AlphaServer GS160/320
Contents
1. La invent AlphaServer GS160 320 CPU Online Addition and Removal Order Number EK GSHPG RM BO1 This document is for system owners and service providers of AlphaServer 160 320 systems It describes how to add and replace CPU modules in a running system and restrictions related to that servicing Hewlett Packard Company August 2002 2002 Hewlett Packard Company UNIX is a trademark of The Open Group in the United States and other countries All other product names mentioned herein may be trademarks of their respective companies HP shall not be liable for technical or editorial errors or omissions contained herein The information in this document is provided as is without warranty of any kind and is subject to change without notice The warranties for HP products are set forth in the express limited warranty statements accompanying such products Nothing herein should be construed as constituting an additional warranty Chapter 1 Overview The AlphaServer GS160 320 systems support the removal and addition of CPU modules while the system is powered on and applications are running This chapter provides background information so that one can safely service these machines and understand requirements about which CPUs can go where Chapter 2 outlines the procedures and gives specific operating system examples of replacing a CPU in a running system Questions and Answers What is the advantage of online service These s2. 320 CPU Addition and Removal Chapter 2 Examples Online addition and replacement of CPUs is used to expand capacity upgrade components and replace failed components while the operating system continues to run Following are the steps required to replace a CPU Step 3 is the physical replacement of the CPU the other steps are performed at the operating system level 1 Take the CPU offline 2 Remove power from the CPU When power is removed the yellow LED on the CPU module lights indicating that the CPU module may be safely removed Physically remove the CPU and insert another CPU module 4 Restore power to the CPU Self test begins and lasts 7 0 seconds The yellow LED on the CPU goes out and the green LED lights 5 Put the CPU online so that the operating system can use it Section 2 1 shows how to handle the CPU modules Examples of the procedure for each operating system are given in the following sections Examples 2 1 2 1 CPU Module Handling Review the information on handling the CPU module before beginning the service procedure Figure 2 1 CPU Removal and Replacement Removal Replacement _ For Module lt Installation J Align Module Color to Frame Color A as Shown CORRECT INCORRECT PK2223 2 2 AlphaServer GS160 320 CPU Addition and Removal CAUTION Always wear an antistatic wrist strap when working on the system See Figure 2 Using both hands one on
3. Examples 2 7
4. console and the SCM monitor so a hard copy record should be stored with the system It would be helpful to have printouts of the SCM show fru and SRM show config commands From these displays one can determine the following e System clock speed setting 9 0 or 9 6 ns e QBB part numbers which indicate the size of B cache supported 54 25043 01 02 4 Mbytes 1 6 AlphaServer GS160 320 CPU Addition and Removal 54 25045 01 02 4 or 8 Mbytes 54 30630 01 02 4 8 or16 Mbytes SCM Example SCM_EO gt show fru FRUname QBBO QBBO QBBO QBBO QBBO QBBO QBBO QBBO 0 E Part part of display omitted PSM PWR AUX CPU0 CPU1 CPU2 CPU3 54 25045 01 C01 is the part number for the QBB that supports a 80 00 00 00 00 00 00 00 54 25045 01 54 25074 01 54 25017 01 54 25123 01 B4166 AA A03 B4166 AA A03 B4166 AA A03 B4166 AA A03 col L01 F01 E01 maximum of 8 Mbytes cache All CPUs in QBBO are B4166 CPUs SRM Example o 2 P00 gt gt gt show config QBB 0 Quad Switch Duplicate Processor 0 Processor 1 Processor 2 Processor 3 1 Tag Hard QBB 0 CPU 0 CPU 1 CPU 2 CPU 3 Up To 4 MB Caches 4 MB Cache 4 MB Cache 4 MB Cache 4 MB Cache Serial SM01300025 NI94770176 NI93970911 NI94171051 SW01040057 SW04300330 SW01040058 SW05100041 Model Other Alias Misc QSA rev 4 QSD revs 0 0 0 0 DTag EV67 EV67 EV67 EV67 rev
5. e various power and mechanical hazards in the system Because these systems use a great deal of power caution must be exercised when servicing them Remove all jewelry when working on the systems The use of ESD straps is required Wrist straps are located on the front and rear doors of system cabinet 1 and on the rear door of the power cabinet Table 1 2 Hazards and Safety Features Hazard Risks Safety Features Exposed Areas AC power Flashing metal AC only in the power cabinet None Shock AC loads fully enclosed Death Securely mated wiring system 48 VDC Flashing metal Securely mated wiring system QBB backplane power Welding of jewelry 48V isolated from cabinets H switch backplane heat Covers CPU module Current limit Power modules Low Flashing metal Short circuit protection QBB backplane voltage Welding of jewelry Current limit H switch backplane power heat Logic modules Power modules System Severe cuts Grill Exposed only when blowers Loss of fingers lower system box out PCI Low voltage power AC loads fully enclosed Under power supply enclosure hazard Interlock connector on the Battery on backplane standard I O 1 2 AlphaServer GS160 320 CPU Addition and Removal What do I need to know about the system clock CPU variants and the system box QBBs Each CPU variant supports a different cache size To use the larger cache size the CPU needs the QBB that supports the larger cache size even though the faster CPUs can be install
6. each module lever place your index finger on the catch and thumb on the edge of the lever just below above the arrow First squeeze to release the lever and then pull both levers away from the module to release it from the QBB backplane Remove the module Insert the replacement module and then press the levers down CAUTION Module should be moved deliberately without any extra movement back and forth in the slot to avoid the discharge of energy To avoid causing a system error close the levers only when the module is properly seated Wait 10 seconds before moving the module again Examples 2 3 2 2 Tru6 4 UNIX Procedure For more information refer to the Tru64 UNIX manual Managing Online Addition and Removal Example 2 1 Replacing a CPU Module sbin hwmgr offline name CPU2 1 hwmgr CPU2 is now offline sbin hwmgr power off name CPU2 2 hwmgr CPU2 is now powered off sbin hwmgr status component ngood STATUS ACCESS INDICT HWID HOSTNAME SUMMARY STATE STATE LEVEL NAME 4 wild one critical offline off CPU2 Remove the CPU insert another CPU 0 sbin hwmgr power on name CPU2 hwmgr CPU2 is now powered on sbin hwmgr online name CPU2 hwmgr CPU2 is now online sbin hwmgr status component id 4 7 STATUS ACCESS INDICT HWID HOSTNAME SUMMARY STATE STATE LEVEL NAME 4 wild one online available CPU2 2 4 AlphaServer GS160 320 CPU Addition and Removal See Table 1 1 f
7. ed in earlier QBBs Furthermore the CPU derives its operating clock speed from the system clock so you need to know the setting of the system clock Most of these components can coexist in the same system But for optimum performance the faster CPUs should be in a system with the 9 0 ns clock setting Table 1 3 compares the B4125 B4166 and B4168 CPUs and Table 1 4 compares the system boxes Table 1 3 Comparison of CPUs CPU Part B cache Speed Number Size in MHz Comments B4125 4 Mbytes 731 The B4125 must operate at this speed and requires the system clock to run at 9 6 ns B4166 4or8 1001 The B4166 runs at this speed and uses all its avail Mbytes able B cache when all CPUs in the system are B4166 or it is in a system with B4168 CPUs the clock runs at 9 0 ns and the CPUs are in a system box with duplicate tag support of the 8 MB cache If the system box duplicate tag supports only 4 MB cache then only 4 of the 8 Mbytes of B cache on the CPU are used 4or8 940 The B4166 runs at this speed and uses all its avail Mbytes able B cache when all CPUs in the QBB are B4166 or it is mixed with B4168 CPUs but the system clock runs at 9 6 ns If the system box duplicate tag sup ports only 4 MB cache then only 4 of the 8 Mbytes of B cache on the CPU are used Uses 4 of 731 The B4166 runs at this speed and uses 4 Mbytes of its its 8 Mbyte B cache when it is in a QBB with a B4125 CPU and the B cache clock runs at 9 6 ns Conti
8. ents BA51A AA 4 Mbytes Though the B4166 and B4168 CPUs have larger B 54 25043 01 02 caches only 4 Mbytes are used when placed in this backplane BA51A BA 4 or 8 Mbytes When CPUs in this backplane are B4166 or B4168 54 25045 01 02 they use 8 Mbytes of B cache If a B4125 is mixed with other speed CPUs on the same backplane all are configured for the 4 Mbyte B cache size BA51A DA 4 8 or 16 When all CPUs in this backplane are B4168 they use 54 30606 01 02 Mbytes 16 Mbytes of B cache If CPUs are mixed B4125 B4166 or B4168 on the same backplane all are configured for the B cache size of the processor that supports the smallest size 4 or 8 Mbytes In summary All variants of system boxes can coexist in a system The B4125 CPUs run at 731 MHz and require a 9 6 ns system clock B4125 CPUs are not supported with the 9 0 ns system clock The B4166 and B4168 CPUs can operate at different speeds and use different amounts of their B cache depending upon the system box the system clock speed and whether the CPU is configured with the B4125 CPU The B4166 CPU requires that the SRM console firmware be at V6 0 or higher and that the microprocessor firmware be at V6 0 or higher The B4168 CPU requires that the SRM console firmware be at V6 3 or higher and that the microprocessor firmware be at V6 3 or higher Overview 1 5 Are there additional restrictions when adding or changing CPUs when the operating system is running Ye
9. nsert another CPU O CPU3 removed from QBBO CPU3 added to QBBO SET CPU POWER ON 3 SYSTEM I CPUSELECTED a selected CPU is attempting the requested transition QBBO now Testing Step 0 I QBBO PSM30 SysEvent CPU_SYNC_INIT Reg0 768F Regl1 3FFF I QBBO PSM30 SysEvent CPU_PINIT_DONE Reg0 768F Regl1 3FFF 2 6 AlphaServer GS160 320 CPU Addition and Removal sHow cpu WFSI27 a Compaq AlphaServer GS160 6 1001 Multiprocessing is ENABLED Streamlined synchronization image loaded Primary CPU 000 CPU sets Active 0 7 Configure 0 7 Powered Down None Potential 0 15 Autostart 031 Failover None See Table 1 1 for the minimum versions of the operating system firmware and WEBES needed to support online service Also be sure to understand the distinctions and restrictions in Chapter 1 relating to the CPU models and the system boxes Refer to Section 2 1 for instructions on removing CPU modules from the system box Example 2 2 shows the OpenVMS commands used in the online removal and replacement of a CPU module Take the CPU offline Remove power from the CPU The yellow LED on the module lights Check the status of CPU 3 Take the module out of the system box and insert another one Restore power to the CPU which starts the CPU The yellow LED on the CPU goes out and the green LED lights Check the status of CPU 3 OpenVMS automatically adds the CPU to the active set of processors 00000
10. nued on next page Overview 1 3 Table 1 3 Comparison of CPUs Continued CPU Part B cache Speed Number Size in MHz Comments B4168 8 or 16 1224 The B4168 runs at this speed and uses all its available Mbytes B cache when all CPUs in the system are B4168 or B4166 the clock runs at 9 0 ns and the CPUs are ina system box with duplicate tag support of the 16 MB cache If the system box duplicate tag supports an 8 MB cache then 8 of the 16 Mbytes of B cache on the CPU are used Uses 4 of 1148 The B4168 runs at this speed when placed in a QBB its 16 MB that supports only 4 Mbytes of B cache when all CPUs B cache in the system are B4168 or B4166 and the clock runs at 9 0 ns Uses 8 of 1001 The B4168 runs at this speed and uses 8 Mbytes of its its 16 MB B cache when the B4168 is placed in the same QBB as B cache the B4166 and the clock is at 9 0 ns 4 or 8 940 The B4168 runs at this speed when all CPUs in the Mbytes QBB are B4168 or B4166 and the system clock runs at 9 6 ns Ifthe system box duplicate tag supports only 4 MB or 8 MB cache then only 4 or 8 of the 16 Mbytes of B cache on the CPU are used Uses 4 of 731 The B4168 runs at this speed and uses 4 Mbytes of its its 16 MB B cache when the B4168 is placed in the same QBB as B cache the B4125 and the clock is at 9 6 ns 1 4 AlphaServer GS160 320 CPU Addition and Removal Table 1 4 Comparison of System Boxes Cache Size System Box Supported Part Number by the Box Comm
11. or the minimum versions of the operating system firmware and WEBES needed to support online service Also be sure to understand the distinctions and restrictions in Chapter 1 relating to the CPU models and the system boxes Refer to Section 2 1 for instructions on removing CPU modules from the system box Example 2 1 shows the Tru64 UNIX commands that can be used in the online removal and replacement of a CPU module Take the CPU offline All bound processes must first be removed from the CPU to be serviced Remove power from the CPU The yellow LED on the module lights Check the status of CPU 2 Take the module out of the system box and insert another one Restore power to the CPU The yellow LED on the CPU goes out and the green LED lights Put the CPU online Check the status of CPU 2 00000 0 Examples 2 5 2 3 OpenVMS Procedure For more information refer to OpenVMS Version 7 3 New Features Manual Example 2 2 Replacing a CPU Module STOP CPU 3 1 SYSTEM I CPUSTOPPING trying to stop CPU 3 after it reaches quiescent state SET CPU POWER OFF 3 2 SYSTEM I CPUSELECTED a selected CPU is attempting the requested transition SHO CPU WFSI27 a Compaq AlphaServer GS160 6 1001 Multiprocessing is ENABLED Streamlined synchronization image loaded Primary CPU 000 CPU sets Active 0 2 4 7 Configure 0 7 Powered Down 3 Potential 0 2 4 15 Autostart 0 31 Failover None Remove the CPU i
12. s pass pass pass pass 1 1 1 1 2 4 731 MHz 2 4 731 MHz 2 4 731 MHz 2 4 731 MHz that the system clock speed is set at 9 6 ns since the B4125 731 MHz CPUs require the 9 6 ns system clock speed If the display shows a CPU running at 1001 or 1224 MHz then we know the system clock speed is 9 0 If you see a CPU running at 940 or 1148 MHz then you can infer that the system clock speed is 9 6 Overview 0 All CPUs in QBBO are running at 731 MHz From this we can infer 1 7 OpenVMS Example show cpu WFSI27 a Compaq AlphaServer GS160 6 940 1 The CPU is identified as 6 940 From this you can infer that the system clock speed is set at 9 6 since the CPU is not running at full speed 1001 or 1224 MHz What is the limit on the number of cycles a CPU module can be power cycled CPU modules must not be power cycled defined as a power on and a power off sequence more than 1 000 cycles This capability is intended to be exercised as needed not by test procedures For More Information Title Part Number Web Site AlphaServer GS80 160 320 Service Manual EK GS320 SV AlphaServer GS160 320 Upgrade Manual EK GS320 UP Tru64 UNIX Managing Online Addition and Removal AA RPUFA TE http www tru 6 4unix compaq com faqs publications pub page pubs page html OpenVMS Version 7 3 New Features Manual AA QSBFD TE http www openvms compaq com 8000 index html ovmsdocset 1 8 AlphaServer GS160
13. s because the firmware recorded what speed modules were in which slots at system startup you should not change the speed of the CPU in that slot The firmware will not take note of the change in the CPU until the next power cycle Terminology clarification Power cycle Occurs when the SRM or SCM power off console command is issued followed by the SRM power on command Power is turned off and on testing begins and the SRM prompt returns Initialize Occurs when the SRM initialize command is issued it causes the I O adapters to initialize The system must then be booted Reboot Occurs at the operating system level causes an SRM initialize command Reset Occurs when the contro panel Reset button is pushed or when the SRM or SCM reset command is issued This causes a system level hardware reset initiating testing and return of the SRM prompt Power is not turned off Note these additional restrictions e All CPUs in a QBB will run at the same speed all QBBs in a GS system operate from one common reference clock e OpenVMS does not support mixing CPUs in an instance of its operating system a partition with a speed difference of greater than 250 MHz Can I removelreplace the primary CPU No In a partitioned system this restriction applies to each primary CPU in each partition How does a service person determine the system environment so he can proceed with CPU online service Some information must be obtained from the SRM
14. ystems were designed to provide high availability so that the operating systems can continue to run while parts of the machine can be serviced or upgraded Some components can be replaced while the operating system continues to run and power is present in the area where the component is to be replaced In other cases power is removed from the area that is to be serviced but other parts of the system remain powered on The online addition and removal of CPU modules means that a faulty CPU module can be replaced while the operating system and applications continue to run In addition the CPU capacity can be expanded without any downtime What is required for online service of CPUs The system must be running versions of firmware operating systems and WEBES that support this function See Table 14 Also with Tru64 UNIX Compaq Analyze must be running in auto analysis mode to support indictment of faulty CPUs Compaq Analyze is one of the WEBES Web Based Enterprise Services tools OpenVMS does not yet support indictment of faulty CPUs but will in a future release Overview 1 1 Table 11 Minimum Versions Required CPU Tru64 UNIX OpenVMS SCM amp SRM WEBES B4166 V5 1A V7 3 V6 0 V4 0 B4168 V5 1A V7 3 1 V6 3 V4 0 Remove and replace only Who may service these systems Only those with appropriate technical training and experience should attempt to service these systems What precautions are to be taken Table 1 2 lists th
Download Pdf Manuals
Related Search