3 Replies Latest reply on Jan 10, 2017 6:52 PM by Intel Corporation

    R2308GZ4GC/S2600GZ "Pwr Unit Redund" after BIOS update - How to fix Power Supply FRU?

    nomorries

      After updating BIOS/firmware with S2600GZ_GL_EFI_BIOS02060005_ME020107328_BMC0127r9958_FRUSDR113.zip, I get a "Power Unit Redund" warning and can't see any "PS1" entries in "Sensor Readings". Both powersupplies have green lights and the system continues running when I unplug one of them.

       

      When rerunning the update I observed this message in the uEFI shell:

      Power supply 1 detected but unable to identify model.

      Sensors for power supply 1 will not be installed.

      When running the command ipmitool fru read 2 fru2.bin and comparing with the good power supply, I see that the first byte is FF instead of 01.

       

      Power Supply 1 FRU (BAD):

      00000000  ff 00 00 00 01 09 00 f5  01 08 19 c5 44 45 4c 54  |............DELT|

      00000010  41 cb 44 50 53 2d 37 35  30 58 42 20 41 ca 45 39  |A.DPS-750XB A.E9|

      00000020  38 37 39 31 2d 30 30 36  c4 30 31 20 20 d1 45 39  |8791-006.01  .E9|

      00000030  38 37 39 31 44 31 32 31  36 30 33 34 38 36 36 c0  |8791D1216034866.|

      00000040  c0 c1 00 00 00 00 00 3e  00 02 18 75 71 ee 02 d0  |.......>...uq...|

      00000050  06 37 05 28 23 b0 36 50  46 20 67 2f 3f 14 1f 52  |.7.(#.6PF g/?..R|

      00000060  c3 00 00 00 85 01 02 0d  49 a7 01 b0 04 74 04 ec  |........I....t..|

      00000070  04 78 00 00 00 30 f2 01  82 0d 09 67 82 b0 04 74  |.x...0.....g...t|

      00000080  04 ec 04 78 00 00 00 dc  05 ff ff ff ff ff ff ff  |...x............|

      00000090  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

      Power Supply 2 FRU (GOOD):

      00000000  01 00 00 00 01 09 00 f5  01 08 19 c5 44 45 4c 54  |............DELT|

      00000010  41 cb 44 50 53 2d 37 35  30 58 42 20 41 ca 45 39  |A.DPS-750XB A.E9|

      00000020  38 37 39 31 2d 30 30 36  c4 30 31 20 20 d1 45 39  |8791-006.01  .E9|

      00000030  38 37 39 31 44 31 32 31  36 30 33 34 38 36 35 c0  |8791D1216034865.|

      00000040  c0 c1 00 00 00 00 00 3f  00 02 18 75 71 ee 02 d0  |.......?...uq...|

      00000050  06 37 05 28 23 b0 36 50  46 20 67 2f 3f 14 1f 52  |.7.(#.6PF g/?..R|

      00000060  c3 00 00 00 85 01 02 0d  49 a7 01 b0 04 74 04 ec  |........I....t..|

      00000070  04 78 00 00 00 30 f2 01  82 0d 09 67 82 b0 04 74  |.x...0.....g...t|

      00000080  04 ec 04 78 00 00 00 dc  05 ff ff ff ff ff ff ff  |...x............|

      00000090  ff ff ff ff ff ff ff ff  ff ff ff ff ff ff ff ff  |................|

      • I think the bad FRU is the source of the problem. Is this correct?
      • How can I write a correct FRU? I tried to use ipmitool fru write 2 fru2.new.bin, but the memory was write protected.
      • Does the BIOS update overwrite the power supply FRU? Was the BIOS update the source of the corruption, or did it just expose the problem?
        • 1. Re: R2308GZ4GC/S2600GZ "Pwr Unit Redund" after BIOS update - How to fix Power Supply FRU?
          estebanv_intel

          This could be a bad power supply that the FRU information is not been detected. Can you try swapping the power supplies to see if you receive the same message?

          You can also try with the previous firmware revision and see if it exhibits the same behavior.

          • 2. Re: R2308GZ4GC/S2600GZ "Pwr Unit Redund" after BIOS update - How to fix Power Supply FRU?
            nomorries

            I tried swapping the powersupplies now. I define PS1 to mean "the power supply in slot 1" (and similarly for PS2). I will refer to the power supply that was PS1 above as PS_BAD, and the one that was PS2 as PS_GOOD.

             

            1. The first thing I did after swapping, was checking "Sensor Health". Entries for PS1 was still absent in "Sensor Health" (i.e. I couldn't see PS_GOOD), and PS2 (PS_BAD) reported all good.
            2. Then I ran UpdateFRUSDR.nsh from S2600GZ_GL_EFI_BIOS02060005_ME020107328_BMC0127r9958_FRUSDR113.zip. The updater complained
              Power supply 2 detected but unable to identify model.
              Sensors for power supply 2 will not be installed.
            3. When checking "Sensor Health" now, only PS1 (PS_GOOD) is visible.
            4. I added the lines

              SET "PS1_750W"
              SET "PS2_750W"

              to the file master.cfg from the BIOS update. After rerunning UpdateFRUSDR.sh with this modification, both power supplies are visible in "Sensor Health" and I no longer get a "Power Unit Redund" warning.

             

            Since master.cfg deals with FRU data, I think this suggests that the stray FF byte is the problem.

            http://www.intel.com/content/dam/www/public/us/en/documents/product-briefs/platform-management-fru-document-rev-1-2-feb-…

            describes the first 8 bytes as containing the COMMON HEADER, of which the first byte denotes the version. The stray FF byte both gives this record an invalid version and breaks the header checksum, which should be sufficient to make the UpdateFRUSDR tool fail to discover the power supply.

             

            Thus I think that I, to correctly fix the issue, need to rewrite the FRU data on PS_BAD. How can I do this when ipmitool fru write reports it as write protected?

            I see that ipmitool has an i2c command, would it be possible to do it with raw i2c requests?

             

            I tried to read up on IPMI a bit, are the following correct?

            • The power supplies connected to the IPMI Messaging Bus.
            • The IPMI messaging bus is an SMBus.
            • The IPMI messaging bus is a not the same as the system (i801) SMBus, and is not (directly) connected to it, thus it's not possible to reach the power supplies without going thru some mediator device.
            • 3. Re: R2308GZ4GC/S2600GZ "Pwr Unit Redund" after BIOS update - How to fix Power Supply FRU?
              Intel Corporation
              This message was posted on behalf of Intel Corporation

              As you can see the ipmi FRU write is write protected and at this moment we are not providing support on how-to by pass it.

              I do understand about rewriting the FRU information of the power supply could clear all this issues. I can share your feedback to our BIOS developing team to take this into consideration to address this in a future BIOS release.

              Let us know if you need further assistance.