exits the ShMM reboots and the Shelf Manager is restarted If this variable is FALSE the Shelf Manager sets the Healthy bit and continues operation EXIT_IF_NO_SHE Boolean FALSE If TRUE the Shelf Manager exits No pene probably resetting the ShMM if no Shelf FRU can be found EXTERNAL_EVENT String 255 This is the path to an executable file No HANDLER or a script file on the Shelf Manager that performs local handling of events via PEF Release 3 2 0 50 July 31 2013 Pigeon Point Shelf Manager User Guide TY
282. n Note The Shelf Manager doesn t support dynamic reconfiguration of the Shelf FRU After any modifications of the Shelf FRU performed via CLI RMCP SNMP and other interfaces the whole shelf must be restarted to accommodate the changes 3 5 4 Other FRU Information Repositories The Shelf Manager itself exposes at least one IPM controller the ShMC at IPMB address 20h For most carriers the Shelf Manager also exposes a physical IPM controller that represents the resources of the carrier board and has an IPMB address derived from the physical address of the carrier While the ShMC is exposed only by the active Shelf Manager and is subject to switchover the physical IPM controller is exposed separately by both active and backup Shelf Managers For both of these IPM controllers FRU Information is stored in Flash files on the ShMM var nvdata bmc fru information for the ShMC var nvdata shelfman fru information for the physical IPM controller For some carriers there may be additional FRUs represented by the ShMC The location of the FRU Information for these FRUs is carrier Specific Reading and writing these FRU Information repositories can be done via the IPMI commands Read FRU Data Write FRU Data addressed to the appropriate FRUs Release 3 2 0 100 July 31 2013 Pigeon Point Shelf Manager User Guide 3 6 Configuring Carrier and Shelf Attributes using HPDL A special description language has been develop
283. n Original Approach Single Multiplexer or Switch Not HPLD based In this approach the isolation algorithm is controlled by the value of the configuration parameter ISOLATE_MUX_ON_GPIO08 and works as follows e The multiplexer switch 7 bit12C address can be specified by the configuration parameter ISOLATE_MUX_ADDRESS The default is 0x70 e When the error ETIMEDOUT is detected by the Linux 12C subsystem while trying to access a device on a subordinate bus this bus is recorded as faulty in the Shelf Manager and the multiplexer switch is held in reset via GPIO E8 This operation causes all the channels to be deselected on the multiplexer switch which allows the Shelf Manager to access devices on bus 0 the main master only 12C bus even if one of the subordinate buses has a fault e The multiplexer switch is re enabled via GPIO E8 the next time some parts of the Shelf Manager tries to access it This allows the Shelf Manager to access devices on the Subordinate buses that are still functional e When an attemptis made to access the faulty subordinate bus by writing the channel number or mask for the faulty bus into the multiplexer switch this operation is not permitted for the ignore count subsequent attempts The faulty bus is not enabled an error is instead returned for the corresponding write operation The parameter ignore count is specified by the configuration parameter ISOLATE_MUX_IGNORE_COUNT e The algorithm knows which v
284. n as the source of the IP address netmask and default gateway for the second RMCP network interface New section 3 4 2 3 describes support for site dependent ShMC cross connects New section 3 6 describes configuring Carrier and Shelf Attributes using HPDL the Pigeon Point Hardware Platform Description Language New section 3 7 describes configuring the cooling management strategy on HPDL based platforms Section 3 8 specifies that the section applies only to systems where HPDL is not used New section 3 13 describes how to configure the Platform Event Trap format Section 4 4 1 describes the event message format for the Redundancy and CPLD state sensor Section 4 4 3 augments the list of reboot causes that are reported by the reboot reason sensor New section 4 7 describes the cooling state sensors Section 4 8 adds information about the direct deadlock detection mechanism Section 7 changes the heading structure to eliminate an unnecessary heading level Release 3 2 0 257 J uly 31 2013 Pigeon Point Shelf Manager User Guide C 10 Release 2 5 3 Section 3 3 corrects the description of the configuration parameter COOLING_MANAGEMENT Section 3 6 3 amends the description of the procedure to locate HPDL data taking into account the new environment variables that can provide the location of that data New section 4 9 describes the provisions for isolating faulty subsidiary buses behind a multiplexer on the ShMM carrier fo
285. n standard output in a human readable format It includes information about which locks are owned by which threads and a per thread history of lock acquisition and release for each currently active Shelf Manager thread The output of the command looks something like this xxx 24 Log for pthread with pid 263 18 17 Jun 22 17 23 27 2007 Succeeded 2ac34530 pid 263 file sdrrep c line 4288 Jun 22 17 23 27 2007 Released 2ac34530 pid 263 file sdrrep c ne 3715 n 22 17 23 27 2007 Acquiring 2ac34ad0 pid 263 file ipmc c ne 387 n 22 17 23 27 2007 Succeeded 2ac34ad0 pid 263 file ipmc c ne 387 1 Ju ae a n 22 17 23 27 2007 Released 2ac34ad0 pid 263 file ipmc c L 1 Ju ae u un 22 17 23 27 2007 Acquiring 7d5ff 740 pid 263 file msg c ne 775 n 22 17 23 27 2007 Succeeded 7q45ff740 pid 263 file msg c HiG u cy ine 484 line 775 Jun 22 17 23 27 2007 Acquiring 2acclf80 pid 263 file ipmf c line 1360 Jun 22 17 23 27 2007 Succeeded 2acclf80 pid 263 file ipmf c line 1360 Release 3 2 0 154 July 31 2013 Pigeon Point Shelf Manager User Guide Jun 22 17 23 27 2007 Acquiring 2ac344a0 pid 263 file msg c line 924 Jun 22 17 23 27 2007 Succeeded 2ac344a0 pid 263 file msg c line 924 Jun 22 17 23 27 2007 Released 2ac344a0 pid 263 file msg c line 932 Jun 22 17 23 27 2007 Acquiring 2ac3
286. nager User Guide e Section 3 14 2 corrects the path to the O penHPI client library e New section 4 4 1 describes in detail the initialization of redundant Shelf Managers e Section 4 10 2 adds a description of how the Shelf Manager retrieves LAN parameters from a DHCP Server e New section 4 10 2 3 describes the configuration of a Linux DHCP Server to enable retrieving LAN Configuration information from it e New section 4 10 4 describes the synchronous assignment of LAN configuration parameters to a board C 17 Release 2 7 1 e Section 3 3 adds new configuration parameters ALLOW_POWER_UNRELATED_FRU_IN_CRITICAL_STATE REAPPLY _POWER_MAX COOLING_STATE e Section 3 7 1 adds a description of the effects of the new configuration variables ALLOW_POWER_UNRELATED_FRU_IN_CRITICAL_STATE and REAPPLY _POWER_MAX COOLING_STATE Section 3 10 1 clarifies that ShMM timestamps correspond to the local time zone Section 4 4 3 adds descriptions of two more states of the Reboot Reason sensor New section 4 4 1 describes the HPI System Event sensor Section 5 2 new section 5 4 3 describes the IPMB board trace mode Section 7 6 adds a description of a new parameter conf for the upgrade script step4hshm C 18 Release 2 7 2 e Sections 7 6 7 7 adds a description of a new parameter 1ip for the upgrade script step4hshm Notes that the option u is no longer supported e New section 3 15 describes configuration of local services C 19
287. nd if itis still 1 samples it again after some delay If the Remote Healthy bit Release 3 2 0 137 July 31 2013 Pigeon Point Shelf Manager User Guide becomes 0 the backup Shelf Manager concludes that the active Shelf Manager is dead In that case it initiates a switchover and assumes the active Shelf Manager role Otherwise if the Remote Healthy bit retains the value of 1 the backup Shelf Manager concludes that the communication link between the Shelf Managers is broken In that case no switchover is initiated instead the backup Shelf Manager repeatedly reinitializes itself and tries to establish a connection with the active Shelf Manager until the communication link is restored Re initialization is achieved by rebooting the ShMM and automatically restarting the Shelf Manager after the reboot Special logic in the Shelf Manager guarantees that it does not try to become active at Startup if the peer Shelf Manager is already active The Shelf Manager uses a watchdog timer to protect against becoming unresponsive due to infinite loops or other software bugs In the event the watchdog timer on the active Shelf Manager triggers that ShMM is reset causing the Remote Healthy bit on the backup ShMM to become 0 and triggering a switchover In addition to the scenarios described above the active Shelf Manager monitors the state of the Local Healthy and Active bits If either of these bits becomes 0 on the active Shelf Manager and the backup Sh
288. nding lines in that script to disable it 3 16 Configuring the Speed of I2C Buses ShMM 700 Only Two of the ShMM 700 I2C buses with logical numbers 2 and 3 support both high speed 400KHz and normal speed 100KHz operation By default both buses operate in high speed mode but this can be changed by means of a Linux kernel startup parameter i2c mxs speed which has the following syntax i2c mxs speed lt speed1 gt lt speed2 gt where lt speedi1 gt and lt speed2 gt can have values 100 or 400 and represent the speed of buses 2 and 3 respectively According to normal Linux kernel naming conventions the name of the parameter is composed of two parts separated by the dot symbol i2c mxs corresponds to the IC driver name and speed s the parameter name within that driver For example to set both buses to normal speed the following parameter should be passed to the Linux kernel i2c mxs speed 100 100 Release 3 2 0 126 July 31 2013 Pigeon Point Shelf Manager User Guide The recommended method to pass this parameter to the kernel is to append it to the value of the U Boot variable bootargs_initrd for example by typing the following at the U Boot prompt shmm700 setenv bootargs_initrd bootargs_initrd i2c mxs speed 100 100 shmm700 saveenv Release 3 2 0 127 July 31 2013 Pigeon Point Shelf Manager User Guide 4 Using the Shelf Manager This section introduces the overall operation of th
289. ndor specific requirements that need to be satisfied for specific shelves e g turn on the Critical TELCO alarm when the thermal condition in the shelf becomes critical Until now vendor specific cooling management was implemented inside the Shelf Manager With the advent of HPDL as a means to describe the architecture of carriers and shelves without additional coding it became desirable to make cooling management modular so that cooling management strategies can be added to the Shelf Manager without modification of the Shelf Manager code and the proper cooling management strategy can be chosen dynamically To Satisfy this goal the following approach is supported in the Shelf Manager e Cooling management modules are implemented as Linux shared libraries so files e A cooling management API is defined this API fully defines the interface between the Shelf Manager and the cooling management module in the form of function calls e Support has been added to the Shelf Manager for dynamically loading the cooling management modules and binding to the functions that they export Release 3 2 0 106 July 31 2013 Pigeon Point Shelf Manager User Guide This approach is utilized only if HPDL is used to describe the carrier and the shelf For non HP DL defined carriers and shelves the cooling management strategy remains bound to carrier specific code in the Shelf Manager 3 7 1 Default Cooling Management Strategy With HPDL a reasonable d
290. ndpoint on the first Shelf Manager and the usb0 endpoint on the other Shelf Manager are in the second range The 4 IP addresses in question can be derived from one IP address for example the IP address assigned to usb0 on the Shelf Manager with the even hardware address and the network mask for which the recommended value is 255 255 255 128 The rules are as follows e To compute the IP address for usb0 on the Shelf Manager with the even hardware address when the file etc readhwaddr is present you should set the least significant bit of REDUNDANCY_IP_ADDRESS to 0 e To compute the IP address for usb1 on the other Shelf Manager you should toggle the least Significant bit in the usb0 IP address on the first Shelf Manager This guarantees that usb0 on Shelf Manager with the even hardware address and usb1 on the Shelf Manager with the odd hardware address are in the same logical network and are not equal to each other e To compute the IP address for usb1 on the Shelf Manager with even hardware address you Should take the IP address for usb0 on the same Shelf Manager and toggle the least non zero bit of the network mask This guarantees that the IP addresses for usb0 and usb1 on the Shelf Manager with the even hardware address are in different logical networks Release 3 2 0 80 July 31 2013 Pigeon Point Shelf Manager User Guide e The last step is to compute the IP address for usb0 on the Shelf Manager with the odd hardware address You should
291. nds is set to channel N For modules commands are wrapped in Send Message commands The following batch of Set LAN Configuration Parameters commands is sent Channel N Parameter Set In Progress 0 data set in progress Channel N Parameter P Address 3 data board IP address Channel N Parameter Subnet Mask 6 data board subnet mask Channel N Parameter Default Gateway 12 data board default gateway IP address Channel N Parameter Set In Progress 0 data commit write Channel N Parameter Set In Progress 0 data set complete 4 10 4 Synchronous Assignment of LAN Configuration Parameters to Boards In the case of obtaining parameters from a DHCP server if the configuration variable BOARD_LAN_ PARAMETERS USE_DHCP is TRUE activation of LAN enabled boards and modules transitioning from M3 to M4 is not affected by the process of retrieving LAN configuration parameters for them a board may reach M4 before its parameters are retrieved and assigned However it is possible to instruct the Shelf Manager to assign LAN configuration parameters to designated boards modules in a synchronous way before allowing them transition to M4 This is done by delaying power assignment to these FRUs until LAN configuration parameters are retrieved and assigned to them This feature is controlled by a Shelf Manager configuration variable BOARD_LAN PARAMETERS SYNCHRONOUS that contains the list of all FRUs that are to be treated this
292. ned on a single chip Correspondingly the firmware for A2F 060 consists of an FPGA fabric image and the program code for the Cortex M3 processor combined in a single composite image Update of the A2F 060 firmware requires special precautions and is implemented as a special case of the reliable upgrade procedure This special case is documented in detail in section 8 6 Itis expected however that A2F 060 firmware will be relatively stable and that most new releases of the Shelf Manager in future will not require an A2F 060 firmware upgrade Release 3 2 0 221 July 31 2013 Pigeon Point Shelf Manager User Guide 8 3 Flash Partitioning and Image Layout The ShMM 700 provides a hardware mechanism that allows swapping of the two Flash partitions at certain addresses under control of software running on the i MX287 This capability is implemented in support of the reliable upgrade procedure and applies to U Boot images stored on dedicated Flash partitions Other firmware images the Linux kernel RFS and application package are stored in files on UBIFS volumes and there are two sets of these images numbered 0 and 1 The choice of the set to use is made in U Boot based on the chosen instance number evaluated by the U Boot Therefore a hardware mechanism is needed only for choosing the U Boot instance and exposing the chosen instance number to the software To accelerate loading of images from a UBIFS partition U Boot implements a special cache w
293. nfiguration parameter VERBOSITY_CONSOLE is omitted The VERBOSITY configuration parameter is a hexadecimal bit mask with each bit enabling output of a specific type of message Table 9 Verbosity Configuration Parameters and Levels VERBOSITY CONFIGURATION PARAMETER VERBOSITY LEVEL 0x01 Error messages 0x02 Warning messages 0x04 Informational messages 0x08 Verbose informational messages 0x10 Trace messages not recommended 0x20 Verbose trace messages not recommended 0x40 Messages displayed for important commands sentto the IPM Controllers during their initialization not recommended 0x80 Verbose messages about acquiring and releasing internal locks not recommended The default debug level is 0x07 enabling error warning and informational messages 3 3 4 Verbosity Console Level Description This parameter controls the verbosity of output sent by the Shelf Manager to the console The VERBOSITY_CONSOLE configuration parameter is a hexadecimal bit mask with each bit enabling output of a specific type of message according to Table 9 If this parameter is omitted the value of the VERBOSITY configuration parameter is used to control the verbosity of output to both the system log and the console 3 4 Setting Up Ethernet The ShMM uses two Ethernet ports one of them being used for shelf external access Since RMCP is the only shelf external interface that is required by ATCA the shelf external Ethe
294. nfigures the network interface Specified by the RMCP_NET_ADAPTER variable using that IP address but with the least significant bit inverted Release 3 2 0 60 July 31 2013 REAPPLY_POWER _ MAX COOLING_ST ATE Pigeon Point Shelf Manager User Guide String 16 DEFAULT NORMAL DESCRIPTION Sets the most severe shelf thermal State in which the Shelf Manager can power a FRU up after that FRU has been powered down due to a Critical thermal alert Possible values of this variable are NORMAL MINOR_ALERT MAJOR_ALERT CRITICAL_ALERT CAN BE OBTAINED FROM SHELF FRU INFO No REDUNDANCY_COM PRESSION_THRES HOLD Number This parameter controls compression of messages sent over the redundancy interface A message is sent in compressed format if its size exceeds the parameter value The value 1 turns compression off Using message compression can improve the speed of redundancy information transfer on relatively slow redundancy links like the serial interface used on ShMM 1500 No REDUNDANCY_ENA BLED Boolean TRUE Run the Shelf Manager in redundant mode No REDUNDANCY_NET _ADAPTER String 16 un undefined The name of the network adapter used for communication between redundant instances of the Shelf Manager usbO is the recommended value if USB network interfaces are used for redundancy No REDUNDANCY_NET _ADAPTER2 St
295. nformation of its subsidiary FR Us If these data are found for a specific FRU they are used to substitute definitions for that FRU from the carrier or chassis HPDL data and or SDRs Release 3 2 0 51 J uly 31 2013 HPDL_NETWORK_E LEMENT_ID_EEPR OM_OFFSET Pigeon Point Shelf Manager User Guide TYPE Number DEFAULT DESCRIPTION If non zero specifies the offset to the Network Element ID data structure in Shelf FRU Information EEPROMs This 33 byte data structure consists of the three separate 11 byte Network Element IDs and is transparent for the Shelf Manager but can be read and modified with CLI commands A positive value of this parameter indicates the offset from the beginning of the EEPROM a negative value means the offset from the end of the EEPROM Zero value indicates the absence of the Network Element ID This parameter makes sense only for HPDL based systems and only if local EEPROMs are used to store Shelf FRU Information CAN BE OBTAINED FROM SHELF FRU INFO Yes IGNORE_FAILED_ DIRECTED_POWER _DOWN Boolean TRUE This parameter tells the Shelf Manager to ignore board power down failure s during a Critical Alert and not to power down all boards in the Shelf in this case This parameter also instructs the Shelf Manager to ignore board power level decrease failure s during a Major Alert and not to decrease power level for all boards in the shelf in
296. nines are available with 1 32 Mbytes of Flash and 64 Mbytes of SDRAM or 2 64 Mbytes of Flash and 128 Mbytes of SDRAM All ShMM 700Rs have 64 Mbytes of Flash and 128 Mbytes of SDRAM The following table shows the main characteristics of the key ShMM variants Table 3 ShMM 500R ShMM 1500R and ShMM 700R Features and Variants FEATURE SHMM 500R SHMM 1500R SHMM 700R CPU NetLogic Au1550 Freescale MPC8343__ Freescale i MX28 Processor core s 333 MHz MIPS 32 250 MHz PowerPC 297 MHz ARM9 SDRAM 64 or 128 Mbytes 128 Mbytes with EEC 128 Mbytes Flash 161 32 or 64 Mbytes 32 or 64 Mbytes 64 Mbytes Ethernet Dual 10 100 Mbit Dual 10 100 Mbit Dual 10 100 Mbit Serial Two one with modem Two one with modem Three one with controls controls modem controls and one with flow control optionally battery backed on ShMM Universal Serial Bus Hostand device ports No Host and device ports USB PCI interface to No Yes No carrier devices Duplex IP MB 0 Yes Yes Yes ATCA watchdog timer Yes Yes Yes Real time clock Yes Yes Yes 1 For new shipments after about August 2008 the low end ShMM 500R model has 32 Mbytes of Flash Previously the low end model had 16 Mbytes of Flash Release 3 2 0 19 J uly 31 2013 Pigeon Point Shelf Manager User Guide FEATURE SHMM 500R SHMM 1500R SHMM 700R carrier General Purpose I O Nine Nine Ten Signals Shelf Manager Yes via CPLD Yes
297. ns in the Linux kernel to capture IPMB traffic on IPMB 0 Specifically the promiscuous mode of 12C adapter operation is used to capture all messages on the IP MB Promiscuous mode configures the adapter to receive all IPMB messages on the bus not just the messages that are addressed to or issued by the ShMM itself However promiscuous mode is supported only on the ShMM 1500 On the ShMM 500 and ShMM 700 due to hardware limitations adapters can only receive messages that are issued by or directed to a ShMM itself so promiscuous mode does not have its intended effect In addition to IP MB 0 traffic the daemon can also collect IPMI messaging traces from boards or modules in the shelf via RMCP sessions with the management controllers on those boards or modules To begin collecting a trace the daemon establishes an RMCP session with the target board using a special type of payload the IP MI Trace Payload that is defined in the HPM 2 LAN Attached IPM Controller Specification Once the RMCP session with the IP MI Trace Payload is established the IPM Controller begins sending trace data to the daemon in the IP MI Trace Payload format and continues to do so until the session is terminated A specific instance of the daemon can collect either local or remote IPMI messaging trace but not both Also when collecting a remote trace it can do so only from one IPM Controller in the shelf To collect a remote trace from each of multiple IPM Co
298. nsor definition if only a subset of the attributes of the sensor need to be redefined see below The PPS supplied SDR compiler utility can be used to produce the binary SDRs from plain text human readable text files The SDR compiler can also decode binary SDR data and produce a human readable text file from it This utility is described in the SDR Compiler User Guide The current version of the SDR compiler as distributed with release 2 2 or later of the Shelf Manager must be used In order to take advantage of this sensor configuration facility you should install the SDR Compiler on either a Linux ora MS Windows system Then you should create an SDR definition text file according to the format described in a SDR Compiler User Guide You can use standard text editors such as vi for Linux or Notepad for MS Windows The standard extension for SDR definition text files is in The SDR Compiler is a command line utility To produce a binary SDR definition file usex_sdr from a text SDR definition file you should use the utility in the compilation mode For example gt python sdrce py test inf user_sdr user_sdr s an optional name of an output file If it is not indicated in the command line a binary file test bin will be created by the SDR Compiler The SDR compiler is written in Python so you also need a Python interpreter version 2 3 or later to run it Python is available for free downloading with support for both Windows and Linux
299. nt Board Management Starter Kits for each of these BMR variants include all the materials necessary documentation schematics bill of materials firmware source code and development tools etc for Intelligent FRU developers to integrate a reference design directly into their boards and take immediate advantage of the fully validated BMR firmware More details including product briefs on the available Pigeon Point BMR variants and corresponding Starter Kits are available at http www pigeonpoint com products html 2 4 Pigeon Point Shelf Manager and ShMM The Pigeon Point Shelf Manager consistent with AdvancedTCA Shelf Manager requirements has two main responsibilities e Manage track the FRU population and common infrastructure of a shelf especially the power cooling and interconnect resources and their usage Within the shelf this management tracking primarily occurs through interactions between the Shelf Manager and the IPM Controllers over IPMB 0 e Enable the overall System Manager to join in that management tracking through the System Manager Interface which is typically implemented over Ethernet Much of the Pigeon Point Shelf Manager software is devoted to routine missions such as powering a shelf up or down and handling the arrival or departure of FRUs including negotiating assignments of power and interconnect resources In addition the Shelf Manager can take direct action when exceptions are raised in the shelf For instan
300. ntext 7 First check to see if you are interacting with the active Shelf Manager You only need to make changes on the active Shelf Manager as it updates the backup with the network configuration changes via the redundancy interface Use the hwri command and look for Active in the output of the command see the underlined fragment of the output below The hwri command reports the current state of the hardware redundancy signals on the ShMM exposed through the CPLD on the ShMM 500 and through the FPGA on the ShMM 1500 and ShMM 700 The flags reported by the hwri command are identical for all Release 3 2 0 88 July 31 2013 Pigeon Point Shelf Manager User Guide three variants though the actual register interface and bit layout is different among those variants If you are not using the active ShMM connect to the other ShMM device and repeat Step 7 hwri HWRI word 3307 1000h Local Presence 0100h Remote Presenc 0002h Local Healthy 0200h Remote Healthy 0004h Local Switchover Request 2000h Active 8 Getthe current IP settings for channel 1 clia getlanconfig 1 Pigeon Point Shelf Manager Command Line Interpreter Authentication Type Support 0x15 None MD5 Straight Password Key Authentication Type Enables Callback level 0x00 User level 0x15 None MD5 Straight Password Key Operator level 0x15 None MD5 Straight Password Key Administrato
301. ntrollers multiple instance of the daemon can be started on the ShMM When collecting a local or remote trace the trace collection daemon runs in either the unattended trace or controlled trace modes When initiated in unattended trace mode the daemon initiates trace collection and starts writing collected data into a file This continues until the daemon is terminated by the user via c or the SIGINT signal is sent to the daemon process operating as a background Linux application The resulting trace file can then be transferred to another host for analysis using the IPMB trace analyzer When started in controlled trace mode the daemon begins to listen for an incoming TCP IP connection from the IPMB analyzer running on another host Once the connection is established the daemon initiates trace collection and starts transferring collected data to the analyzer over TCP IP This continues until the GUI terminates the connection The daemon can be terminated by the user via C or the SIGINT signal sent to the daemon process operating as a background Linux application The daemon command line has the following syntax ipmb_traced OPTIONS The options are described in the following table Release 3 2 0 170 July 31 2013 Pigeon Point Shelf Manager User Guide Table 15 Options for the IPMB Analyzer Trace Collection Daemon OPTION NAME DESCRIPTION u lt file_name gt unattended lt file name gt Specify a file for collecting trac
302. on lt fru id gt is the FRU device ID of that location lt file gt is the name of the file on the ShMM that contains the new Shelf FRU Information image The Shelf Manager must be running when this command is issued After the Shelf FRU Information is updated the shelf must be completely restarted to accommodate the changes Release 3 2 0 99 July 31 2013 Pigeon Point Shelf Manager User Guide To update the Shelf FRU Information stored in a Flash file with the contents of the file newdata bin update a single location with the IPMB address 20h and FRU device ID 1 like this clia frudataw 20 1 newdata bin To update the Shelf FRU Information stored in SEEPROMs with the contents of the file newdata bin use two commands to update locations with the IPMB address 20h and FRU device IDs 1 and 2 like this clia frudataw 20 1 newdata bin Clia frudataw 20 2 newdata bin To update the Shelf FRU Information stored on separate IPM controllers in the shelf use several usually two commands with the corresponding IPMB addresses and FRU IDs FRU ID is usually 1 in this case Here is an example of updating the Shelf FRU Information in a shelf where the Shelf FRU Information is stored on the IPMCs at 62h and 64h Clia frudataw 62 1 newdata bin clia frudataw 64 1 newdata bin This method can be also used to update any other FRU Information repository in the shelf if the IPMB address and the FRU device ID of that repository are know
303. on Point Shelf Manager User Guide lt INFO gt shmm7 700_hal c 164 lt INFO gt shmm700_hal c 186 2 3x Cleaning image var upgrade boot rfs 1 lt INFO gt shmm700_hal c 189 available ALLOW N FO gt shmm700_hal c 125 lt INFO gt shmm700_hal c 164 lt INFO gt backend c 772 8 53 2 File var upgrade boot rfs 0 of device 4 0 oned to file var upgrade boot rfs 1 of device 4 1 Write storage state available gt not Cleaning image 1 sentry shmm700 app Burning image lt INFO gt backend c 806 2377014 bytes were written from file tmp _shmm700_sentry shmm700 app to dev 8 1 lt INFO gt backend c 850 Erasing user data on candidate bank lt INFO gt backend c 860 Copy lt INFO gt shmm700_hal c 116 lt INFO gt shmm700_hal c 116 lt INFO gt shmm7 700_hal c 123 lt INFO gt shmm700_hal c 123 lt INFO gt shmm700_hal c 116 lt INFO gt shmm7 700_hal c 117 confirmed 4 ALLOW lt INFO gt shmm700_hal c 123 lt I O AG be un CO WO 10 11 T2 T3 14 NF a O gt frontend c 96 La initial PASSED die 6 53 53 ds 4 5i st before storage mount fter storage mount after sync PASSED after processing var and etc PASSE PASSED after boot select after reboot PASSE after confirm PASS 8 8 3 Example 3 This example shows an unsuccessful reliable upgrade P ower is
304. on a topic to go to it Setting Up U Boot Setting Up Shelf Manager Configuration File Setting Up Ethernet Configuring the FRU Information Configuring Carrier and Shelf Attributes using HPDL Configuring the Cooling Management Strategy Configuring Local Sensors Setting the Auxiliary Firmware Revision Setting Up the Clock Setting Up and Using ShMM Power On Self Tests Configuring External Event Handling Configuring the Platform Event Trap Format Configuring the IntegralHP Interface Configuring System Services Configuring Speed of Master Only 12C Buses ShMM 700 only 3 2 Setting Up U Boot Ona power up reset of the ShMM the hardware starts executing the U Boot firmware in Flash The firmware performs basic initialization of the ShMM and unless the user explicitly disables the Autoboot feature thus forcing the firmware to switch to the maintenance user command interface commences booting the Linux kernel Linux is booted from the kernel and root file system images residing in Flash U Boot relocates the kernel image to RAM sets up kernel parameters and passes control to the kernel entry point For ShMM 500 U Boot 1 1 4 Apr 27 2005 19 17 09 CPU Aul550 324 MHz id 0x02 rev 0x00 Board ShMM 500 S N 00 00 00 00 00 00 00 00 00 03 03 03 DRAM 64 MB Flash 16 MB Bale serial Out serial Err serial Release 3 2 0 22 July 31 2013 Pigeon Point Shelf Manager User Guide Net AulX00 ETHERNET Hi
305. on radial shelves C 3Release 2 3 0 e Section 3 2 1 corrects references to the U Boot environment variable gatewayip that was previously incorrectly referenced as gateway e Section 3 4 2 1 explains the configuration parameter INITIAL _SLOW_LINK_DELAY Introduces the parallel usage of two network interfaces In parallel mode instead of having a Single RMCP network address that is switched between the two network interfaces the Shelf Manager supports RMCP on both interfaces with different IP addresses as two separate IPMI channels channels 1 and 2 e Section 3 4 6 adds an option for IP addresses for the Shelf Manager to be assigned by a DHCP server e Section 3 3 introduces the new configuration parameters COOLING_FAN_DECREASE_TIMEOUT COOLING_FAN_INCREASE_TIMEOUT CPLD_ACTIVE_WORKAROUND Release 3 2 0 255 J uly 31 2013 Pigeon Point Shelf Manager User Guide DEFAULT_GATEWAY_IP_ADDRESS2 DEFAULT_RMCP_IP_ADDRESS2 INITIAL SLOW_LINK_DELAY FAN_LEVEL_STEP_DOWN FAN_LEVEL_STEP_UP IPMB_RETRY_TIMEOUT_MSEC MAX NODE_BUSY_RETRANSMISSIONS NORMAL STABLE_TIME PREFERRED_DHCP_SERVER SYSTEM MANAGER _TRUNCATES_SEL USE_DHCP USE_SECOND_CHANNEL Section 3 9 adds coverage for the Auxiliary Firmware Revision being set when the Shelf Manager is started It is read from the file var nvdata aux fw revision Section 3 10 1 adds an option for the Shelf Manager to obtain date and time via the Network Time Protocol Section 4 6
306. onment Variables ccccccccccccsteceeeenseeeeenieeeeeeneeeseeneeeteenieeessenieeeees 24 3 2 2 Assigning Values to Environment VariableS eeesseeeeeeeeeeieneeeinneerirneerenreerennee 30 3 2 3 Configuring U Boot Environment Variables for the Shelf Managetr 21 000 31 3 2 4 Establishing the Secondary RC Script 2 cccccccceeececseeceeseeceeeeeceeeeeeneeeeeeeessenenees 33 3 3 SETTING UP SHELF MANAGER CONFIGURATION FILE ceecceceeeeeeeeeeeeeeeseeeeeseaeeesaeeeeeeeees 34 3 3 1 Carrier specific Configuration File 2s ccccccecescceceseeeenseceeeeeceeeeessaetseneeeeeeeetsnnetees 71 3 3 2 Obtaining Configuration Variables from Shelf FRU Information c cc0ccccceeee 71 3 3 3 Verbosity Level Description 00 ceceecceeesteeeceenneeeeeenneeeceeneeeceseeecsenineeeeenineeentnnnees 73 3 3 4 Verbosity Console Level D SCLIDLION 11 cccccceceeeceeeeeceeneeceneteceeeteneeteeeeersinetees 73 3 4 SETTING UP ETHERNET assess ste ccavedccvesaaish eesunactdvachich oyealt cnsedsednnee aiaia Erraina 73 3 4 1 Usage of the First Ethernet Interface c cccccesseeeeeenseeeeeenieeeeeenneeeseeteeeseenineees 73 3 4 2 Usage of the Second Ethernet Interface ccccccceccccseeceeneeteeetecneeseeeeeeeeessenenens 75 3 4 3 Using the SAMM 500 and ShMM 700 Alternate Software Redundancy Interface 79 3 4 4 Using the SAMM 1500 Alternate Software Redundancy Interface cc0 82 3 4 5 Changing the Default ShMM Network Parameters
307. oot filesystem image in Flash This variable is set automatically by U Boot during bootstrap Note on ShMM 700 this variable contains the absolute starting address of the RFS image in RAM This variable is used in the ubifs net load_ubifs and load_ubifs_alt commands to specify the address in physical memory where the RFS image should be loaded and where itis expected to be during ShMM boot The default value is 0x46000000 rmcpaddr Default IP address for the RMCP service serverip IP address of the TFTP server set_mtdparts ShMM 700 only U Boot command that is invoked to set up the MTD configuration in U Boot to perform boot volume access The command is invoked by mount_ubifs during bootstrap start _rc2 dae mons Instructs the secondary startup script to start or not start the snmpd boa and shel fman daemons after bootup Default is y time_proto Protocol used to retrieve time from a network time server possible values are ntp and rdate time server Time server for synchronization at runtime If this variable is not specified time is extracted from the hardware clock at system startup timezone Local time zone in CCCn format where n is the offset from GMT and optionally negative while CCC identifies the time zone The default is UTC ubifs ShMM 700 only U Boot command executed to perform system boot from the Flash drive The command performs setting bootargs variable loading bo
308. or one LAN channel Parameter sets for a specific slot are ordered they are assigned to a board module in the order in which the Shelf Manager polls LAN channels of a board module see the definition of the configuration variable BOARD_LAN PARAMETERS CHANNEL LIST In general parameters sets are retrieved by the Shelf Manager dynamically on demand During initialization after the Shelf FRU Info is found and the shelf Address Table is known the Shelf Manager retrieves parameter sets for all currently active boards and modules in states M3 and M4 and performs the assignment After initialization when a new board or a module is installed in the shelf and reaches state M3 the Shelf Manager retrieves the corresponding parameter sets for itand performs the assignment This assignment is performed synchronously before powering up the payload of the corresponding board or module 4 10 2 Obtaining LAN Configuration Parameters on the Shelf Manager Level There are three ways to obtain LAN configuration parameters on the Shelf Manager level 1 Retrieved from a DHCP Server In this case when the configuration variable BOARD_LAN PARAMETERS USE_DHCP is set to TRUE the Shelf Manager retrieves parameter sets from a DHCP server The same algorithm for finding the appropriate DHCP server is used as in the case of obtaining Shelf Manager IP addresses via DHCP Release 3 2 0 159 July 31 2013 Pigeon Point Shelf Manager User Guide The queries are base
309. orporation of Pigeon Point products Release 3 2 0 12 J uly 31 2013 Pigeon Point Shelf Manager User Guide Figure 1 Management Aspects and Potential Pigeon Point Product Sites in an Example AdvancedTCA Shelf Potential Pigeon Point Product Sites ATCA AMC Specification Elements 1 Shelf Managers PP Shei Manager Running Pigeon Point E shot Management Controller shMic Shelf Manager on ShMM IPM Controllers Several Variants ShMM 500R m L ShMM Module e ShMM 1500R SS AMC Module e ShMM 700R Other Field Replaceable Unit FRU AdvancedTCA Board and Optional Rear Tr ition Module RTM 2 IPM Controllers ee Moa TIND Variants for Distinct Types of Boards and FRUs Fan Tray Running Pigeon Point 1 N BMR Firmware Optionally Implemented in SmartFusion Mixed Signal FPGA 2x Redundant Bused or Radial IPMB 0 2x Redundant Radial Internet Protocol Capable Iransport An AdvancedTCA Shelf Manager communicates inside the shelf with IPM Controllers each of which is responsible for local management of one or more Field Replaceable Units FRUs such as boards fan trays or power entry modules Management communication within a shelf occurs primarily over the Intelligent Platform Management Bus IPMB which is implemented on a dual redundant basis as IPMB 0 in AdvancedTCA The PICMG Advanced Mezzanine Card AdvancedMC or AMC specification AMC 0 defines a hot swappable mezzanine form factor desi
310. orrect 0000 20 14 cc 82 a0 2d 00 32 c0 00 bf ee ern Aree Frame 3 8 bytes on wire 8 bytes captured Arrival Time Jun 24 2008 20 58 03 818209000 Time delta from previous captured frame 0 500135000 seconds Time delta from previous displayed frame 0 500135000 seconds Time since reference or first frame 0 579642000 seconds Frame Number 3 Frame Length 8 bytes Capture Length 8 bytes Frame is marked False Protocols in frame i2c ipmi Inter Integrated Circuit Data Bus I2C 2 Target address 0x41 Flags 0x00000000 Intelligent Platform Management Interface No corresponding response Header Get Sensor Reading Request from 0x20 to 0x82 Target Address 0x82 Target LUN 0x00 NetFN Sensor Event Request 0x04 wee 00 Target LUN 0x00 0001 00 NetFn Sensor Event Request 0x04 Header checksum Ox6e correct Source Address 0x20 Source LUN 0x00 SeqNo 0x29 wee 00 Source LUN 0x00 1010 01 Sequence Number 0x29 Command Get Sensor Reading 0x2d Data Sensor Number 4 Data checksum 0x0b correct 0000 82 10 6e 20 a4 2d 04 Ob SREY gees 5 4 2 Collecting IPMI Traces on IPMB 0 in Controlled Mode When the trace daemon is run in the controlled mode the trace is transferred directly to the trace analyzer via a TCP IP connection The following stepwise procedure shows how to collect a controlled trace and open it for analysis in the GUI The GUI specific steps are followed by cor
311. orts converting a live shelf from ShMM 500 based management to ShMM 700 based management without losing manageability of the shelf This support minimizes the risk of upgrading shelves in the field to the ShMM 700 Itis important to understand the special aspects of this support e This conversion is only supported in one direction going to ShMM 700 based management e Concurrent presence in the shelf of both types of ShMMs is only supported for transition purposes Stable dual redundant management requires a pair of ShMMs from the same family e This functionality is only supported if the ShMM 500 uses a USB based Software Redundancy Interface If the ShMM 500 uses an Ethernet based Software Redundancy Interface and this functionality is desired please contact Pigeon Point Systems The switchover procedure is not much different from a standard ShMM 500 to ShMM 500 or ShMM 700 to ShMM 700 switchover procedure and is described below where ShMM 500 700 carriers are referenced for physical board removals and insertions 1 Remove the backup ShMM 500 carrier Insert a ShMM 700 carrier into the vacant slot Wait until it comes up as a backup Shelf Manager Issue a clia Switchover command on the ShMM 700 The ShMM 700 should immediately become active and the ShMM 500R will automatically reboot The ShMM 500 will keep rebooting without becoming a backup Shelf Manager and can be removed at any time 6 Remove the second ShMM 500 carrier 7
312. p tftp_info_2 TFTPSERVER 192 168 1 253 BOOTFILE tftpboot ssh_script_eth0_ShM2 The example configuration file below shows how to configure the Linux DHCP server DHCPD to provide IP addresses as well as TFTP server and bootfile information to the Shelf Manager Fixed predefined addresses are used for that purpose This file should be located as etc dhepd conf on the system hosting the DHCP server Other DHCP servers such as those on non Linux operating systems are configured differently allow booting allow bootp option domain name tst option subnet mask 255 255 255 0 option domain name servers 192 168 1 100 option ntp servers 192 168 1 50 option routers 192 168 1 253 option vendor class identifier PPS min lease time 4294967295 default lease time 4294967295 use host decl names on ddns update style ad hoc subnet 192 168 1 0 netmask 255 255 255 0 host client00 option dhcp client identifier 0 0 0 0 0 0 0 10 fixed address 192 168 1 140 option tftp server name 192 168 1 253 option bootfile name tftpboot ssh_script_ethO_ShM1 host client0Ol option dhcp client identifier 0 0 0 0 0 0 0 11 fixed address 192 168 1 141 option tftp server name 192 168 1 253 option bootfile name tftpboot ssh_script_eth1_ShM1 host client0O2 option dhcp client identifier 0 0 0 0 0 0 0 20 fixed address 192 168 1 142 option tftp server name
313. pgrade utility places the embedded U Boot kernel and RFS images into the provisional Flash bank and makes the provisional Flash bank active The utility is called with no hook option so that the default mode of copying non volatile data applies only the directory var nvdata and SSH key files are copied On ShMM 700 the reliable upgrade utility places the embedded U Boot kernel RFS images and application package into the candidate locations and makes the candidate image set active The default mode of copying non volatile data applies only the directory var nvdata and SSH key files are copied As with the reliable upgrade facility Pigeon Point recommends upgrading the backup Shelf Manager then the active Shelf Manager After the upgrade of the active Shelf Manager a Switchover occurs and the roles of the physical Shelf Managers active vs backup become reversed If itis desired to preserve the original role assignment an additional switchover is needed this can be done via the HP I Shelf Manager Failover control associated with the Virtual Shelf Manager resource Release 3 2 0 251 July 31 2013 Pigeon Point Shelf Manager User Guide See the Pigeon Point HPI User s Guide for more information and examples of the HP l based Shelf Manager upgrade procedure Release 3 2 0 252 July 31 2013 Pigeon Point Shelf Manager User Guide Appendix A Converting to ShMM 700 based Shelf Managers in ShMM 500 managed Shelves The ShMM 700 supp
314. pports leaving cold sensitive FRUs that are too cold at the time of shelf startup unpowered until they warm up This may be useful for shelves installed in severe climate conditions A special Pigeon Point OEM specific record must be present in the Shelf FRU Information to activate this feature This record contains the list of PMB addresses and FRU IDs of FRUs considered cold sensitive A cold Sensitive FRU is considered too cold if any temperature sensor associated with it generates an event at startup that indicates that the temperature is below the lower non recoverable threshold Such a FRU is not immediately powered on but kept in the state M3 until all of its temperature sensors are above their lower non recoverable thresholds then it is powered on 3 7 2 Configuring a Specific Cooling Management Strategy To configure a specific cooling management strategy the following steps need to be taken 1 2 2 2 3 Create a new cooling management strategy module shared library or choose one of the several libraries supplied with the Shelf Manager in the RFS directory lib that are based on cooling strategies from certain vendors If a new module is created perform the following steps Implement the specific cooling management functionality using the Shelf Manager Cooling Management AP I which is described in a separate document Choose a name for the shared library that conforms to the following pattern libcooling_ lt xxx gt so whe
315. print buffer at 1003c910 Pthread lock log print buffer at 10040940 17 11 28 027 522 Custom SDR initialization file is absent important to understand that replacement SDRs must be closely coordinated with SDRs that are defined within the Shelf Manager Therefore additions or modifications to the set of replacement SDRs in a shelf should only be undertaken in close cooperation with the shelf supplier The following rules apply to SDRs that are used to configure local sensors referenced as replacement SDRs below Every replacement SDR must be a Full Sensor Record type 01 Compact Sensor Records type 02 are not supported and are never used by the Shelf Manager for local sensors The following fields are mandatory in every replacement SDR note that this list is smaller than the list of mandatory fields for normal SDRs processed by the SDR compiler e Sensor Owner ID e Sensor Number e Sensor Initialization If a field from a replacement SDR is used as a result of the operation of other rules to replace an attribute for a target sensor and the field is not specified in the replacement SDR a value zero 0 is used for this field The tuple Sensor Owner ID Sensor Number identifies the sensor that is to be configured The Sensor Owner ID is a literal IPMB address so for sensors on the physical IPM controller two instances of the corresponding replacement SDR must be present one each for the IPMB addres
316. propriate instance of the application package and unpacks it into certain directories Release 3 2 0 220 July 31 2013 Pigeon Point Shelf Manager User Guide in the root RAMFS file system thus creating application binaries in the RAMFS file system on the fly Also there are two instances of writable directories etc and var on the ShMM 700 so that each of the two image sets has its own instance of these directories These directories contain writable application data When the corresponding image set is loaded the corresponding instance of etc and var directory is found on the Flash and mounted as etc and var in the current file tree When a stable firmware configuration is established in one of these image sets it is designated as the confirmed image set When a new firmware configuration is installed it goes in the other image set which is initially designated candidate Once a new firmware configuration in the candidate set is validated that image set is designated the confirmed image set and continues to be uses until a future upgrade cycle starts the process over again ShMM reliable upgrades are enabled by hardware mechanisms which ensure that if defective firmware is present in the candidate image set and the candidate image set fails to boot or initialize an automatic rollback takes place to the safe software copy in the confirmed image set Even if the candidate image set boots and initializes successfully but embedded soft
317. provides a comprehensive set of textual commands that can be issued to the Shelf Manager via either a physical serial connection or a telnet connection Release 3 2 0 18 July 31 2013 Pigeon Point Shelf Manager User Guide e Web based Interface This interface enables essentially the same functionality as the CLI with access to the Shelf Manager via a web browser Finally the Pigeon Point Shelf Manager can optionally include an IntegralHP subsystem which provides access to the Shelf Manager via the Hardware P latform Interface HPI IntegralHP operates within the Shelf Manager fully leveraging the Shelf Manager s facilities for managing the elements and events in the shelf IntegralHP also takes advantage of the mature redundancy framework of the Shelf Manager to deliver a fully redundant HP service Using these mechanisms the System Manager can access information about the current state of the shelf including current FRU population sensor values threshold settings recent events and overall shelf health These aspects of ATCA s System Manager Interface are considered to be the Pigeon Point shelf external interfaces They are documented separately in the Shelf Manager External Interface Reference 2 4 4 Pigeon Point ShMM Shelf Management Mezzanines The Pigeon Point Shelf Manager executes on the ShMM a small Shelf Management Mezzanine with ShMM 500R ShMM 1500R and ShMM 700R variants SOMM 500R and ShMM 1500R mezza
318. q IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req IPMI ATC Req TCP 192 168 1 65 lt live capture in pro Packets 262 Displayed 262 Ma Profile Default Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID Get Device ID QF wf pS seq Ox3e seq Ox3e seq 0x3f seq 0x3f seq 0x00 seq 0x00 seq 0x01 seq 0x01 seq 0x02 seq 0x02 seq 0x03 seq 0x03 seq 0x04 seq 0x04 seq 0x05 seq 0x05 seq 0x06 seq 0x06 seq 0x07 MB m 8 4 To stop the capture session go to the Capture submenu and select Stop In a command line context redirect the output of net cat to the tshark utility instead of the Wireshark GUI application Like Wireshark tshark can take the standard input option to indicate the source of the data Specify the IP address of the ShMM and the port on which the trace daemon listens for incoming connections 19000 by default as input to net cat Options x and V as in the case of unattended mode can be used to request a hexadecimal and or detailed dumps of the messages Any additional options needed for filtering and analysis of the message stream see section 5 4 6 should also be specified at this time Next execute the actions from step 5 the trace is then cap
319. r MAX_EVENT_SUBS Number 60 The maximum timeout for an event Yes A el subscriber in seconds between the moment when an event arrives and the moment when the subscriber retrieves this event from the Shelf Manager If this timeout is exceeded the subscriber is considered dead and is automatically unregistered MAX_INCOMING_I Number 128 The size of the internal Shelf Manager No EME REQUNS TS queue for incoming IPMB requests Incoming IPMB requests are stored in this queue before processing MAX_NODE_BUSY_ Number 255 The maximum number of Yes TRANSMISSIONS transmissions of an IPMB command if the receiver always returns the completion code Node Busy in response MAX_OEM_FILTER Number 16 The maximum number of PEF OEM No S event filters available MAX_PENDING_EV Number 1024 The maximum number of outstanding No event notifications for each active subscriber MAX_PENDING_IP Number 192 The maximum number of pending No MB_REQUESTS IPMB requests awaiting response MAX_SEL_ENTRIE Number 1024 The maximum number of entries in No S the System Event Log CAUTION Large values of this parameter can significantly degrade Shelf Manager performance MAX_SESSIONS Number 32 The maximum number of No simultaneous IPMI sessions The maximum allowed value for this parameter is 64 MAX_USERS Number 32 The maximum number of IPMI users No The maximum allowed value for this parameter is 64 Release 3 2 0 58 J uly 31 2013 Pigeon Point Shel
320. r 84 lt I gt 14 35 22 693 199 Operational state for SA 84 FRU 0 is set to M7 from stored SDR lt I gt 14 35 22 701 199 SDR Repository broadcasting Get Device ID lt I gt 14 35 22 763 224 CLI subscribing for deadlocks pid 224 lt I gt 14 35 22 784 224 Registering 224 facility to the triggering watchdog lt I gt 14 35 22 806 225 FRU data write thread started successfully lt I gt 14 35 22 834 199 Generating list of the trusted Shelf FRU Information sources lt I gt 14 35 22 836 199 FRU 1 on 0x20 IPMC was added to the Shelf FRU trusted list lt I gt 14 35 22 842 205 Creating new SEL Journal lt I gt 14 35 22 850 199 Controller 20 FRU 0 ATCA state set to M1 prev M0 cause 0 locked 0 lt I gt 14 35 22 852 199 Before_init_fru_hs_state lt I gt 14 35 22 854 199 set the FRU 1 initial state Release 3 2 0 132 July 31 2013 Pigeon Point Shelf Manager User Guide lt I gt 14 35 22 856 199 Controller 20 FRU 1 ATCA state set to M1 prev M0 cause 0 locked 0 lt I gt 14 35 22 859 199 Registering shelf FRU notification 0x4eb18c lt I gt AAS 35222 2861 199 Registering shelf FRU notification O0x2ab44fdc lt I gt 14 35 22 863 199 Shelfman Running lt I gt 14 35 22 865 226 Shelf FRU registering for SDR Repository notifications lt I gt 14 35422 873 226 Added 20 1 to the fru list treated as Shelf FRU In
321. r both Shelf Managers A total of 6 IP addresses can be assigned via DHCP A specific IP address is designated by a particular value of the Client Identifier that is passed from the Shelf Manager to the DHCP server In the default implementation the Client Identifier is based on the Shelf Address string that is Stored in the Shelf FRU Info plus the Request Identifier Request ID byte at the end The Request ID byte has the following format e Request ID bits 7 4 Shelf Manager number 0 1 2 with 0 for logical Shelf Manager e RequestID bits 3 0 Interface number 0 for eth 1 for eth1 If the Shelf Address string in the Shelf FRU Info is empty a hard coded string of 7 zero bytes is used In that case the Client Identifier values are implemented as follows Currently the Client Identifier values are hardcoded as follows 00 00 00 00 00 00 00 10 Shelf Manager 1 eth0 00 00 00 00 00 00 00 11 Shelf Manager 1 ethl 00 00 00 00 00 00 00 20 Shelf Manager 2 eth0 00 00 00 00 00 00 00 21 Shelf Manager 2 ethl 00 00 00 00 00 00 00 00 Logical Shelf Manager eth0 00 00 00 00 00 00 00 01 Logical Shelf Manager ethl However these Client IDs can be redefined for a specific carrier inside the corresponding ShMM carrier specific module To use this feature in the Shelf Manager itis necessary to define the configuration parameter USE_DHCP in the Shelf Manager configuration file etc shelfman conf as follows U
322. r can revert to the original images if he she detects that the new images are not acceptable for any reason To do this the user invokes rupgrade rollback This sequence can be performed manually by an operator However if necessary one can easily automate the above sequence via a script that runs on a host machine that is connected to the ShMM over the network and remotely issues appropriate commands to the rupgrade and backend utilities on the ShMM 8 8 Reliable Upgrade Examples This section provides reliable upgrade examples for ShMM 700 8 8 1 Example 1 This example shows a reliable upgrade of all four components U Boot kernel RFS image and application package copying ete and var nvdata non volatile directories to the candidate locations All images are taken from the local tmp which implies that they have Release 3 2 0 233 July 31 2013 Pigeon Point Shelf Manager User Guide already been copied there in some unspecified way The U Boot image is taken from tmp u boot bin the kernel image is taken from tmp sentry kernel the RFS image is taken from tmp sentry rfs the application image is taken from tmp sentry app tgz The upgrade procedure is started from the serial console Comments are interspersed in the console log to provide additional background on the steps of the upgrade procedure First the rupgrade utility is started from the command prompt in upgrade mode The parameters show that all four images a
323. r carriers that have GP 10 control of that multiplexer C 11 Release 2 6 0 Entire document adds coverage of ShMM 1500R Sections 3 1 3 10 1 changes the default time zone name from UTCO to UTC Section 3 3 adds new configuration parameters ACTIVATE LOCAL _WITHOUT_SHELF_FRU ATCA_TESTER_COMPATIBILITY DEFAULT_VLAN_ID DEFAULT_VLAN_ID2 DHCP_FOR_RMCP_ONLY TURBO_MODE_MIN FAN FAILURES New section 3 4 6 describes assigning VLAN IDs in accordance with version 2 0 of the IP MI Specification Section 3 12 1 in the PEF configuration example corrects control 7 tO control 1 New section 4 5 1 describes operation in ShMM 500 based Shelves with radial IPMB 0 New section 4 5 2 describes operation in ShMM 1500 based Shelves with radial IPMB 0 New section 5 introduces the use of the IPMI analysis tools Section 7 3 adds the tables describing the Flash partitions fora SHMM 1500 with 32 MB and 64MB Flash devices Section 7 6 adds descriptions of new options and specifiers for the reliable upgrade utility a R C Section 7 8 2 adds an example reliable upgrade scenario for the SaMM 1500 C 12 Release 2 6 1 Section 3 2 1 adds descriptions of new U Boot variables log_max and log_remote Section 3 3 adds new configuration parameters ENABLE_INTEGRALHPI INTEGRALHPI_DOMAIN_ID TACHOMETER_THRESHOLD_UPDATE_DELAY Section 3 3 corrects the description of the configuration parameter REDUNDANCY_NET_ADAPTER2
324. r level 0x15 None MD5 Straight Password Key OEM level 0x00 IP Address 206 25 139 28 IP Address Source Static Address Manually Configured 0x01 MAC Address 00 50 c2 22 50 30 Subnet Mask 0 0 0 0 IPv4 Header Parameters 0x40 0x40 0x10 Primary RMCP Port Number 0x026f S B econdary RMCP Port Number 0x0298 C generated ARP Control 0x02 Enable BMC generated ARP Response Gratuitous ARP Interval 2 0 seconds Default Gateway Address 206 25 139 3 Default Gateway MAC Address 00 00 00 00 00 00 Backup Gateway Address 0 0 0 0 Backup Gateway MAC Address N A Community String public Number of Destinations 16 Destination Type N A Destination Address N A 802 1q VLAN ID 0 disabled VLAN priority 0 Cipher Suite Entry count 15 Supported Cipher Suite IDs Oh 1h 2h 3h 4h 5h 6h 7h 8h 9h Ah Bh Ch Dh Eh Cipher Suite Privilege Levels ID 00h Priv Level User 2 ID Olh Priv Level User 2 Release 3 2 0 89 July 31 2013 Pigeon Point Shelf Manager User Guide Administrator OEM Proprietary OEM Proprietary OEM Proprietary OEM Proprietary ID 02h Priv Level Proprietary 5 ID 04h Priv Level Proprietary 5 ID 06h Priv Level User Administrator 4 ID 08h Priv Level Proprietary 5 ID OAh Priv Level Administrator 4 ID OCh Priv Level Proprietary 5 ID OEh Priv Level
325. rate Q 2 ccccccccscsesceessseesinees 108 3 8 CONFIGURING LOCAL SENSORS sccccccceessstecececececseseeseceeeeeceeseseeaeceseeseeeseaeaeceeeeeeseees 109 3 9 SETTING THE AUXILIARY FIRMWARE REVISION 0cccecceececsesseaececeeseeesseaeceeeeeeseees 114 Release 3 2 0 3 July 31 2013 Pigeon Point Shelf Manager User Guide 3 10 SETTING UP THE CLOCK 5 secrete ceased cctv ths tence peccenengea dees estebba Mee canctaseletadeevodsdeatseintserees 114 3 10 1 Obtaining Date and Time from a Time S rvel cccccceceeececeseeteneceeeessnneeteaes 115 3 11 SETTING UP AND USING SHMM POWER ON SELF TESTS 0cssesececeeeceesssseaeeeeeeeeeeees 116 3 12 CONFIGURING EXTERNAL EVENT HANDLING cccccceeseseecesececeeseseeaececeesesesssaeaeceseeseeeeas 118 3 12 1 Detailed Steps to Configure External Event HandlinG ccccccccssceesseteseeesees 118 3 12 2 External Event Handler Operation 2c cccccccecseceeseeceeeeecenecessneseneeseeeeessnneetiaes 120 3 13 CONFIGURING THE PLATFORM EVENT TRAP FORMAT cscccccessessaeceseeeceessseaeeeeeeeesenses 121 3 13 1 Parsed Example of SNMP Trap 2 ccccsceceeceeceeeeeenseceeeeecenneessneseeesseneecsnneesaes 122 3 14 CONFIGURING THE INTEGRALHPI INTERFACE 0ssecccecceeceesesneaececeeseeesssaeaeceeeeeeseees 123 3 14 1 HPI Domain Support in Int QralHPl 21 ccccccccecceceeeceeneececeeecseneesneeeseseecsnneesaes 123 3 14 2 HPI SNMP Subagent Support ccccc
326. ration variables SHELF_FRU_IPMB SOURCE1 and SHELF_FRU_IPMB_SOURCE2 must be defined in the configuration file etc shelfman conf These variables are of the Number type and contain the IP MB addresses of the two designated IPM controllers that represent Shelf FRU Information For example to limit the search for the Shelf FRU Information to IPM controllers at 66h and 68h respectively these variables should be defined as follows note the use of Ox prefix for hexadecimal addresses SHELF_FRU_IPMB_SOURCE1 SHELF_FRU_IPMB_SOURCE2 0x66 0x68 However most known ShMM based shelves provide two SEEPROMs that are connected to the Shelf Manager via the master only I2C interface usually with each SEEPROM residing on its own bus behind an 12C multiplexer Each of these two SEEPROMs stores a copy of the Shelf FRU Information providing the needed redundancy Some shelves do not provide even this type of storage in that case itis possible to Store the Shelf FRU Information on the ShMM itself as a single Flash file var nvdata shelf_fru_info The redundant Shelf Managers each have their own copy of that Flash file and synchronize them using the redundancy protocol so even in that case some degree of redundancy is still preserved The Shelf Manager configuration as represented in shel fman conE must be aligned with the mechanisms for accessing Shelf FRU Information that are provided by the shelf The default Release 3 2
327. re lt xxx gt is the name of the strategy typically the vendor name Place the shared library either in the directory 1ib or in the directory var bin on the ShMM Set the value of the configuration variable COOLING_MANAGEMENT in etc shelfman conf to the name of the strategy lt xxx gt above and restart the Shelf Manager thereafter whenever it starts the Shelf Manager loads and uses the designated shared library for cooling management Release 3 2 0 108 July 31 2013 Pigeon Point Shelf Manager User Guide 3 8 Configuring Local Sensors Note This section applies only to shelves where HPDL is not used With HPDL sensor SDRs are defined as part of the carrier and chassis definition and do not require additional configuration from a file on the ShMM Local sensors on the ShMM can be configured when the ShMM is started This capability applies to sensors that are associated with either 1 the Shelf Manager or 2 the physical IPM controller that takes its IPMB 0 address from the hardware address of the ShMM carrier slot Only sensor attributes that are defined in Sensor Data Records such as thresholds hysteresis values sensor name linearization parameters etc can be configured at this time The Sensor Device Records SDRs defining these sensors are read from the file var nvdata user_sdr This file must contain an array of binary SDRs that are compliant with the IPMI specification However these SDRs can contain only partial se
328. re is a trace collection dialog trace filter window and others These facilities are described in detail in the Wireshark user documentation and man pages Section 5 4 later in this document provides a step by step procedure to use selected GUI elements for tracing and analyzing IPMB data When analyzing RMCP or RMCP based IPMI network traffic the main GUI window looks similar but with some small differences as shown in Figure 7 below Figure 7 Main Window IPMI over network trace analysis ethO Capturing Wireshark O0 X File Edit View Go Capture Analyze Statistics Help as a A S M 47 ER aaan nme g Filter rmcp j 7 p Expression Clear Apply gt Time Source Destination Protocol Info z e 5 5 5 X vvyvvvvy Frame 33 65 bytes on wire 65 bytes captured Ethernet II Src UniwillC_2e 23 71 00 03 0d 2e 23 71 Dst Cisco_9a c7 f0 00 16 c7 9a c7 f0 Internet Protocol Src 80 240 102 57 80 240 102 57 Dst 192 168 1 65 192 168 1 65 User Datagram Protocol Src Port 50122 50122 Dst Port asf rmcp 623 Remote Management Control Protocol Class IPMI IPMI v1 5 Session Wrapper session ID 0x0 Intelligent Platform Management Interface 0000 00 16 c7 9a c7 fO 00 03 Od 2e 23 71 08 00 45 00 q E 0010 00 33 00 00 40 00 40 11 cl a7 50 fO 66 39 cO a8 3 P f9 0020 01 41 c3 ca 02 6f 00 1f 97 73 06 00 ff 07 00 00 y E AS PEER 0030 00 00 00 00 00 00 00 O9 20
329. re to be updated with the Shelf Manager non volatile data and configuration file preserved as well rupgrade skip checksums copy etc k file tmp sentry kernel r file tmp sentry rfs u file tmp u boot bin a file tmp sentry app tgz The upgrade utility attempts to invoke a validation script to check the images in tmp currently Supplied If any of the specified file designators is not found in tmp the utility stops and a message like the following is produced lt INFO gt frontend c 449 Downloading file file tmp u boot bin lt ERR gt download_easy c 51 Failed to download file file tmp u boot bin Error The utility proceeds to copy the images to the specified destinations in candidate Flash lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend lt INFO gt frontend 449 Downloading file file tmp u boot bin 449 Downloading file file tmp sentry kernel 449 Downloading file file tmp sentry rfs 9 5 5 oo Oo 449 Downloading file file tmp sentry app 753 Initiating partial upgrade 73 uboot tmp _tmp_uboot 745 kernel tmp _tmp_kernel 755 rfs tmp _tmp_rfs lt INFO gt frontend c 765 app tmp _tmp_app lt INFO gt frontend c 820 Calling backend to handle partial upgrade sbin backend uboot tmp _tmp_uboot kerne
330. ree values can be used a2f upgrade dat in the case of a normal A2F upgrade a2 auto rollback dat in the case of an A2F upgrade rollback or empty if no A2F upgrade is in progress Note U Boot expects the a2 auto rollback dat file to be Stored on the boot volume and to contain an image of the currently running firmware The image is used to rollback an A2F upgrade if new candidate firmware is not programmed successfully or fails any of the post upgrade tests addmisc Appends quiet and console Settings to bootargs This variable is normally not modified Note not present on ShMM 700 Release 3 2 0 24 July 31 2013 ENVIRONMENT VARIABLE alt_image Pigeon Point Shelf Manager User Guide DESCRIPTION ShMM 700 only This variable contains the rollback image set index This variable is used by load_ubifs command to determine which of the Linux kernel and RFS images to boot if the current image set is broken and there is no reliable upgrade running The variable is set automatically by U Boot during bootstrap and should not be changed by the user its value is always opposite to image_sel baudrate Serial port baud rate default is 115200 bootargs Command line to be passed to the Linux kernel May contain references to other U Boot environment variables which is resolved at run time On both ShMM 500 and ShMM 1500 the default value is root dev ram rw console ttyS0 115200 On ShMM 700 the defa
331. responding guidance for collecting a controlled trace int shark 1 Onthe ShMM start ipmb traced with the c and w flags daemon f ipmb_traced c w Release 3 2 0 181 July 31 2013 Pigeon Point Shelf Manager User Guide On the analyzer host use net cat to feed the data to Wireshark over the TCP IP network specify the IP address of the ShMM and the port on which the trace daemon listens for incoming connections 19000 by default C gt ne exe 192 168 1 62 19000 c Program Files Wireshark wireshark exe k i The k option instructs Wireshark to start capturing data immediately The standard input is Specified as the interface that provides the trace data To avoid repeatedly typing the long command a Windows batch script can be created instead rtwireshark bat C distrib nc exe 1 19000 ProgramFiles Wireshark wireshark exe k i amp exit Given that this script is defined invoke it from the command prompt or create a shortcut on the desktop and specify the IP address as an input parameter Figure 10 Capture Options Window rtwireshark Properties General Shortcut Options Font Layout Colors Compatibility ey ttwireshark a Target type MS DOS Batch File Target location distrib Target C distrib rtwireshark bat 192 168 1 62 Start in C distrib Shortcut key None Bun Normal window Comment Eind Target Change Icon 2 Execute actions that cr
332. ress is assigned to the first network interface and the ShMM is not accessible over the network until the Shelf Manager starts and the RMCP IP address is assigned However it may be useful in some cases to assign an IP address to the RMCP network interface and have the ShMM accessible over the network as soon as the operating system is booted In that case it is also desirable that when the Shelf Manager is started for the RMCP IP address to coexist with the originally assigned IP address rather than replacing it To achieve this configuration it is necessary to instruct the Shelf Manager to assign the RMCP IP address not to the first network adapter itself ethO but to its first alias ethO 1 The initial IP address is assigned in that case to the network adapter itself et hO during the start of the operating system This initial assignment happens in the initialization script etc rxe itis accomplished by 1 enabling the U Boot variable re_ifconfig setting it to y 2 assigning the original IP address to the U Boot variable ipaddr for example shmmx500 setenv rce_ifconfig shmmx500 setenv ipaddr 192 168 1 240 and 3 changing the value of the Shelf Manager configuration parameter RMCP_NET ADAPTER to ethO 1 In a redundant configuration the U Boot variable ipaddr is allowed to have the same value on both ShMMs The actual initial IP address assigned to each of the two redundant ShMMs is based on the value of ipaddr but is modified d
333. ressed Data Size 844843 Bytes 825 kB Load Address 80100000 Entry Point 802bc040 Verifying Checksum OK Uncompressing Kernel Image OK Loading Ramdisk Image at bfc40000 Image Name sentry RFS Ramdisk Image Created 2005 06 27 10 51 03 UTC Image Type MIPS Linux RAMDisk Image gzip compressed Data Size 2465924 Bytes 2 4 MB Load Address 00000000 Release 3 2 0 212 July 31 2013 Pigeon Point Shelf Manager User Guide Entry Point 00000000 Verifying Checksum OK Starting kernel init started BusyBox v0 60 5 2005 06 15 14 45 0000 multi call binary etc re Mounted proc etc rce Mounting filesystems etc rce Mounted dev pts etc rc Mounted dev mtdblock0O to var etc rce Mounted dev mtdblock10 to var upgrade At this point in the execution of the re script itinvokes rupgrade_tool c to check whether a reliable upgrade is in progress The tool returns 0 confirming that an upgrade is in progress Given that result the re script continues with the startup process etc rc Checking the reliable upgrade watchdog timer activated etc rc Mounted ram disk to var log etc re Started syslogd and klogd etc rc Mounted ram disk to var tmp etc rce Setting hostname shmm 193 Since a reliable upgrade is in progress the watchdog timer is strobed once more in the re script etc rce Strobing the reliable upgrade watchdog timer etc rc Mounted dev mtdblockl to etc et
334. rface Reference for the detailed information about these CLI commands In addition to providing an IP address related information a DHCP server can be configured via the TFTP Server Name and Bootfile Name options to provide the name of a TFTP server and identify a bootfile on that server For the benefit of other applications that may run on the ShMM in parallel with the Shelf Manager the DHCP client if it receives this additional information stores it in the ShMM file system A separate file is used for each of the 6 network interfaces processed by the DHCP client according to the following rules tmp tftp_info_0 for the first network interface on the ShMM with the lower Hardware Address tmp tftp_info_1 for the second network interface on the ShMM with the lower Hardware Address tmp tftp_info_2 for the first network interface on the ShMM with the higher Hardware Address tmp tftp_info_3 for the second network interface on the ShMM with the higher Hardware Address tmp tftp_info_4 for the first RMCP accessible interface tmp tftp_info_5 for the second RMCP accessible interface The first four files above are stored on the corresponding ShMM only the last two are stored on both redundant ShMMs Each file consists of two lines The first line represents the TFTP server name and the second line represents the boot file name for example Release 3 2 0 93 July 31 2013 Pigeon Point Shelf Manager User Guide cat tm
335. ring 16 un undefined The name of the second network adapter used for communication between redundant instances of the Shelf Manager if the dual USB network interface is used for this purpose usb1 is the recommended value if USB network interfaces are used for redundancy If the configuration variable REDUNDANCY_NET_ADAPTER is setto usbO this variable defaults to usb1 No Release 3 2 0 61 July 31 2013 REDUNDANCY_NET MASK Pigeon Point Shelf Manager User Guide TYPE IP address DEFAULT 0 0 0 0 DESCRIPTION The netmask to assign to the redundancy IP address by default if 0 the netmask is determined automatically from the class of the IP address 255 255 255 128 is the recommended value if USB network interfaces are used for redundancy CAN BE OBTAINED FROM SHELF FRU INFO No REDUNDANCY_POR Number 1040 The TCP port used for interactions No between redundant instances of the Shelf Manager REDUNDANT_IP_A P address None The IP address used for redundant No DDRESS communications This address actually specifies a pair of IP addresses that differ only in the least significant bit They are assigned to redundant Shelf Managers according to their hardware addresses RESERVATION_RE Number 10 The maximum number of times the Yes TRIES Shelf Manager retries the Reserve Device SDR command RMCP_NET_ADAPT String 16
336. rk interfaces see description of active standby mode in 3 4 2 1 the Shelf Manager Supports RMCP on both interfaces with different IP addresses as two separate IPMI channels channels 1 and 2 Each channel has its own setof LAN configuration parameters that includes the IP address network mask default gateway etc Both addresses are available via the Get Shelf Manager IP Addresses command The System Manager in this case is responsible for switching over to a different IP address if the currently used IP address becomes unavailable or maintaining two parallel RMCP sessions to both addresses If IP addresses on both interfaces become unavailable the System Manager can access one of the IP addresses on the backup Shelf Manager and initiate a switchover in a non standard way for example log in to the backup ShMM via telnet and issue the command clia switchover The active active mode is enabled by setting the configuration parameter USE_SECOND_CHANNEL t0 TRUE In this case the configuration parameter DEFAULT _RMCP_IP_ADDRESS2 should be set to a non zero IP address This IP address becomes the default IP address for the channel 2 and the configuration variables DEFAULT_RMCP_NETMASK2 and DEFAULT_GATEWAY_IP_ADDRESS2 specify the network mask and default gateway IP address for the channel 2 respectively The values of the above mentioned configuration parameters have no effect if the corresponding values in the channel configuration
337. rnel image Release 3 2 0 199 July 31 2013 Pigeon Point Shelf Manager User Guide OFFSETINFLASH SIZE IN Device NODE MOUNTED As CONTENT IN MBYTES MBYTES ON STARTUP 33 5 1 dev mtdcharl etc The etc J FFS2 file dev mtdblock1 system 34 5 1 75 dev mtdchar0 Nar The Aar J FFS2 file dev mtdblock0 system 36 25 15 75 dev mtdchar4 Not mounted The Linux root file system rfs image 52 8 dev mtdcharl2 Not mounted The first half of the dev mtdblock12 app1_jffs application specific J FFS2 partition 60 0 25 dev mtdchar3 Not mounted The U Boot firmware image 60 25 3 75 dev mtdcharl1 Not mounted The second half of dev mtdblock11 the app_jffs application specific FFS2 partition The following tables provide a summary of the Flash partitions maintained by Linux for each of the ShMM 1500 variants Table 22 Flash Partitioning for 32MB Flash ShMM 1500s OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED As CONTENT IN MBYTES MBYTES ON STARTUP dev mtdchar3 Not mounted The U Boot firmware image 0 25 0 5 dev mtdcharl0 var upgrade The first half of the dev mtdblock10 var upgrade J FFS2 filesystem 0 75 1 dev mtdchar2 Not mounted The Linux kernel dev mtdblock2 image 1 75 1 Idev mtdchar1 etc The etc J FFS2 dev mtdblock1 filesystem 2 75 1 75 dev mtdchar0 Nar The var FFS2 dev mdtblock0 filesystem 45 7 75 dev mtdchar4 Not mounted The Linux root file d
338. rnet port is referenced as the RMCP port though the other shelf external interfaces HTTP TELNET FTP are accessible via this port as well The other Ethernet port can be used for communication between redundant Shelf Managers or potentially for other purposes if the redundancy link is provided by another mechanism 3 4 1 Usage of the First Ethernet Interface Since the RMCP Ethernet port is directly connected to the site network the IP address should be set up appropriately for that network For example if the site uses the IP address range 192 168 0 x the RMCP Ethernet port should be Set to a unique IP address within that range such as 192 168 0 2 In a redundant ShMM setup itis Release 3 2 0 73 July 31 2013 Pigeon Point Shelf Manager User Guide important to note that only one ShMM the active ShMM has the RMCP IP address enabled on the RMCP Ethernet port The backup ShMM assigns the same IP address to the RMCP Ethernet port but only enables it when that ShMM assumes the active role This way the RMCP IP address maintains availability in a failover situation If the configuration variable PROPAGATE_RMCP_ADDRESS is Set to TRUE in that case the backup ShMM assigns a conjugate to the RMCP IP address to its RMCP Net Adapter that is itassigns an address that is different in the least significant bit of the fourth byte 3 4 1 1 Assigning an Additional IP Address to the First Network Interface In the default configuration no IP add
339. rom the current image set to the new image set erase all erase both the var and etc directories for the new image set copy only the SSH keys from the current image set skip checksums skip the verification of checksums in the images used for upgrade The following verbosity levels are supported none do not output verbose messages to stdout do not write them to the log file file do not output verbose messages to stdout do write them to the log file all output verbose messages to stdout and write them to the log file The following command variants invoke the HPM 1 upgrade mode rupgrade p lt URL gt copy config copy etc copy var copy all erase all skip checksums rupgrade package lt URL gt copy config copy etc copy var copy all erase all skip checksums Here p lt URL gt 0r package lt URL gt specifies the location of the new image in HPM 1 format which contains some combination of the new U Boot kernel RFS and application package images Other optional parameters have the same meaning as for a non HPM 1 upgrade The following command variant invokes the A2F060 upgrade mode see section 8 6 rupgrade upgrade a2f lt URL gt skip checksums The following command invokes rollback mode rupgrade rollback In this mode the rollback takes place to the other image set which is assumed to be the image set used before the last upgrade Rollback doesn t affec
340. rt In contrast to the SaMM 500 1500 the ShMM 700 firmware includes a fourth component that contains just the application binaries for the Shelf Manager and accompanying applications in many cases it will be sufficient to upgrade only this component in order to upgrade to the next version of the Shelf Manager which should result in faster upgrades On the other hand the RFS image on the ShMM 700 contains only system files and executables and does not contain application binaries and application specific files which makes it smaller and less prone to change in successive versions of the ShMM 700 firmware Similar to the ShMM 500 1500 two sets of upgrade component images exist on the ShMM 700 The storage and layout of upgrade components on ShMM 700 is however different from ShMM 500 1500 Only the U Boot images are assigned to dedicated Flash partitions Kernel and RFS images are stored as files on a single partition formatted under the UBIFS file system U Boot on ShMM 700 is able to read files from UBIFS partitions and therefore is able to locate and load the proper kernel and RFS images Also there exist two instances of the application component package one of which represents the candidate firmware and the other of which represents the confirmed firmware They are stored on another user accessible UBIFS Flash partition ina compressed form When the RFS is loaded and the initial startup script etc re Is executed it picks up the ap
341. s 165 4 12 SHMM TESTS AVAILABLE VIA THE DIAGNOSTIC INFRASTRUCTURE 2 ceceeeeeteeeeseeeeeeees 166 5 USING THE IPMI ANALYSIS TOOLS 0 cscccseeeseseeeseeeeeseeeeneeeeseaeseseeeenseeeeseeeeseaeseneeneneeees 169 5 1 IN TAIS SECTION S jeaeccheucaceceita ssctieses einan eaaa E aaea aaea a aaa a aai a aaaea 169 5 2 IPMB TRACE COLLECTION DAEMON IPMB_TRACED sscccceseeeceeeeecseesessesseeeesaeeeeaeeees 169 5 3 IPMI TRACE ANALYZER 1 0 c0sctccccv send cccwacddaacnveenetobuaddencncch daccvasedaeeeeeetandedes ddectecntccawashdaeewec 173 5 3 1 Introduction to the Wireshark GUI ccccccccccccecseceesseceeeeeceneeeseeseeeeeeeeeersnneetaes 173 5 3 2 Introduction to Terminal orienteO Wireshark ccccccccecceeeeseeeceenteeeeesnneeeeeenneees 176 5 3 3 Installing IPMI Analyzer SOfWAPC 21 2c ccccccceceeeeeeseceeeeeceneeessaeeteeeseeeeersnetetiaes 176 5 4 COLLECTING AND ANALYZING A TRACE USING THE GUI AND COMMAND LINE TOOLS 176 5 4 1 Collecting IPMI Traces on IPMB 0 in Unattended MOde c ccccccccssccesenees 176 5 4 2 Collecting IPMI Traces on IPMB O in Controlled MOCeC cccccsccccssssseceseeeees 181 5 4 3 Collecting an IPMI Trace in Board Trace Mode 22 cccsccccseceessesseseecenstessinees 184 5 4 4 Collecting Traces of IPMI Traffic Over the Network ccccccccccsssstecssstseeeeeneees 184 5 4 5 Analyzing a Trace Using the GUI 2 cccscccceeecesseeeeeneeceeeeeceeneesnseseeneeceeneeenanens 187
342. s can be specified in the Shelf FRU Information the main reason is that many variables control the Shelf Manager initialization behavior which happens before the Shelf FRU Information is found For such variables overriding the value when the Shelf FRU Information is found could not have any effect since the choice they control has typically already been made If such variables are specified in the Shelf FRU Information they are parsed successfully but value assignments are silently ignored Release 3 2 0 35 July 31 2013 Pigeon Point Shelf Manager User Guide The configuration variable SHELF_MANAGER_CONFIGURATION_IN_SHELF_FRU_INFO controls whether this functionality of the Shelf Manager is enabled By default the value of this variable is FALSE and the functionality is disabled The following table lists the configuration parameters that are currently supported Table 7 Shelf Manager Configuration Parameters DEFAULT DESCRIPTION CANBE OBTAINED FROM SHELF FRU INFO 2_X_SYSTEM Boolean None If specified this parameter explicitly No designates the current shelf as CompactP Cl if TRUE or AdvancedTCA if FALSE If not Specified the choice of the shelf type is made automatically This parameter should not be specified unless itis necessary to override an incorrect hardware detection algorithm for the shelf type ACTIVATE_LOCAL Boolean FALSE If setto TRUE both IPM controllers No pe ieee exposed by the active Shelf Manager
343. s case Release 3 2 0 160 July 31 2013 Pigeon Point Shelf Manager User Guide An example of a LAN parameters configuration file is given below Parameters for a board in physical slot 192 168 1 101 255 255 255 0 192 168 192 168 1 102 255 255 255 0 192 168 192 168 1 103 255 255 255 0 192 168 PRR a errr ie ea Parameters for an AMC in the first AMC slot of a board in slot 1 l 192 168 1 104 255 255 255 0 192 168 1 1 pl 192 168 142 105 255 255 255 0 192 168 1121 l 192 168 1 106 255 255 255 0 192 168 1 1 Parameters for an AMC in the second AMC slot of a board in slot 1 12 192 1684 15107 255 255 255 304 192 168 1061 1 2 192 168 1 108 255 255 255 0 192 168 1 1 Parameters for a board in physical slot 192 168 1 121 255 255 255 0 192 168 192 168 1 122 255 255 255 0 192 168 192 168 121 123 255 255 255 0 192 168 NO NM NB 3 PRPeDN 4 10 2 1 Using the FRU Info Compiler to Place LAN Configuration Data in the Shelf FRU Information The FRU Information Compiler implements special syntax that can be used to incorporate a LAN configuration parameters file into a FRU Information image that is being created For example the following instructions add a compressed LAN configuration parameters file subsidiary lan param gz to the current FRU Information image PPS Board AMC LAN Configuration File subsidiary_lan_param gz 4 10 2 2 Using the Shelf Manager Comman
344. s from the Shelf Manager to the multiplexer on the master only 12C bus but must be FALSE if the HPDL based reset mechanism is used No LOCAL_SHELF_FR U Boolean TRUE Create a local FRU 1 on the Shelf Manager that exposes the Shelf FRU Information obtained from the file var nvdata shelf fru_i nfo No M7_TIMEOUT Number The maximum time in seconds fora FRU to stay in M7 state after the expiration of this time the FRU is automatically transitioned to M0 1 the default stands for forever Setting this parameter to O completely prevents FRUs from going into state M7 Yes MAX_ALERT_POLI CIES Number 64 The maximum number of PEF Alert Policies available The maximum allowed value for this parameter is 128 No MAX_ALERT_STRI NGS Number 64 The maximum number of PEF Alert Strings available No MAX_DEFERRED_A LERTS Number 32 The maximum number of outstanding PEF alerts No MAX_EVENT_FILT ERS Number 64 The maximum number of PEF event filters available Release 3 2 0 57 J uly 31 2013 Pigeon Point Shelf Manager User Guide TYPE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO MAX_EVENT_SUBS Number 64 The maximum number of entities that No CRIBERS can simultaneously subscribe to receive event notifications from the Shelf Manage
345. s of the active Shelf Manager Forced switchover can also be initiated via the CLI switchover command with the option force addressed to the backup Shelf Manager As discussed briefly in section 2 4 2 redundant ShMMs share three Hardware Redundancy Interface HRI data items between them for six items total these items together indicate for each ShMM the health presence and switchover request state of the ShMM itself referenced as the local state and its peer referenced as remote state Each ShMM variant has a different hardware implementation of these signals At the software level each of the Shelf Managers can access a register that provides a view of these data items The register is implemented in a PLD a CPLD Complex Programmable Logic Device on the ShMM 500 or an FPGA Field Programmable Logic Array on the ShMM 1500 and ShMM 700 thatis part of each ShMM For details of the HRI implementation for each architecture please see the appropriate ShMM Hardware Architecture specification In the remainder of this section these HRI data items are referenced informally as the Healthy Presence and Switchover Request bits each with Local and Remote variants In addition the set of HRI data items also includes an Active bit which is 1 on the active Shelf Manager This section and most other Shelf Manager user documentation refers to the device that implements the HRI functionality as a CPLD though that is not strictly true on ei
346. s the number of milliseconds that the Shelf Manager waits for all AXle modules to reach State M4 If the timeout expires and not all AXle modules have reached State M4 the Shelf Manager terminates the wait issues an error message and proceeds to the next State of power sequencing No AXIE_SEQUENCIN G_WAIT_DELAY Number 100 The delay in milliseconds between Successive attempts by the Shelf Manager to poll the state of hardware and software signals and advance the Current power sequencing state if it can be advanced No AXIE_SYSTEM_MO DULE_IPMB_ADDR ESS Number 0x82 The IPMB address of the AXle system module No Release 3 2 0 40 July 31 2013 Pigeon Point Shelf Manager User Guide TYPE DEFAULT DESCRIPTION CAN BE OBTAINED FROM SHELF FRU INFO AXIE_TIMING_AS Boolean FALSE This variable specifies how the E No YMMETRIC MATCH Keying logic in the Shelf Manager matches AXle timing links If this variable is setto TRUE asymmetric matching applies that is links tagged as Instrument Slot Inputs match links tagged as System Slot Outputs on the other side If this variable is set to FALSE regular matching rules apply to AXle timing links This configuration variable affects only AXle configurations in non AXle configurations its value is ignored BOARD_LAN_PARA String 64 The list of IPMI channel numbers on Yes ee boards and modules that are
347. sec timeout period This timeout period is considered sufficient for any software that boots after the reset to proceed to the point where it is able to call rupgrade_tool c which strobes the upgrade WDT in case it is active without having to strobe the upgrade WDT in the interim The utility performs a strobe of the upgrade WDT just before resetting the ShMM c v Proceed with the reliable upgrade procedure after the ShMM is booted from the provisional Flash The invocation of rupgrade_tool cis done from the etc rc Script As described below certain situations discovered by rupgrade_tool c imply a failure in the upgrade procedure and require corrective actions including those resulting in the need to soft reset the ShMM This means that an invocation of rupgrade_tool c may not return and instead may result in a reset of the ShMM If a reset takes place it reverts the ShMM to the software installed in the persistent Flash If the upgrade WDT is active and has fired at any step prior to invocation of rupgrade c this means that the ShMM already reverted to the software in the persistent Flash In this scenario the utility disables the upgrade WDT and returns to the use of persistent Flash and terminates the upgrade procedure If the upgrade WDT is active but has not fired this means that the ShMM successfully booted up to this point from the provisional Flash The utility strobes the upgrade WDT and exits with the return code o
348. sed by the Shelf Manager but is propagated to set the Linux time zone For instance if the time zone name XXX0 is used the date command produces output like the following Thu Sep 9 21 24 24 XXX 2004 Daylight saving time is not supported Here is an example of a time zone definition for US Eastern Time timezone EST5 Here the digit 5 specifies that the time zone is 5 hours west of GMT Any three letters can replace EST they are used to identify the time zone in for example the Linux date command output For timezones with half hour or quarter hour differences from an integer offset e g UTC 05 30 itis possible to use the etc localtime file Due to limited RFS space there is no time zone database on ShMM but such file can be copied from a Linux system e g usr share zoneinfo Asia Calcutta and installed ona ShMM as etc localtime If this file is detected at ShMM startup the Tz variable is not set so date related library calls use the time zone specified in etc localtime 3 11 Setting Up and Using ShMM Power On Self Tests The available P ower On Self Test POST tests are built into U Boot The choice of the available POST tests that are actually executed is controlled by the dedicated U Boot environment variables post_normal and post_poweron Release 3 2 0 116 July 31 2013 Pigeon Point Shelf Manager User Guide The value of the environment variable post_normal contains names of tests that are executed on each
349. sed to configure IP MI sensors on the carrier and in the shelf These collections of SDRs are compiled with the SDR compiler stored as binary data and retrieved by the Shelf Manager during initialization The Shelf Manager uses these saved SDRs for the sensors described in HPDL that it creates during initialization 3 6 1 Compiling HPDL Definitions HPDL definitions in text format are compiled by the HPDL compiler into a binary format that conforms to the ASN 1 basic encoding rules ISO 8825 and stored in an output file The HPDL compiler executable is available both for Linux x86 and Windows operating systems with the names hpdlc and hpdic exe respectively The Linux version of the compiler can be run on various Linux distributions including Red Hat Mandriva Linux and Ubuntu The command line for the HPDL compiler has the following synopsis hpdic d v lt in file gt lt out file gt The parameter lt in file gt specifies the input file that contains the HPDL descriptions in text format The optional parameter lt out file gt specifies the output file if this parameter is omitted the name of the output file is based on the input file name with the suffix bin appended at the end Release 3 2 0 101 July 31 2013 Pigeon Point Shelf Manager User Guide Option d causes the HPDL compiler to output additional debugging information when compiling the input file Option v causes the HP DL compiler to identify its ver
350. send IP MI trace packets in the appropriate format over the RMCP session while the session is open when the session closes the controller stops tracing IP MI messaging In this mode the trace daemon instead of collecting the IP MI trace locally on the Shelf Manager establishes a RMCP session with the HPM 2 IPMI Trace Payload with the target board and forwards the data received over this session to the trace analyzer in controlled mode or stores it in a file in unattended mode For example the following command line can be used to start the trace daemon in the board trace mode with a board that has the IP address 192 168 1 149 The specified channel mask instructs the IPM controller to trace the IPMB 0 and IPMB L buses The trace daemon operates in controlled mode daemon f ipmb_traced c w b 192 168 1 149 m 0x81 5 4 4 Collecting Traces of IPMI Traffic Over the Network When collecting the traces of IP MI traffic transferred over the network RMCP or RMCP sessions the trace daemon is not used Wireshark or tshark collects the trace directly from one of the local network interfaces on the system on which that tool runs that is the analyzer host Usually the analyzer host is also the one that participates in the IP Ml oriented communication However if the analyzer host s local network interface supports promiscuous mode Wireshark or tshark can collect a trace of PMI communication between two separate systems that are no
351. ser programs or scripts executed on the ShMM External event handling is implemented via the PEF alerting mechanism As an extension of this mechanism all alerts with a destination channel set to the System Interface Channel Fh are delivered by the Shelf Manager to the standard input of a designated external event handler program This program is specified via the configuration parameter EXTERNAL_EVENT_HANDLER 3 12 1 Detailed Steps to Configure External Event Handling To use external event handling the user should take the following steps 1 Prepare a script or executable for external event handling and place it on the Shelf Manager Flash file system for example into the directory var bin 2 Specify the location of the external event handler in the Shelf Manager configuration file etc shelfman conf as the value of the configuration parameter EXTERNAL_EVENT_HANDLER for example EXTERNAL_EVENT_HANDLER var bin ext_handler sh If the configuration parameter EXTERNAL_EVENT_HANDLER is not defined or is defined as an empty value external event handling is disabled 3 Start the Shelf Manager 4 Enable PEF via the CLI if not enabled yet clia setpefconfig control 1 5 Enable alerting if not enabled yet clia setpefconfig action_control 1 Release 3 2 0 118 July 31 2013 Pigeon Point Shelf Manager User Guide 6 Set alert strings if necessary clia setpefconfig alert_string 1 test string
352. ses associated with each of the two redundant dedicated ShMC slots The fields Entity ID and Entity Instance if specified in the replacement SDRs and different from 0 0 replace the corresponding attributes of the sensor The Sensor Initialization field identifies the specific attributes to be redefined and indicates what fields in the rest of the replacement SDR are applicable It is specified as a list of symbolic constants the following constants are defined e THRESHOLDS indicates that the sensor thresholds are to be replaced with the corresponding threshold values specified in the replacement SDR e HYSTERESIS indicates that the sensor hysteresis values are to be replaced with the corresponding hysteresis values specified in the replacement SDR e SENSOR_TYPE indicates that the sensor type is to be replaced with the corresponding fields Sensor Type and Event Reading Type from the replacement SDR e EVENTS indicates that the event masks are to be replaced with the corresponding fields from the replacement SDR The field Sensor Capabilities cannot be replaced If and only if the symbolic constant SENSOR_TYPE has been specified in the Sensor Initialization field the fields Sensor Type and Event R eading Type from the replacement SDR replace the corresponding attribute of the sensor even if the values of these fields are 0 Release 3 2 0 111 July 31 2013 Pigeon Point Shelf Manager User Guide e fand only if the symbolic constant
353. significant extension to support them well IPMI defines a management infrastructure that is widely used across the PC and server industry PICMG 3 0 the AdvancedTCA specification defines the necessary extensions to IPMI PICMG Specifications are based on revision 1 5 of IPMI However the Pigeon Point Shelf Manager conforms to revision 2 0 of the IPMI specification and implements numerous IP MI 2 0 features including RMCP Virtual LAN support and firmware firewall commands PICMG has significantly extended IP MI to cover the needs of open modular architectures In fact about 30 of the 656 pages of PICMG 3 0 are devoted to hardware platform management including the definition of 38 new commands ten new FRU Information data structures several quite complex and 3 new sensor types The AdvancedMC and MicroTCA specifications add another 208 pages of hardware platform management coverage The strategy for the Pigeon Point Shelf Manager is to fully support these extensions and also map them to other platform architectures such as CompactPCI Note AdvancedTCA has adopted the term shelf for alignment with typical practice in telecommunications applications Traditionally for instance in the CompactPCI specifications the term chassis has been used with essentially the same meaning Figure 1 shows the logical elements of an example AdvancedTCA shelf identified in terms of the ATCA Specification and potential sites for inc
354. sion Error messages if any are output on the standard error stream of the HP DL compiler For example the following command line compiles the input file acb3 hpd 11 into the binary file acb3 hpdl bin gt hpdlc acb3 hpdl 3 6 2 Compiling SDRs Most of the attributes of sensors referenced in HPDL files are actually defined through Sensor Data Records SDRs The details of these SDRs are specified in text files and compiled to binary by a Separate Python based compiler This compiler is available from Pigeon Point for both Linux and Windows operating systems there is a separate User Guide for this compiler For example the following command line compiles the input SDR file acb3 sdr into the binary SDR file acb3 sdr bin gt python sdrce py acb3 sdr 3 6 3 Deploying HPDL Data to the ShMM File System The Shelf Manager first looks for HPDL data and SDRs in FRU Information areas Shelf FRU Information for the shelf level data and carrier FRU Information for the ShMM carrier data If the data is not found there it is retrieved from the following files on the ShMM file system var nvdata carrier_data Binary HP DL definitions for the carrier var nvdata carrier_sdrs Binary SDRs for the carrier sensors var nvdata chassis_data Binary HP DL definitions for the shelf var nvdata chassis_sdrs Binary SDRs for the shelf sensors If the Shelf Manager cannot retrieve HP DL data from these locations it uses the
355. sponding stage e lt description gt is a description of the corresponding stage The list of stages and the corresponding description of each stage is summarized in Table 27 below The status file can be used to track the progress of the reliable upgrade procedure it is also used by the reliable upgrade utility itself to synchronize the state of the upgrade procedure across ShMM reboots Table 27 Reliable Upgrade Stages STAGE NUMBER DESCRIPTION 1 Images have been downloaded to tmp and verified 2 Boot volume mount is about to begin Boot volume mount is complete 4 Candidate has been marked as non valid 5 U Boot image write is complete 6 Kernel image write is complete 7 RFS image write is complete 8 9 Boot volume umount is complete Application package has been removed 10 New application package write is complete 11 Sync is complete 12 var and etc have been processed 13 First strobe after reboot has been done 14 Upgrade is confirmed Release 3 2 0 224 July 31 2013 Pigeon Point Shelf Manager User Guide 8 5 Reliable Upgrade Utility A special user space utility rupgrade is provided on the ShMM This utility has several modes of operation and allows the user to perform multiple tasks including reliable upgrades of various ShMM software components status checks for the most recent upgrade and manual rollbacks to the previous software versions Another utilit
356. st image Release 3 2 0 149 July 31 2013 Pigeon Point Shelf Manager User Guide 4 6 Automatic SEL Truncation The System Event Log SEL on the Shelf Manager stores events from all IPM controllers and FRUs in the shelf and can easily exceed its maximum capacity To prevent overflows the Shelf Manager can automatically truncate the SEL removing the oldest entries as SEL approaches its limits The automatic truncation algorithm works as follows The two configuration parameters are defined SEL_HIGH_WATERMARK and SEL_LOW_WATERMARK The first parameter relates to the percentage of free entries in the SEL and the second one relates to the percentage of occupied entries in the SEL When the percentage of free SEL entries falls below SEL_HIGH_WATERMARK the truncation thread wakes up and starts deleting the oldest entries from the SEL using the IPMI command Delete SEL Entry The thread works until the percentage of occupied entries in the SEL falls below SEL_LOW_WATERMARK then it stops and stays dormant until the percentage of free SEL entries falls below the SEL_HIGH_WATERMARK again This algorithm is enabled by default but can be turned off by setting the configuration variable SYSTEM _MANAGER_TRUNCATES_SEL t0 TRUE In that case automatic truncation is turned off but the Shelf Manager assists the System Manager by indicating the current state of the SEL via the sensor SEL State of the type Event Logging Enabled as follows in accordanc
357. st name or IP address lt path gt relative path to the image file path origin depends on HTTP server configuration TFTP Download the image tftp lt host gt lt path gt tftp 192 168 1 253 shmm using the TFTP 700 rootfs image protocol lt host gt host name or IP address lt path gt relative path to the image file path origin depends on TFTP server configuration Release 3 2 0 227 July 31 2013 Pigeon Point Shelf Manager User Guide PROTOCOL DESCRIPTION URL SYNTAX EXAMPLE URL NAME FTP Download the image ftp lt user gt l lt pass gt lt host ftp user password 192 using the FTP gt lt path gt 168 1 253 shmm700 rootf protocol s image lt user gt user name lt pass gt password lt host gt host name or IP address lt path gt relative path to the image file path origin depends on FTP server configuration SCP Download the image scp lt user gt lt host gt lt path scp user 192 168 1 253 using the SSH SCP gt Ishmm700 rootfs image protocol lt user gt user name scp user 192 168 1 253 lt host gt host name or ip home user shmm700 roo address tfs image lt path gt relative path to the image file path origin is usually the user s home directory An absolute path can be specified by beginning it with The password must be entered interactively by the user In upgrade mode both regular an
358. stalled image terminating the Shelf Manager is not necessary To program a new FPGA image to the routing element FPGA s issue the following command ripmb_prog lt image_name gt where lt image_name gt is the name of the upgrade image with a dat extension By default the ripmb_prog utility tries to determine how many FPGA devices are connected to the J TAG chain and programs each of them with the upgrade image IMPORTANT all routing element FPGAs on the Shelf Manager must be programmed with the same FPGA image If the upgrade procedure is interrupted at any point it must be re started to ensure all FPGAs get fully programmed When the programming is successful the following message s are displayed by the ripmb_prog utility depending on the number of FPGA devices to be programmed Image file radial dat Determining the number of devices in the JTAG scan chain 2 Programming devic Sai ag ee oe Bak GS Reg as Roe dl Sees BOS w Ecosse Be eS SUCCESS Programming devic 2a a PEA sarc ca ew a E te lees Bete Gee whee ee SUCCESS bh To obtain the version of the currently installed routing element image issue the following command ripmb_prog v This command displays the version of the routing element FPGA design and firmware in the following format lt fpga_version gt lt firmware_version gt for example 1 0 0 0 1 0 If this command displays veRston mIsmatcu the FPGAs must be re programmed with the late
359. states M3 M5 during startup after the Shelf FRU Info is found After the initialization Shelf Manager performs this procedure during board or module activation when the board or module reaches state M3 The Shelf Manager maintains internal records to avoid Sending configuration parameters to the same board or module twice To detect what LAN channels a specific board or module has the Shelf Manager sends it the command Get Channel Info with a channel number picked sequentially from the BOARD_LAN PARAMETERS CHANNEL LIST Ifa board or module responds with the Channel Medium Type 802 3 LAN for some channel the Shelf Manager sends the command Get LAN Configuration Parameters IP Address Source for that channel If the IP Address source 3 address loaded by BIOS or system software the Shelf Manager assigns configuration parameters for this channel Commands are sent on IPMB 0 for modules commands are directed to the module address on IPMB L and are wrapped with Send Message commands directed to the IPMB 0 address of the carrier board Parameter sets are obtained in order as they are needed for a specific slot If a parameter set cannot be obtained for instance if parameter sets are exhausted for a specific slot configuration parameters are not assigned The Shelf Manager sends the configuration parameters to a board or module as a set of Set LAN Configuration Parameters commands over IPMB 0 however the channel parameter in these comma
360. sts If one of the threads fails to strobe the internal watchdog the actual watchdog does not get strobed and ultimately the ShMM resets initiating a failover to the backup Shelf Manager In addition currently on all platforms except ShMM 700 lock data structures are periodically checked directly for the presence of a deadlock See section 4 8 for further background CAN BE OBTAINED FROM SHELF FRU INFO No DEVICE_POLL_TI Number 10 The time in seconds between Yes MEOUT successive polls of the IPMB 0 devices by the Shelf Manager via sending the Get Device ID command to them DHCP_FOR_RMCP_ Boolean FALSE If this variable is set only the RMCP Yes PADS IP addresses are assigned via the DHCP mechanism private IP addresses of the ShMMs are not touched by DHCP DHCP_SERVER_AD P address None This parameter is the IP address of Yes DRESS the DHCP server it applies only if the variable USE_DHCP S TRUE If this parameter is omitted or set to 0 0 0 0 and the USE_DHCP variable is TRUE the Shelf Manager accepts address information from any DHCP server that responds to its broadcast discovery request Release 3 2 0 48 July 31 2013 DHCP_SPECIAL C LIENT_ID_ FORMA T Pigeon Point Shelf Manager User Guide Boolean DEFAULT FALSE DESCRIPTION If this parameter is set to TRUE the DHCP Client ID has a special format that is used by some Pigeon Point
361. support encryption 4 4 Redundant Operation The active Shelf Manager exposes the ShMC device address 20h on IPMB manages IP MB and the IPM controllers and interacts with the System Manager over RMCP and other shelf external interfaces It maintains an open TCP connection with the backup Shelf Manager It communicates all changes in the state of the managed objects to the backup Shelf Manager The backup Shelf Manager does not expose the ShMC on IPMB does not actively manage IPMB and IPM controllers nor interact with the System Manager via the shelf external interfaces with one exception noted below Instead it maintains the state of the managed objects in its own memory volatile and non volatile and updates the state as directed by the active Shelf Manager These state updates are not applied immediately but are cached on the backup Shelf Manager The backup Shelf Manager may become active as the result of a switchover Two types of Switchover are defined e Cooperative switchover the active and backup Shelf Managers negotiate the transfer of responsibilities from the active to the backup Shelf Manager this mode is supported via the CLI switchover command issued on the active or backup Shelf Manager Release 3 2 0 136 July 31 2013 Pigeon Point Shelf Manager User Guide e Forced switchover the backup Shelf Manager determines that the active Shelf Manager is no longer alive or healthy and forcefully takes on the responsibilitie
362. t any key to stop autoboot 0 shmm500 For SAMM 1500 U Boot 1 1 4 Jun 15 2006 17 49 12 MPC83XX Clock configuration Coherent System Bus 99 MHz Core 249 MHz Local Bus 24 MHz CPU MPC83xx Rev 1 1 at 249 975 MHz Board ShMM 1500R PCI 32 bit 33 MHz I2C ready DRAM 128 MB FLASH 64 MB PCr Bus Dev VenId DeviId Class Int 00 17 21172 0001 ff00 00 CRS serial Quti serial Err serial FPGA firmware version 1 12 carrier id 0 Net TSECO TSEC1 Hit any key to stop autoboot 0 shmm1500 For ShMM 700 PowerPrep start initialize power Battery Voltage 0 80V No battery or bad battery detected Disabling battery voltage measurements Feb 24 201207 45 29 FRAC 0x92926152 memory type is DDR2 ait for ddr ready 1 power 0x00820616 Frac 0x92926152 start change cpu freq hbus 0x00000003 cpu 0x00010001 start test memory accress ddr2 0x40000000 finish simple test U Boot 2009 08 Feb 15 2012 18 50 25 Freescale i MX28 family CPU 297 MHz BUS 99 MHz EMI 130 MHz GPMI 24 MHz T2C ready DRAM 128 MB SFGEN N25Q512A detected total size 64 MB Release 3 2 0 23 July 31 2013 Pigeon Point Shelf Manager User Guide A2F SPICOMM protocol v1 6 M3 firmware v0 7 FPGA design v0 41 0 0 A2F A2F firmware version 0 7 A2F Last reset cause HARD A2F Device type A2FO60M3E FG256 A2F MSS clock frequency 40 MHz A2F Fabric clock frequency 2
363. t is set ODh The Shelf Manager is Active but senses the CPLD Active bit set on other ShMM 6 Event Data 2 7 4 Event S everity 3 0 The highest previous offset 7 Event Data 3 bits 8 15 of the ShMM CPLD state on the ShMM 500 1500 7 1b Reboot was caused by watchdog expiration 6 1b Interrupt happened 5 1b Active 4 1b Local presence 3 1b Hot Swap latch open Release 3 2 0 141 July 31 2013 Pigeon Point Shelf Manager User Guide 2 1b Remote switchover request 1 1b Remote healthy 0 1b Remote presence On the ShMM 700 this byte should be ignored its value is always 00h 4 4 3 Reboot Reason Sensor Both active and standby Shelf Managers also expose a sensor that indicates the reason for the last reboot This discrete sensor is associated with the physical Shelf Manager has number 129 name Reboot Reason and OEM event reading type code DDh The state mask for the sensor indicates the cause of the last reboot its bits have the following meanings 0 The reboot reason is unknown 1 The reboot was caused by a switchover operation 2 The reboot was caused by a forced switchover operation 3 The reboot was caused by the CLI command terminate 4 The reboot was caused by loss of the HEALTHY bit 5 The reboot was caused by loss of the ACTIVE bit 6 The reboot of the Backup ShMM happened because the redundancy connection was broken but the Active ShMM was s
364. t mounted The Linux kernel image 12 0 25 dev mtdchar3 Not mounted The U Boot firmware image 12 25 3 75 dev mtdchar4 Not mounted The Linux root file system rfs image Table 20 Flash Partitioning for 32MB Flash ShMM 500s OFFSETINFLASH SIZE IN Device NODE MOUNTED AS CONTENT IN MBYTES MBYTES ON STARTUP 0 0 5 dev mtdcharl0 var upgrade The first half of the dev mtdblock10 var upgrade J FFS2 filesystem 0 5 1 dev mtdchar2 Not mounted The Linux kernel dev mtdblock2 image 1 5 1 Idev mtdchar1 etc The etc J FFS2 dev mtdblock1 filesystem 2 5 1 75 dev mtdchar0 Nar The var FFS2 dev mdtblock0 filesystem 4 25 7 15 dev mtdchar4 Not mounted The Linux root file dev mtdblock4 system rfs image 12 0 25 dev mtdchar3 Not mounted The U Boot firmware image 12 25 3 75 Idev mtdchar11 Not mounted The first half of the dev mtdblock11 app_jffs application specific FFS2 partition 16 0 5 dev mtdcharl0 var upgrade The second half of dev mtdblock10 the var upgrade J FFS2 file system 16 5 1 dev mtdchar7 Not mounted The other Linux dev mtdblock7 kernel image Release 3 2 0 198 July 31 2013 Pigeon Point Shelf Manager User Guide OFFSETINFLASH SIZE IN DEVICE NODE MOUNTED As CONTENT IN MBYTES MBYTES ON STARTUP 17 5 1 dev mtdchar6 Not mounted The other etc dev mtdblock6 J FFS2
365. t the A2F 060 firmware state because the A2F 060 has its own auto rollback mechanism that is utilized during an A2F reliable upgrade procedure Additional command variants invoke other modes of the reliable upgrade utility Get Upgrade Status mode output current upgrade status on stdout rupgrade s rupgrade status Release 3 2 0 226 July 31 2013 Pigeon Point Shelf Manager User Guide Help mode output a brief help summary on stdout rupgrade h rupgrade help Get Version mode output the version of the utility to stdout rupgrade version Add Hook Mode rupgrade add hook lt URL gt In this mode the utility adds a new hook script specifically the script file designated by the lt URL gt to the list of scripts executed during the upgrade procedure Scripts are stored in the directory etc hook to remove a hook script just delete the corresponding file from this directory The following table lists the protocols that are recognized in the URLs by the reliable upgrade utility Table 28 Protocols Recognized in the URLs by the Reliable Upgrade Utility PROTOCOL DESCRIPTION URL SYNTAX EXAMPLE URL NAME File Get image from a file lt abs_path gt file tmp rootfs image local ShMM file lt abs_path gt absolute path to the image file on the ShMM starts with HTTP Download the image http lt host gt lt path gt http 192 168 1 253 shm from an HTTP server m700 rootfs image lt host gt ho
366. t the analyzer host if at least one of them is on the same LAN with the analyzer host The following stepwise procedure shows how to collect an IPMl over network trace and open it for analysis in the GUI The GUI specific steps are followed by corresponding guidance for collecting a similar trace in tshark 1 On the analyzer host start the GUI C gt wireshark In the main menu go to Capture submenu and select Options This opens the Capture Options dialog In the dialog use the Capture Interface window to Release 3 2 0 184 July 31 2013 Pigeon Point Shelf Manager User Guide Specify the target Choose the name of the network interface through which the IP MI communication takes place e g ethO Additionally you can use the Update list of packets in real time checkbox to specify that the GUI should show the trace in real time For instance Figure 12 Capture Options Window analyzing networked IPMI traffic Wireshark Capture Options O x Capture 4 Interface eth0 IP address 80 240 102 57 fe80 203 dff fe2e 2371 Link layer header type Ethernet 4 gt v Capture packets in promiscuous mode Limit each packet to bytes Capture File s Display Options File Browse v Update list of packets in real time Use multiple files Automatic scrolling in live capture Hide capture info dialog Name Resolution Stop Capture v Enable MAC name resolution after Enable network name r
367. t variables do not apply to the second network interface Changing the RMCP network parameters is a two step process First the U Boot network environment variables need to be updated then the booted active ShMM module network settings need to be updated using the Shelf Manager command line interface CLIA Specific steps are shown below 1 Attach a serial port console connection to the ShMM module This typically is 115200 Baud N 8 1 for ShMM 500 ShMM 1500 and ShMM 700 2 Reset the ShMM carrier and press the space bar to interrupt the automatic boot up process You should see on ShMM 500 U Boot 1 1 2 Apr 27 2005 19 17 09 Board ShMM 500 S N 00 00 00 00 00 00 00 00 00 03 03 03 DRAM 64 MB Flash 16 MB In serial Out serial Err serial Net AulX00 ETHERNET Hit any key to stop autoboot 0 shmm500 On ShMM 1500 you should see the following U Boot 1 1 4 Jun 15 2006 17 49 12 MPC83XX Clock configuration Coherent System Bus 99 MHz Core 249 MHz Local Bus 24 MHz CPU MPC83xx Rev 1 1 at 249 975 MHz Board ShMM 1500R PCI1 32 bit 33 MHz T2C ready DRAM 128 MB FLASH 64 MB PCr Bus Dev VenId DeviId Class Int 00 17 1172 0001 ff00 00 Release 3 2 0 83 J uly 31 2013 In Out Err FPGA Net Pigeon Point Shelf Manager User Guide serial serial serial firmware version 1 12 TSECO TSEC1 Hit any key to stop autoboot shmm1500 carrier id 0 0
368. ted by the daemon either from a file or directly from a ShMM via a TCP IP connection Wireshark and its terminal oriented varianttshark are two specific applications that can play this role In addition Wireshark and tshark can also collect and analyze IPMI packets that are transferred over an Internet Protocol IP based network via RMCP or RMCP both of which are layered above UDP 5 1 In This Section This section contains the topics listed below J ust click on a topic to go to it e PMB Trace Collection Daemon ipmb_t raced e PMI Trace Analyzer e Collecting and Analyzing a Trace Using the GUI and Command Line Tools 5 2 IPMB Trace Collection Daemon ipmb_traced The IPMB trace daemon can operate in unattended trace mode or controlled trace mode In the unattended trace mode the trace data is collected into a file ina ShMM local or NFS mounted file system When the trace collection has been completed the file can be transferred to the development host then read and reviewed via the trace analyzer In the controlled trace mode the trace data is transferred directly to the trace analyzer via a TCP IP connection The trace analyzer can display and analyze the collected data on a real time basis Release 3 2 0 169 July 31 2013 Pigeon Point Shelf Manager User Guide The IPMB trace collection daemon ipmb_ traced runs on the ShMM and takes the role of IPMB trace collector The daemon interacts with the IP MB specific extensio
369. ters in the Shelf Manager configuration file etc shelfman conf as follows REDUNDANCY_NET_ADAPTER usb0 REDUNDANCY_NET_ADAPTER2 usbl One additional consideration relates to the definition of the subnet mask for the redundancy network interfaces In the legacy case when only one redundant network adapter is used two different IP addresses are derived from the redundancy IP address specified in etc shelfman conf They are assigned to the two endpoints of the redundancy connection and differ only in the least significant bit However when two redundancy network adapters are used four different IP addresses are used one for each of the endpoints two endpoints on each of the two redundant Shelf Managers To ensure proper operation the two endpoints on the same Shelf Manager usb0 and usb1 must belong to different logical networks while usb0 on one Shelf Manager and usb1 on the other Shelf Manager must belong to the same logical network This is achieved by dividing the IP address Space into two ranges These ranges logical networks are defined by the subnet mask given by the parameter REDUNDANCY_NETMASK from the configuration file etc shelfman conf If the network mask is 255 255 255 128 then the first range is 192 168 1 0 192 168 1 127 and the other is 192 168 1 128 192 168 1 255 The usb0 endpoint on the first Shelf Manager and the usb0 endpoint on the other Shelf Manager are in the first range The usb0 e
370. th kernel and RFS images into RAM and booting the system If at least one of the images is broken and there is no reliable upgrade in progress the command also outputs an error message and performs a Single attempt to boot the ShMM using the alternative rollback kernel and RFS images ver The U Boot version string The value is taken from the U Boot image The version string contains the base U Boot version and the build timestamp as well For example on the ShMM 700 the variable contains something like U Boot 2009 08 Feb 24 2012 07 44 12 3 2 2 Assigning Values to Environment Variables To assign a value to an environment variable on either ShMM variant use the format Release 3 2 0 30 July 31 2013 Pigeon Point Shelf Manager User Guide shmmxx00 setenv lt variable name gt lt new_value gt For example shmmxx00 setenv bootdelay 1 Once all of the environment variables have been properly set you need to save them back out to the Flash so that they remain after the ShMM is powered down The saveenv command is used for this purpose shmmxx00 saveenv The setenv functionality is also available as a Linux utility with the same usage To display U Boot variables at the shell prompt use the additional getenv utility or issue the setenv command without parameters 3 2 3 Configuring U Boot Environment Variables for the Shelf Manager When U Boot is started for the first time the following default env
371. the Shelf Manager configuration file and the IPMI LAN configuration parameters The value of 0 0 0 0 for an IP address type of configuration parameter is interpreted as undefined The Shelf FRU Information should specify the RMCP IP address the Shelf Manager uses it and propagates it to the LAN configuration parameters RMCP is available in the absence of the Shelf FRU Information only if the configuration parameter RMCP_WITHOUT_SHELF_FRU is defined and set to TRUE In this case the Shelf Manager uses the IP address stored in the channel parameters If there are no stored channel parameters it uses the IP address specified in the boot parameters of the ShMM Some Shelf Manager configuration variables can get their values from the Shelf FRU Info This may be useful to users who need to associate certain configuration parameters with a specific instance of a shelf In that case even if ShMMs or ShMM carriers are moved between shelves the configuration parameters specified in the Shelf FRU Information stay with the shelf and override the parameter values specified in configuration files on the ShMM If the same variable is specified in a configuration file on the ShMM and in the Shelf FRU Information the value from the Shelf FRU Information overrides the value from configuration files Note that configuration files are parsed early during the Shelf Manager initialization while the Shelf FRU can be found substantially later Not all variable
372. the ipmb_traced process ID using the ps command ps 39 ttysod root 0 S ipmb_traced w u var ipmb_traced log e Send the SIGINT signal to the ipmb_t raced process kill 2 39 4 Copy the trace file to the analyzer host cd var ftp 172 16 1 24 Connected to 172 16 1 24 220 ts FTP server Version wu 2 6 2 5 ready ftp gt put ipmb_traced log 5 On the analyzer host start the GUI C gt wireshark 6 Inthe main menu go to File then Open and browse to the trace file Open the trace file Release 3 2 0 177 J uly 31 2013 Pigeon Point Shelf Manager User Guide Figure 8 Open Capture File Window Wireshark Open Capture File Ed home serge Places Name v Modified E g Search Desktop 07 15 2008 Recently Used B Documents Wednesday B root E Download 04 18 2008 E ffmpeg export 2008 07 07 07 08 2008 B libmad 0 15 1b 07 07 2008 7E libpcap 0 9 8 06 23 2008 B Linux Wireless HowTo 04 21 2008 B madwifi 0 9 4 04 19 2008 E mpeg2dec 0 4 1 07 08 2008 B Music 04 18 2008 _ E Pictures Today at 14 57 Adds BSMOVE Ime tects nannan 7 Desktop directory Filter V Enable MAC name resolution Size C Enable network name resolution Enable transport name resolution Elap Cancel This will open the trace in the GUI main window Figure 9 GUI Main Window Unattended mode ipmb_trac File Edit View Go Capture Analyze Statistics H
373. the kernel command line to automatically configure the corresponding kernel network interface This variable is not assigned by default ipdevice Device corresponding to ipaddr Default value is etho ipldevice Device corresponding to Sipladdr On ShMM 500 and ShMM 1500 the default is eth1 On ShMM 700 usb0 is the default ip2device Device corresponding to ip2addr Not assigned by default but can be assigned if more than two network interfaces are used on the ShMM e g a USB or serial based network interface can be designated this way kernel start The absolute starting address of the kernel image in Flash This variable is set automatically by U Boot during bootstrap Note on ShMM 700 this variable contains the absolute starting address of the kernel image in RAM This variable is used in ubifs nfs net load_ubifs load_ubifs_alt to specify the address in physical memory where the kernel image should be loaded and where it is expected to be during system boot The default value is 0x42000000 load_ubifs ShMM 700 only U Boot command that loads current Linux kernel and RFS images from the Flash drive into RAM Images are selected on the basis of the image_sel variable value The kernel image is loaded at the loadaddr address and the RFS image is loaded at the rfsaddr address This command is invoked by ubifs explicitly Release 3 2 0 2il July 31 2013 ENVIRONMENT VARIABLE load
374. ther the ShMM 1500 or the ShMM 700 The backup Shelf Manager recognizes the departure of the active Shelf Manager when the Remote Healthy or Remote Presence bit becomes 0 The Remote Presence bit monitors the presence of the peer Shelf Manager this bit changing to 0 on the ShMM 500 and ShMM 1500 means that the board hosting the peer Shelf Manager has been removed from the shelf For the ShMM 7OOR this bit changing to 0 indicates that the peer Shelf Manager has been reset or removed from the shelf The Remote Healthy bit is set by the peer Shelf Manager during initialization this bit changing to 0 means that the remote Shelf Manager has become not healthy typically has been powered off or reset Another situation that needs some action from the backup Shelf Manager is when the TCP connection between the Shelf Managers gets closed This happens when either the communication link between the two Shelf Managers is broken or the shel fman process on the active Shelf Manager terminates in a voluntary or involuntary way or due to a software exception Also since the keepalive option is enabled on the TCP connection it closes shortly after the active ShMM is switched off or reset In the case of Shelf Manager termination itis possible that the TCP connection is closed before the Remote Healthy bit becomes 0 So in order to determine why the TCP connection closed the backup Shelf Manager samples the state of the Remote Healthy bit immediately a
375. this case If this parameter is FALSE all boards in the shelf are powered down if board s that caused the Critical Alert cooling state fail to power off also the power level for all boards in the shelf is decreased if decreasing power level fails for board s that caused the Major Alert cooling state Yes Release 3 2 0 52 J uly 31 2013 INITIALIZATION _SCRIPT Pigeon Point Shelf Manager User Guide TYPE String 256 DEFAULT DESCRIPTION This parameter specifies a command line that is executed by the active Shelf Manager during initialization This initialization script can perform additional platform specific initialization of the Shelf Manager when itis started the Local Healthy and Active bits in CPLD are already set The Shelf Manager waits for the completion of the initialization script The initialization script is run only during the startup of the Shelf Manager itis not run after a Switchover CAN BE OBTAINED FROM SHELF FRU INFO No INITIAL_FAN_LE VEL Number This parameter specifies the initial fan level that the Shelf Manager applies to fan trays Usually fan level values are in 0 15 range where 0 is the Slowest and 15 is the fastest possible fan speed This parameter has an alias CTCA_INITIAL_FAN_LEVEL for CompactPCI shelves INITIAL _SLOW_L INK_DELAY Number The initial delay in seconds before the Shelf Manager starts testing the inte
376. thus becomes associated with a different hub board In this case from the perspective of the hub board the ShMC switchover becomes non transparent which may be undesirable if for example each hub board hosts a System Manager The configuration parameter SWAPPED_CROSS_CONNECTS can be used to avoid this non transparency If the configuration parameter SWAPPED_CROSS_CONNECTS is defined and set to TRUE and if the hardware address of the ShMC is odd the Shelf Manager swaps the values of the configuration parameters RMCP_NET ADAPTER and RMCP_NET_ADAPTER2 AS a result after a switchover the RMCP addresses on the newly active Shelf Manager correspond to the Same hub boards as before the switchover Release 3 2 0 78 July 31 2013 Pigeon Point Shelf Manager User Guide 3 4 2 4 Bonded Usage of the Two Network Interfaces The two physical network interfaces ethO and eth1 can be used in a bonded fashion using the kernel bonding driver that joins several physical devices into one logical device and selects an active device for data transfers In this mode both physical network interfaces that can be used for outside connections are enslaved by the bonding driver at system startup and the Shelf Manager uses the single bonding interface for RMCP based communication This mode is useful when both devices are connected to the same network switch hub and it does not matter which device is used at any given moment as long as the Shelf Manager remains availab
377. tically by U Boot during bootstrap rootpath ShMM 700 only This variable specifies the NFS root to be used by the Linux kernel when booting the ShMM via the bootcmd_nfs command Default value is root fs quiet Instructs the kernel upon bootup not to print progress messages to the Serial console Default is quiet ramargs Sets the kernel command line in the boot args variable as appropriate for the root filesystem to be mounted from a ramdisk Note not present on ShMM 700 ramdisk Specifies what r s image should be used by the net boot option ramsize Size of the system memory in bytes Default setting calculated from the SDRAM configuration encoding in the build time configuration block Note not present on ShMM 700 re_ifconfig Allows the etc rc script to set up the IP address instead of shelfman Defaultis n allow shel fman to set up IP addresses Release 3 2 0 29 July 31 2013 ENVIRONMENT VARIABLE re2 Pigeon Point Shelf Manager User Guide DESCRIPTION Specifies secondary RC script that is to be invoked This is the carrier Specific startup script Default is etc re shmm700 hpdl1 on ShMM 700 and etec re carrier3 on ShMM 500 and ShMM 1500 or other appropriate script for the relevant target platform This variable must be set to a carrier Specific value matching the carrier on which the ShMM is installed rfs_start The absolute starting address of the r
378. till alive 7 The reboot happened due to an error during the Shelf Manager startup 8 The reboot was caused by the ShMM hardware watchdog 9 The reboot was initiated by software a reboot system call 10 The ShMM has been power cycled 11 The reboot happened because of an SDRAM error 12 The reboot was caused by the IP MI Cold Reset command The sensor reading byte is always 0 and does not have any meaning NOTE in case of a Shelf Manager reboot initiated by software this sensor may report two different reasons based on system shutdown speed If the system shuts down within the default setting of 2 seconds i e while the watchdog timer has not expired the reboot reason will be a reboot call state mask 9 Otherwise the reboot reason will be the hardware watchdog To avoid this non deterministic result itis possible to increase the WATCHDOG_TIMEOUT configuration variable value for example WATCHDOG_TIMEOUT could be increased to 5 seconds 4 5 Operation in Shelves with Radial IPMB 0 Some shelves implement radial control of IPMB 0 links in the shelf In that case the segment of IPMB 0 leading to each IPM controller in the shelf can be turned on and off individually by the Shelf Manager This applies individually to both the IPMB A and IPMB B portions of IPMB 0 The operation of the Shelf Manager in such shelves is different from the shelves with a simple bused IPMB 0 Release 3 2 0 142 July 31 2013
379. tion 2 35 18 834 199 2 35 18 836 199 8 0 192 1040 2 35 18 838 199 8 0 64 1041 2 35 18 841 199 35 18 843 199 35 21 844 199 failed 148 235 21 846 199 135 21 883 199 sfully value 8 13522152951 199 x2C 135221 995 199 gnored 135 22 045 199 gnored 135722 049 199 at 0x2C 35 122 054 199 at 0x2C 35 22 058 199 x2C 235 22 062 199 at 0x2C 2 35 22 067 199 at 0x2C 1 35 22 071 199 at Ox2C 3 2 0 Pigeon Poi raw LOCK L nt Shelf og print Manager ver 32 0 Built on buffer at 1006c720 Pthread lock log print buffer at 10070f 70 Reading co Shelfman Custom SDR Carrier se shm_crypt Device GUI Netmask no Interface Input redu Ou Ac Co nnecting nfigura tion file etc shelfman conf Using E initia EPROM for Shelf FRU Set watchdog timeout to 2 seconds lizatio error 0 n file is absent t to PPS o_init initia lization succeeded D tput redundancy socket successful tos B4A841B0 5EBD 11DE 0080 t specified in two adapter redundancy defaulting to 255 255 255 128 usbl activated successful ndancy socket successfully bound to ly ly bound to tive listening thread created successfully 192 168 0 65 1040 x Running in Active mode connection to Redundancy protocol initialized successfully Reboot reason sensor was registered Shelfman ADM1026 ADM
380. tion to Terminal oriented Wireshark The Wireshark tool suite also includes a command line oriented utility called tshark This utility can be used to analyze traffic using only a command line approach when running a GUI application is not practical The t shark tool has the same overall functionality as the GUI but lacks its interactive capabilities Section 5 4 also includes step by step examples of using tshark which is fully documented in the corresponding manual pages man tshark To getonline help use tshark 5 3 3 Installing IPMI Analyzer Software The IPMB trace daemon is included in the software image on the ShMM by default and does not require any specific installation The Wireshark client applications and much more regarding Wireshark can be obtained from www wireshark org The netcat nc utility is needed as a mediator between the IPMB trace daemon and Wireshark tshark for real time IPMI analysis with IPMB trace daemon operating in controlled mode Linux distributions typically include this utility Its Windows variant ne exe can be downloaded from htto www securityfocus com tools 139 The source code of the utility is available at http netcat sourceforge net download php 5 4 Collecting and Analyzing a Trace Using the GUI and Command Line Tools An IP MB trace can be collected in either of two ways 1 unattended mode where it is stored in a file during collection and analyzed later or 2 controlled mode w
381. to the network context in which the ShMM is used 3 2 4 Establishing the Secondary RC Script The secondary RC script gets invoked when the system configuration is established during the boot process It is called from the primary RC script etc re The secondary script is a carrier Specific startup scriptand is etc re carrier3 by default on ShMM 500 1500 and etc re shmm700 hpd1 on ShMM 700 or some other script that is appropriate for that platform A typical name for this scriptis etc re lt target_platform gt Release 3 2 0 33 July 31 2013 Pigeon Point Shelf Manager User Guide The name of this carrier specific startup script is defined by the U Boot environment variable re2 The variable re2 is the one environment variable that must definitely be changed fora ShMM and its carrier to work properly in a shelf The RC2 script sets up environment variables CARRIER and CARRIER_OPTIONS These variables inform the Shelf Manager about the carrier on which itis installed and define carrier Specific options as necessary for each supported carrier By default the values of these environment variables are propagated to the corresponding configuration variables see Table 7 Shelf Manager Configuration Parameters CARRIER and CARRIER_OPTIONS The configuration variables in their turn are retrieved and used by the Shelf Manager The U Boot variable start_rc2_daemons instructs the secondary startup script to start or not start the daemons snmpd
382. to boot during bootup The variable is set automatically by U Boot during bootstrap Release 3 2 0 26 July 31 2013 ENVIRONMENT VARIABLE io_config Pigeon Point Shelf Manager User Guide DESCRIPTION ShMM 500 only Determines if the Programmable Serial Controllers PSCs used for the IPMB 0 interface on the ShMM 500 are configured for the dual slave address configuration y n Default setting y If Setting is not y the IP MB 0 interface does not work properly ipaddr IP address used by the primary on chip Ethernet interface This variable configures the network interface specified by ipdevice automatically ifthe re_ifconfig variable is set to y Note that the system startup script sets the least significant bit of this variable to the least significant bit of the Hardware Address for the ShMM carrier that is if the Hardware Address is an even value the last bit in the IP address is set to 0 otherwise it is set to 1 This is done in the startup script etc netconfig to support coordinated IP address configurations on redundant ShMMs To disable this functionality simply remove the etc readhwaddr file ipladdr IP address used by the secondary Ethernet interface This variable can be passed as a part of the kernel command line to automatically configure the corresponding kernel network interface ip2addr IP address used by the third network interface if present This variable can be passed as a part of
383. transition to Non recoverable The current cooling State is now Critical Alert 137 Fans State Severity 7 Management 00h transition to OK The fans Subsystem State is Normal no thresholds are Health 28h crossed on fan tachometer sensors 01h transition to Non Critical from OK The fans state is now Minor Alert non critical thresholds are crossed for some tachometer Release 3 2 0 151 J uly 31 2013 Pigeon Point Shelf Manager User Guide sensors the previous fans state was Normal 02h transition to Critical from less severe The fans state is now Major Alert critical thresholds are crossed for some tachometer Sensors the previous fans state was either Normal or Minor Alert 04h transition to Non Critical from more severe The fans state is now Minor Alert the previous fans state was either Major Alert or Critical Alert non recoverable thresholds are crossed for some tachometer sensors 05h transition to Critical from Non recoverable The current fans State is Major Alert the previous fans state was Critical Alert 06h transition to Non recoverable The current fans State is now Critical Alert 4 8 Deadlock Detection The Shelf Manager contains built in facilities for detecting internal deadlocks A deadlock happens if two or more threads in the Shelf Manager application acquire locks in an incorrect order so that one thread owns the lock requested by
384. tself will be faster An example of such a partial upgrade is shown as Example 2 at the end of this section The backend utility has the following synopsis Only the command line parameters that are intended to be used by external programs are shown here the remaining parameters not shown here are for the exclusive use of the rupgrade utility Output the current image set index 0 or 1 to stdout backend c backend get current Confirm and finalize the current upgrade backend confirm Cancel abort the current upgrade backend cancel Strobe the reliable upgrade watchdog timer backend strobe Help mode output a brief help to stdout backend h backend help Get Version mode output the version of the utility to stdout backend version Table 29 Standard Source File Names for Upgrade Components COMPONENT NAME SOURCE FILE NAME U Boot sentry shmm700 u boot Linux kernel sentry shmm700 kernel RFS sentry shmm700 rfs Application Package sentry shmm700 app The image upgrade works as follows For each of the specified source images the image is copied to the ShMM via the protocol specified in the URL FTP TFTP etc The image is copied to the RAM file system of the ShMM specifically the copy is to the tmp directory which is cleaned up after each reboot and after that it is validated Image validation involves verification of the image checksum and is performed Release 3 2 0 229 July
385. tshark r ipmb_traced log 1 Readi 2 Readi 3 Readi 4 Readi 5 Readi 6 Readi 7 Readi 8 Readi 9 Readi 10 Readi 11 Readi 12 Readi 0 000000 ng seq 0x28 0 079507 ng seq 0x28 0 579642 ng seq 0x29 0 585695 ng seq 0x29 0 679566 ng seq 0x2a 0 686105 ng seq 0x2a 0 698566 ng seq 0x2b 0 713453 ng seq 0x2b 0 724908 ng seq 0x2c 0 739889 ng seq 0x2c 0 751980 ng seq 0x2d 0 767020 ng seq 0x2d I2C 1 gt 0x41 T2C 1 gt 0x10 I2C 2 gt 0x41 I2C 2 gt 0x10 I2C 1 gt 0x41 I2C 1 gt 0x10 I2C 2 gt 0x41 I2C 2 gt 0x10 I2C 1 gt 0x41 I2C 1 gt 0x10 I2C 2 gt 0x41 I2C 2 gt 0x10 IP IP IP IP LE IP IP IP IP TE TP IP I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA I ATCA Req Rsp Req Rsp Req Rsp Req Rsp Req Rsp Req Rsp Ge Ge Ge Ge Ge Ge Ge Ge Ge Ge Ge Ge Sensor Sensor Sensor Sensor Sensor Sensor Sensor Sensor Sensor Sensor Sensor Sensor Use the v option to obtain a detailed dump of the messages use x option to obtain a hexadecimal dump of messages as well tshark r ipmb_traced log x V Frame 1 8 bytes on wire 8 bytes captured Arrival Time Jun 24 Time delta from previous captured frame Time delta from previous displayed frame Time sin
386. tured by tshark and sent to the command output For example ne 192 168 1 62 19000 Capturing on Standard input 1214325501 948358 Reading seq 0x27 1214325501 999629 Reading seq 0x27 1214325502 002796 Reading seq 0x28 1214325502 017845 Reading seq 0x28 1214325502 038082 Reading seq 0x29 1214325502 038245 Reading seq 0x29 Release 3 2 0 tshark i I2C 1 I2C 1 I2C 2 I2C 2 I2C 1 I2C 1 gt 0x41 gt 0x10 gt 0x41 gt 0x10 gt 0x41 gt 0x10 183 IP IP IP IP IP IP I A I A I Al I A I Al I A TCA TCA TCA TCA TCA TCA Req Rsp Req Rsp Req Rsp Get Ge Get Ge Get Get Sensor Sensor Sensor Sensor Sensor Sensor July 31 2013 Pigeon Point Shelf Manager User Guide 1214325502 041232 I2C 2 gt 0x41 IPMI ATCA Req Get Sensor Reading seq 0x2a To stop the capture session terminate the tshark process e g by pressing C 5 4 3 Collecting an IPMI Trace in Board Trace Mode If some board in the shelf supports an HP M 2 IPMI trace facility the trace daemon can be used as a bridge between the board and the trace analyzer application To support this mode the target IPM controller must be specifically configured to begin tracing the specified IPMI messaging interfaces when an RMCP session with the HPM 2 IPMI Trace Payload is established to a certain payload instance number and
387. turned off after the boot from the candidate locations but before the reliable upgrade is finalized After turning the power back on the rollback to the confirmed images occurs This reliable upgrade is initiated from the serial console The default copy mode is used copy SSH keys and var nvdata All four images are assumed to be already in tmp Release 3 2 0 a a a a after storage umount a a P fter marking image set invalid fter cleaning previous apps set fter writting new apps set PASSED D ED etc Write confirmed 0 gt 0 OW Write candidate 1 gt 1 OW Write upgrade watchdog 0 gt 1 ALLOW Write upgrade watchdog 1 gt 2 ALLOW Write confirmed 0 gt 1 ALLOW Write upgrade_state in progress 2 gt Write upgrade watchdog 1 gt 0 ALLOW fter u boot burn PASSED fter kernel burn PASSED fter rootfs burn PASSED upgrade stages log PASSED ASSED PASSED PASSED PASSED 243 July 31 2013 Pigeon Point Shelf Manager User Guide rupgrade skip checksums k file tmp sentry kernel r file tmp sentry rfs u file tmp u boot bin a file tmp sentry app lt INFO gt frontend c 449 Downloading file file tmp u boot bin lt INFO gt frontend c 449 Downloading file file tmp sentry kernel lt INFO gt frontend c 449 Downloading file file
388. ual LANs VLANs in accordance with version 2 0 of the IP MI specification That is RMCP RMCP traffic to and from the Shelf Manager can be passed over a Specific virtual LAN the ID of that virtual LAN is stored in the LAN configuration parameters and is automatically added to the outgoing RMCP RMCP packets and Stripped from the incoming packets in accordance with the IEEE 802 1q specification Separate VLAN IDs are supported for the two LAN channels 1 and 2 that can carry RMCP RMCP traffic To assign a VLAN ID to a channel use the IPMI Set LAN Configuration command the CLI setlanconfig command or the CLI analog in the SNMP or WEB interface Changing the VLAN ID may disrupt current RMCP RMCP communication on the affected channel so it should be used with caution The VLAN ID is a 12 bit unsigned number the value 0 indicates that virtual LANs are not used on the channel Once set the VLAN ID is stored persistently in the LAN configuration parameters on the ShMM Flash file system Example Get the current VLAN ID settings for channel 1 where no VLANs are currently in use clia getlanconfig 1 vlan_id Pigeon Point Shelf Manager Command Line Interpreter 802 1q VLAN ID 0 disabled Set the VLAN ID for channel 1 to a different value Clia setlanconfig 1 vlan_id enabled 5 Pigeon Point Shelf Manager Command Line Interpreter VLAN ID set successfully The IPMI 2 0 specification allows distinct VLAN IDs to be used for the outgoing
389. udes the following actions obtaining the images to copy locally or via the network copying the images to the provisional Flash terminating the Shelf Manager instance running on the ShMM 500 if any copying non volatile data to the provisional Flash resetting the ShMM and instructing it to boot from the provisional Flash However if the specifier R has been used in the command line only the last action in the list above occurs that is no image copying takes place Because of the last step an invocation of rupgrade_tool s typically does not return and instead resets the ShMM If rupgrade_tool s does return it indicates that the reliable upgrade procedure has failed and was terminated before proceeding to reset the ShMM in order to boot from the provisional Flash Before the first step of the upgrade procedure is initiated by the utility it removes the var upgrade status file In other words the status of the previous upgrade procedure session if any is lost and overwritten by the status of the new upgrade procedure session as soon as rupgrade_tool s is called There can be one or more lt dst gt lt sre gt specifiers in a call to rupgrade_tool s Each such specifier defines the name of a to be installed upgrade image file and where the file is to be installed in the Flash of the ShMM lt dst gt defines the destination of a newly installed upgrade image and can be any of the following e u Upgrade the U Boot
390. uit Data Bus I2C 1 Target address 0x10 Flags 0x00000000 Intelligent Platform Management Interface Response to 1 Responded in 0 079507000 seconds Header Get Sensor Reading Response from 0x82 to 0x20 Target Address 0x20 Target LUN 0x00 NetFN Sensor Event Response 0x05 00 Target LUN 0x00 0001 01 NetFn Sensor Event Response 0x05 Header checksum Oxcc correct Source Address 0x82 Source LUN 0x00 SeqNo 0x28 00 Source LUN 0x00 1010 00 Sequence Number 0x28 Command Get Sensor Reading 0x2d Completion code Command Completed Normally 0x00 Data Sensor Reading 50 Event Messages Enabled Sensor scanning Enabled 1 Event Messages Enabled l Sensor scanning Enabled lt 0 Reading status unavailable False Threshold comparisons assertions byte 0 0 Reserved State 7 asserted False O Reserved State 6 asserted False 0 At or above UNR threshold State 5 asserted False 0 At or above UC threshold State 4 asserted False 0 At or above UNC threshold State 3 asserted False Release 3 2 0 180 J uly 31 2013 Pigeon Point Shelf Manager User Guide 0 At or below LNR threshold State 2 asserted False 0 At or below LC threshold State 1 asserted False 0 At or below LNC threshold State 0 asserted False Data checksum Oxbf c
391. ult value is console ttySP0 115200n8 mtdparts m25p80 1 512K uboot0 512K uboot1 256K envvars0 2 56K envvars1 256K ubifs_cache 30976K boot 32M user quiet fec_mac ethaddr root dev ram bootargs_comm on ShMM 700 only U Boot command that sets up common kernel bootargs parameters It s called by bootargs_initrd and bootargs_nfs commands to setup console mtdparts quiet fec_mac kernel arguments bootargs_init rd ShMM 700 only U Boot command that sets up kernel bootargs string so that the kernel shall be booted with RFS image stored in RAM It executes bootargs_common command and then extends bootargs string with root dev ram kernel parameter bootargs_nfs ShMM 700 only U Boot command that sets up kernel bootargs string to prepare the system to be booted with NFS volume as RFS It executes bootargs_common command and then extends bootargs string with the following NFS parameters root dev nfs ip ipaddr ethO off nfsroot serverip rootpath v3 tcp So the kernel shall use ipaddr to configure eth0 interface and access serverip host to use its exported rootpath Share as root bootcmd U Boot command executed to accomplish auto booting On both ShMM 500 and ShMM 1500 normally this is something similar to bootm BFB00000 BFC40000 which starts the Linux image Stored in Flash On ShMM 700 the boot command is more complex because of the reliable A2F upgrade support and because boot
392. umber 0x00 M3 gt M4 cause 0 lt I gt 14 35 23 853 217 ipmc_cooling_scan_sensors Reread sensors on the SA 0x20 lt I gt 14 35 23 864 229 SA Oxfc FRU 0 is ACTIVATING lt I gt 14 35 23 881 218 Adjust power for 20 00 from 0 to 5000 lt I gt 14 35 23 882 218 Sending sync point lt I gt 14 35 23 891 233 SA 0x20 FRU 0 is OPERATIONAL lt I gt 24 35 lt 23 905 229 Tasklet ACTIVATE fc 0 is_local_address 1 lt W gt 14 35 23 907 229 Local FRU 0 at SA OxFC is not listed in Shelf FRU Info still activate lt I gt 14 35 23 918 212 Controller 20 FRU 1 ATCA state set to M2 prev M1 cause 2 locked 0 lt I gt 14 35 523 920 212 Controller 20 FRU 2 ATCA state set to M2 prev M1 cause 2 locked 0 lt I gt 14 35 23 942 221 sdrrep Skipping 20 read already in progress lt I gt 14 35 23 944 221 sdrrep Skipping FC read already in progress lt I gt 14 35 23 947 214 Hot Swap event SA 0x20 FRU 1 sensor_number 0x02 M1 gt M2 cause 2 lt I gt 14 35 23 963 214 Hot Swap event SA 0x20 FRU 2 sensor_number 0x03 M1 gt M2 cause 2 lt I gt 14 35 23 973 234 SA 0x20 FRU 1 is ACTIVATING Release 3 2 0 134 July 31 2013 Pigeon Point Shelf Manager User Guide lt I gt 14 35 23 978 230 sdrrep_full_ipmc_sdrs_update_thread for 20 lt I gt 14735723979 213 Controller FC FRU 0 ATCA stat
393. urrent image set has number 1 the upgrade utility saves the contents of U Boot image 0 to a file and copies the U Boot image from partition 1 to partition 0 This is done to ensure that if the upgrade process is interrupted the ShMM 700 can still recover by booting from partition 0 see below e The upgrade utility reboots the ShMM Release 3 2 0 231 July 31 2013 Pigeon Point Shelf Manager User Guide When U Boot sees the a2 upgrade dat file in the boot partition at startup it initiates an A2F upgrade procedure via the A2F TAG pins This procedure is performed with Flash remapping logic turned off so that U Boot has direct access to the Flash and U Boot image 0 is always the active one Also the built in i MX287 watchdog timer is enabled during the upgrade so that if the upgrade process gets stuck the ShMM 700 still has a potential to recover If the upgrade is successful U Boot boots Linux from the current image set When Linux boots it performs some basic tests and if everything works as expected it disables the built in i MX287 watchdog timer and turns on the Flash remapping logic It then restores the contents of the first U Boot partition if the backup file is present and renames the A2F upgrade file on the boot partition to a2 auto rollback dat to complete the upgrade process Ifan a2 auto rollback dat file was already present itis renamed as a2 manual rollback dat before overwriting it with the current upgrade fi
394. urther Note that the initialization script is run only at the startup of the active Shelf Manager and is not called after a Switchover The corresponding command line can also be specified as the value of the configuration parameter INITIALIZATION_SCRIPT in the Shelf Manager configuration file Release 3 2 0 130 July 31 2013 Pigeon Point Shelf Manager User Guide To run the Shelf Manager from the command line type the following daemon f shelfman lt options gt For example here is a typical log of the Shelf Manager starting with options that cause it to route log information both to the console and to the log file and to use EEPROMs for Shelf FRU Information daemon f shelfman lcs sf lt gt 14 May 17 lt gt 14 lt gt 14 KKK lt I gt 14 lt I gt 14 lt I gt 14 lt w gt 14 lt I gt 14 lt I gt 14 lt I gt 14 001849 lt I gt 14 config lt I gt 14 lt I gt 14 192 16 lt I gt 14 192 16 lt I gt 14 lt I gt 14 lt I gt 14 backup lt I gt 14 lt I gt 14 succes lt I gt 14 at O lt E gt 14 128 i lt E gt 14 128 i lt I gt 14 0 4 lt I gt 14 O25 lt I gt 14 at o0 lt I gt 14 0 7 lt I gt 14 0 8 lt I gt 14 0 9 Release 2 35 17 338 199 2013 22 00 07 235 17 348 1 99 s357 T7950 199 2 35 17 362 199 2 35 17 384 199 35e L77392 199 1357184810 199 2 35 18 812 199 35 18 818 199 2 35 18 824 199 OOOEEC 2 35 18 829 199 ura
395. values of the following environment variables to locate the information CARRIER_HPDL Binary HPDL definitions for the carrier CARRIER_SDRS Binary SDRs for the carrier sensors CHASSIS_HPDL Binary HP DL definitions for the shelf CHASSIS_SDRS Binary SDRs for the shelf sensors Therefore when HPDL definitions cannot be placed in some part of a FRU Information EEPROM they are placed in the RFS and the environment variables listed above provide the locations inside the RFS somewhere below the etc directory Release 3 2 0 102 July 31 2013 Pigeon Point Shelf Manager User Guide To override the default HPDL definitions stored in the RFS a user should copy the files generated by the HPDL and SDR compilers to the directory var nvdata on the ShMM file system by using for example FTP The directory var nvdata is in the persistent part of the ShMM file system thatis the file content persists across ShMM reboots 3 6 4 Deploying HPDL Data to FRU Information Areas FRU Information areas are the recommended deployment location for HPDL data and SDRs for the carrier and shelf chassis in carrier FRU Information and shelf FRU Information respectively FRU Information for a specific FRU Fan tray Power Entry Module can also be used it provides FRU specific HPDL data that amends definitions for that FRU in the shelf chassis HP DL description FRU Information is usually stored in EEPROM type devices HPDL data and SDRs are
396. ve Linux kernel e 8 Inactive U Boot e 9 Inactive RFS e 10 Active application image ShMM 700 only e 11 Inactive application image ShMM 700 only Release 3 2 0 167 July 31 2013 Pigeon Point Shelf Manager User Guide TEST NAME TEST PARAMETERS NAME TYPE TEST DESCRIPTION 1 DESCRIPTION 2C TEST 6 ous INT32 2C bus The test tries to access the specified I2C Bus Test number to test l2C adapter device and enumerate all Acceptable values are 0 l2C devices present on that bus by and higher bus 0 reading from all possible 2C represents the main master addresses on the bus and checking only 12C bus going to the which addresses are acknowledged ShMM carrier and often beyond the carrier to the Shelf other bus numbers are platform specific Default value is 0 PMB TEST 7 ous INT32 IP MB bus The test queries the IPMB bus driver IPMB Bus number to test for extended information about the Test Acceptable values are 0 specified bus which includes state and higher 0 represents flags and error flags This information IPMB A 1 represents is then converted to a textual format IPMB B higher values and reported in the results string represent radial IPMB links Default value is 0 Release 3 2 0 168 July 31 2013 Pigeon Point Shelf Manager User Guide 5 Using the IPMI Analysis Tools The Shelf Manager software includes tools for analyzing the IP MI messaging traffic
397. verification recommended for execution on each boot up not available on ShMM 700 e uart UART verification specific to the Aul550 processor recommended for execution on each boot up e ethernet Ethernet MACs test specific to the Au1550 processor recommended for execution on each boot up e i2c Master Only I C test recommended for execution on each boot up The names of the tests can be used in values of the post_poweron and post_normal environment variables Starting from release 2 4 1 the IPMI command Get Self Test Results directed to the logical Shelf Manager IPMB address 20h returns the results of the POST performed by U Boot at the startup of the ShMM If all tests have passed the status code 0x55 is returned If any tests have failed the Release 3 2 0 117 July 31 2013 Pigeon Point Shelf Manager User Guide device specific failure code 0x59 is returned The third byte of the response contains the following bit mask in that case 7 5 Reserved 4 1b Ethernet test failed 3 1b UART test failed 2 1b U Boot CRC test failed 1 1b C test failed 0 1b Memory test failed 3 12 Configuring External Event Handling Sometimes there is a need for specific IP MI events to be processed by a user defined application or script that is external to the Shelf Manager but still executing on the ShMM This is called external event handling This feature allows extensions of Shelf Manager functionality by u
398. w data Release 3 2 0 172 J uly 31 2013 Pigeon Point Shelf Manager User Guide 5 3 IPMI Trace Analyzer The recommended IP MI trace analyzer for Pigeon Point Shelf Manager applications is a network traffic analyzer program called Wireshark See www wireshark org with its terminal oriented variant t shark Wireshark provides a convenient interface for a complete analysis of layered protocol stack data The Wireshark design allows developers to easily add support for new protocols entire protocol stacks and trace collection data sources in a collaborative manner There are versions of Wireshark for both Windows and Linux environments All these considerations combine to make Wireshark a great framework for IP MI trace analyzer applications The following sections describe the key user interfaces of Wireshark that are relevant to IPMI trace analysis Please consult the extensive user documentation for Wireshark at www wireshark org for more details Previously Pigeon Point Systems had done a private extension of Ethereal the predecessor to Wireshark see www ethereal com to perform the IPMB trace analyzer function Since Wireshark is now recommended for this role only that application is documented here 5 3 1 Introduction to the Wireshark GUI This section provides a brief introduction to the GUI section 5 4 provides a step by step example of using it Figure 6 provides a view of the GUI main window which is composed of thr
399. ware integrity checks run at the application level during initialization indicate that the new software image is not fully functional hardware mechanisms can be invoked to reset the ShMM and revert to the safe confirmed image set Ata higher level the reliable upgrade utility uses these hardware mechanisms and also implements software mechanisms for carrying out reliable upgrades properly For the benefit of the user the progress of the current reliable upgrade session and its final outcome Success or failure is logged in a special file var upgrade status located on the User UBIFS partition The reliable upgrade procedure implemented on ShMM 700 was designed primarily to handle reliable upgrades of the Pigeon Point Shelf Manager firmware In addition however the procedure can be applied to arbitrary embedded applications running on the ShMM To achieve that the procedure supports hook scripts that can be invoked at certain stages of the reliable upgrade procedure and can be used for customization of the procedure This support for hook scripts is briefly discussed in section 8 5 however the rest of this Section 8 mainly concentrates on reliable upgrades in the context of Shelf Manager firmware where hook scripts are not used Also a special procedure is defined for the reliable upgrade of A2F 060 firmware The Microsemi A2F 060 is a hardware component of the ShMM 700 that includes an FPGA fabric and an ARM Cortex M3 subsystem combi
400. way Using synchronous assignment of LAN configuration parameters can indefinitely hold a target FRU in state M3 so it should be used with caution Release 3 2 0 164 July 31 2013 Pigeon Point Shelf Manager User Guide The string value of the configuration parameter BOARD_LAN_ PARAMETERS SYNCHRONOUS represents the list of descriptors for FRUs that require synchronous assignment of LAN configuration parameters This list is meaningful only if the configuration parameter BOARD_LAN PARAMETERS _USE_DHCP is Set to TRUE The following grammar applies to this value lt descriptor list gt lt descriptor gt lt descriptor list gt lt descriptor gt lt descriptor gt lt ipmb address gt lt fru designator lt fru designator gt lt fru device id gt lt fru device id gt lt site type gt lt site number or wildcard gt lt site number or wildcard gt lt site number gt lt site number gt Descriptors in the list are comma separated Each descriptor consists of the IPMB address and the FRU designator either the FRU device ID or the site type and site number of the target subsidiary FRU The IPMB address can be specified alone or the FRU designator can be specified as a wildcard in either case synchronous assignment applies to all FRUs associated with that IP MB address A wildcard descriptor applies to all FRUs in the shelf Ifa FRU designator is specified itis
401. witchover is disabled if the value of this parameter is equal to 1 Another configuration parameter INITIAL SLOW_LINK_DELAY specifies the time interval from the start of the Shelf Manager during which detection of physical disconnection is not performed This allows using Ethernet links that are slow to start and need some time after shelf power up to establish the physical connection In the case of a Shelf Manager switchover as opposed to the IP address switchover mentioned above caused by a broken link a special situation occurs when the first hub board is inserted into a shelf with no hub boards and cross connect Shelf Manager links From the physical connection perspective this is a transition from a situation where both physical network links were broken on both Shelf Managers to a situation where at least one physical network link is present on both Shelf Managers However if in this case due to the specifics of the hub board the link to the backup Shelf Manager comes up earlier than the link to the active Shelf Manager an undesirable switchover may occur To prevent a switchover in this situation the backup Shelf Manager can delay for a specified number of seconds doing a switchover after the original switchover request from the active Shelf Manager if the physical link to the backup Shelf Manager is present If during this time the physical link to the active Shelf Manager also comes up the active Shelf Manager will stop
402. y 31 2013 Pigeon Point Shelf Manager User Guide o Event Data 2 AND Mask O Event Data 2 Compare 1 o Event Data 2 Compare 2 FF Event Data 3 AND Mask FF Event Data 3 Compare 1 o Event Data 3 Compare 2 Note when an event handler is placed in the directory var bin as assumed in the example above care must be taken to preserve the state of that directory across reliable firmware upgrades see 7 6 for details Also note that the var bin directory is erased when the U Boot variable flash_reset is set to y see 6 3 for details 3 12 2 External Event Handler Operation The external event handler can be a user provided binary executable program or a shell script The handler is invoked by the Shelf Manager when an event matching a relevant PEF filter occurs During that invocation a pipe is created between the Shelf Manager and the external event handler This pipe serves as standard input for the external event handler For each event the Shelf Manager submits a single line of text into the pipe in the following format raw raw data alert_string alert string data The raw data are the 16 bytes of the SEL record with the characters representing each byte separated by a colon from the next byte For example raw 36 3 2 27 df 49 0 20 0 4 1 18 6f a1 0 78 alert_string test The external event handler should read its standard input line by line and process each line as a separate event There are possible two
403. y backend is provided that is not intended to be invoked directly by an interactive user This utility performs more specific actions needed for reliable upgrade like starting strobing and stopping the reliable upgrade watchdog timer getting and setting the number of the candidate and confirmed image set etc It can be invoked from system scripts like the system startup script etc re Also the rupgrade utility uses the services provided by the backend utility during its operation The rupgrade utility is fully functional only when called from the superuser root account If run from a non root account the utility can be used for read only operations like getting the upgrade Status As the rupgrade utility operates in upgrade mode see below it logs the status of each action as it progresses through the stages of the procedure to the status file var upgrade status In the case of a failure the reliable upgrade utility adds a record to this file indicating that the reliable upgrade has completed unsuccessfully and then terminates with an appropriate error code The rupgrade utility outputs any informational messages to stdout and any error messages to stderr All messages are also recorded to the var upgrade 1log txt file The rupgrade utility can operate in several modes depending on the command line parameters Upgrade mode rupgrade u lt URL gt uboot lt URL gt k lt URL gt kernel lt URL gt r lt URL gt
404. y completed var upgrade status is an ASCII format file that contains one or more new line terminated records each describing the status of a particular step in the upgrade procedure The format of a record line is as follows lt step gt lt status gt Release 3 2 0 202 July 31 2013 Pigeon Point Shelf Manager User Guide where lt step gt is an integer ranging from 1 to 14 with Step 14 corresponding to a completed upgrade session and status is a human readable string describing status of the current step of the upgrade procedure session Refer to the ML User Guide for a list and explanation of these steps The status file is used by the reliable upgrade utility to maintain a software protocol atop the reliable upgrade procedure hardware mechanisms to reliably determine the status of the upgrade procedure and proceed as appropriate 7 6 Reliable Upgrade Utility A special user space utility is provided that is used for carrying out the reliable upgrade procedure as well as checking the status of the most recent upgrade The utility can be called only from the superuser root account Any attempt to run the utility from a non Superuser account is rejected If called with any of the s c or options the utility is being used to carry out the reliable upgrade procedure While in the upgrade procedure the utility logs to var upgrade status the status of each action it performs as it proceeds through the Steps of the upgrad
405. ytes 2 0MB copied 12 323593 seconds 167 2KB s Extraction result 0 Placed var tmp to ram disk Setting hostname shmm700 Started syslogd and klogd Strobing the reliable upgrade WDT Mounted O etc to etc Calling etc rc shmm700 hpdl etc readhwaddr Board Hardware Address 0x12 etc net etc net etc net etc net etc re settings etc re etc re etc re etc re etc re config etc hosts has valid shmm700 192 168 0 23 entry config Updating etc profile sentry with IP settings config Starting bin inetd config Starting time synchronization script shmm700 hpdl Updating etc profile sentry with specific shmm700 hpdl Starting snmpd shmm700 hpdl Starting httpd shmm700 hpdl Starting Shelf Manager for carrier type PPS700 shmm700 hpdl Command line shelfman sf shmm700 hpdl Strobing the reliable upgrade WDT Release 3 2 0 248 July 31 2013 lt gt May lt gt lt gt kkk 1108711 872 17 2013 22 20 2 11 08 11 913 112083 11 915 lt I gt 11 08 11 963 Pigeon Point Shelf Manager User Guide Built on 122 Pigeon Point Shelf Manager ver 3 2 0 4 122 Lock log print buffer at Oxla2d74 122 Pthread lock log print buffer at 0x1a7594 122 Reading configuration file etc shelfman conf Finally the user checks the status of the reliable upgrade by calling rupgrade s rupgrade s lt INFO gt frontend c 3 7 Last upgrade log

Shelf Manager User Guide

Contents

Download Pdf Manuals

Related Search

Related Contents