Powering off an expansion unit

This procedure provides detailed instructions for powering off an expansion unit.

1. If you have not already done so, record the expansion unit's location code and machine type-model-serial numbers (MTMS).
2. Can the expansion unit be powered off with the server powered on (the server is HMC-managed), and does the customer want to perform the procedure with the server powered on?
   • No: The expansion unit must be powered off by powering off the server. See Powering on and powering off a system for instructions on powering off the server, then return to the procedure that sent you here.
   • Yes: Go to step 3.
   Note: If the activated firmware level of the server to which the expansion unit is attached is lower than SF235, there is a small chance that bus numbers for the PCI buses in the expansion unit will change if the I/O backplane in the expansion unit is replaced with the server powered off. If this occurs on a server with multiple logical partitions, the I/O slot resources in the expansion unit will have to be reassigned to the appropriate partitions. Additionally, AIX® logical resource names of the I/O resources in the expansion unit may change requiring some manual reconfiguration in the partition. If the I/O backplane is replaced with the server powered off, be prepared to perform these reconfiguration activities after the procedure. If the I/O backplane is replaced with the server powered on, either to firmware standby or firmware running, reconfiguration will not be required.

   If it is desirable to replace the I/O backplane with the server powered off, but configuration changes cannot be tolerated, powering off the server and then powering it on to firmware standby state before performing the procedure might be a satisfactory alternative.

   The activated firmware level of the server is displayed in the upper right corner of the ASMI welcome page.

3. Verify that the expansion unit being powered off is the correct one. If the expansion unit has an alphanumeric display on the control panel, you can verify that the type, model, and serial number of the expansion unit displayed on the user interface match that shown on the control panel display. Otherwise, you can activate the identify indicator on the expansion unit. To activate the Identify indicator on the expansion unit, do the following steps for your particular interface:
   • From the HMC V7 and later, use Select managed system > Operations > LED Status > Identify LED. Select the desired expansion unit and use Activate LED to turn on the LED and thus verify that the tower is the correct one. Then, use Deactivate LED to turn off the LED.
   • From the HMC V6R1.3 or earlier, use Service Focal Point > Service Utilities > Identify LED processing.
   • From the IBM Systems Director Management Console (SDMC), complete the following steps:

     1. On the Welcome page, under the Resources tab , select the server.
     2. Click Actions > Service and Support > Hardware > Identify LED.
     3. Select the desired expansion unit and click Activate LED to turn on the LED and thus verify that the expansion unit is the correct one. Then, click Deactivate LED to turn off the LED.
   • From IBM i use System (or Dedicated) Service Tools > Start a service tool > Hardware service manager > Packaging hardware resources > Concurrent maintenance > Toggle LED blink off/on.
   • Are you powering off the correct expansion unit?
     • No: Choose the correct expansion unit. Then perform this procedure again.
     • Yes: Go to step 4.
4. The system will prevent the expansion unit from being powered off if there are failed or inactive links in the HSL loop that the expansion unit is on, other than the links adjacent to the expansion unit. If an attempt is made to power off the expansion unit under these circumstances, a descriptive error message will be presented. Determine if there are failed or inactive links by following one of the these procedures, if applicable: If your system is managed by an HMC, see Verify the RIO/HSL loop with an HMC. If you have IBM i, see Verify the RIO/HSL loop without an HMC.

   Are there external links that are not operational in the HSL loop, other than those adjacent to the expansion unit being powered off?

   • No: Go to step 5.
   • Yes: Repair the failed or inactive links, except those adjacent to the expansion unit being powered off. If you need assistance, contact your service provider. Then perform this procedure again.
5. Verify that the SPCN loop is complete. A 10009135 reference code in the error log indicates an incomplete SPCN loop. A complete SPCN loop is not required, but is highly recommended. Powering off an expansion unit with an incomplete SPCN loop could cause lost contact with other expansion units on the SPCN loop. The expansion units that the system lost contact with will continue to function; however, the system may log additional errors and will not be able to monitor those units. Work with the customer to determine whether the SPCN loop is complete, whether correction is necessary, and then continue with the next step.
6. If the system has only partitions running AIX or Linux, skip to step 10, if not go to step 7.
7. The system will prevent the expansion unit from being powered off if the expansion unit is configured as switchable and the system that owns the HSL I/O bridge (the logical owner of the expansion unit) is not the system that provides power to the expansion unit (the system whose SPCN the expansion unit resides in, or SPCN system). If an attempt is made to power off the expansion unit under these circumstances, a descriptive error message will be presented. Determine if the expansion unit is configured as a switchable unit, either by asking the customer or referring to Determining if an expansion unit is configured as switchable under OptiConnect .
   Note: Switchable expansion units will exist only if the server makes use of Switchable Independent Disk Pools.
   Is the expansion unit configured as switchable?

   • No: The expansion unit is configured as private to this system. Go to step 10.
   • Yes: Go to step 8.
8. Determine if the logical owner of the expansion unit is the SPCN system by asking the customer or referring to Determining if a expansion unit is configured as switchable under OptiConnect .
   Is the logical owner of the expansion unit also the SPCN system?

   • No: Work with the customer to switch logical ownership of the HSL I/O bridge in the expansion unit to the SPCN system. This should be done by using the OptiConnect user interface, a feature of IBM i. If that is not successful due to problems with the OptiConnect user interface, it may be attempted as described in Switching ownership of an expansion unit's switchable resources. . However, this should be used only as a last resort since it may result in Cluster Resource Group (CRG) problems. If the switch cannot be performed, contact your service provider for assistance. Otherwise, continue with step 9 from the SPCN system.
   • Yes: Go to step 9.
9. The system will prevent the expansion unit from being powered off if it contains resources included in an active Cluster Resource Group (CRG). If an attempt is made to power off the expansion unit under these circumstances, a descriptive error message will be presented. Determine if the expansion unit contains resources included in an active CRG by asking the customer.
   Does the expansion unit contain resources included in an active CRG?

   • No: Go to step 10.
   • Yes: Work with the customer to end each active CRG that contains resources in the expansion unit.
10. Determine if any of the I/O resources in the expansion unit are owned by partitions running Linux.
    Are any of the I/O resources in the expansion unit owned by partitions running Linux?

    • No: Go to step 11.
    • Yes: Power off the partitions running Linux and go to step 11.
11. Determine if any of the I/O resources in the expansion unit are owned by partitions running AIX. Are any of the I/O resources in the expansion unit owned by partitions running AIX?
    • No: Go to step 12.
    • Yes: Do the partitions have a code level that supports the powering off and on of an expansion unit with the partition running? The required level for AIX 5.2 is 5.2.0.61 or later and the required level for AIX 5.3 is 5.3.0.22 or later. You can determine the level of AIX by entering the command lslpp -l devices.chrp.base.rte from the AIX command line.
      • Yes: Does the HMC have a code level that supports power off and on of an expansion unit with I/O resources assigned to active partitions running AIX? The required HMC code level is version 4, release 5.0 or later. You can view the HMC code level by accessing: Licensed Internal Code Maintenance > HMC Code Update. The code level appears in the content area.
        • Yes: Go to step 12.
        • No: Power off the partitions that are running AIX and that own I/O resources in the expansion unit, then go to step 12.
      • No: Power off the partitions that are running AIX and that own I/O resources in the expansion unit, then go to step 12.
12. Work with the customer to end I/O activity to the expansion unit by performing the following steps. If all partitions that own I/O resources in the expansion unit have been powered off, skip this step and go to step 13.
    • In IBM i partitions:
      • Work with the customer to vary off all configuration descriptions (for example, ctl, device, line, network server, etc.) that use resources in the expansion unit. This can be done by using the vrycfg command. The system will prevent the expansion unit from being powered off unless all configuration descriptions that use resources in the expansion unit are varied off. If an attempt is made to power off the expansion unit when there are configuration descriptions that use resources in the expansion unit that are varied on, a descriptive error message will be presented.
        Note: If configuration descriptions that use resources in the expansion unit cannot be varied off, the procedure must be delayed until they can be varied off or the partitions that have such configuration descriptions must be powered off.
      • If the expansion unit contains disk unit IOPs, IOAs, or both, and the disk units are not mirrored to disk units in other expansion units, work with the customer to end all communications activities to these disk units by ending all subsystems.
        Note: The system allows you to power off the expansion unit even if this is not done, however, doing so may result in a disk unit attention SRC and a partial or complete hang of the partition until the expansion unit is powered on.
      • If the expansion unit contains disk unit IOPs, IOAs, or both, and these disk units are mirrored to disk units in other expansion units, you do not need to end all subsystems. Mirroring will assist during the service action.
    • In AIX partitions:
      • Work with the customer to unconfigure all logical device resources that use resources in the expansion unit. This can be done by using the rmdev command. The system will prevent the expansion unit from being powered off unless all logical device resources that use resources in the expansion unit are unconfigured. If an attempt is made to power off the expansion unit when there are configured logical device resources that use resources in the expansion unit, a descriptive error message is presented.
      • If logical device resources that use resources in the expansion unit cannot be unconfigured, the procedure must be delayed until they can be unconfigured or the partitions that have such resources must be powered off.
      • Following is an example of the type of message that is displayed if there are logical AIX resources that must be unconfigured. It shows two messages. The first indicates that logical device resource pci11 with physical name U5294.001.23B2424-CB1-C01 must be unconfigured. To unconfigure this resource enter the AIX command rmdev -R -l pci11. The second indicates that logical device resources pci6 and ent0 with physical name U0595.001.435DF30-CB1-C02 must be unconfigured. To unconfigure these resources enter the AIX command rmdev -R -l pci6.

        Reporting partition type: RPA
        
        
        Return code type: Failure (hard stop or user intervention required).
        
        Message:
        com.ibm.hsc.client.rmc.HSCRMCException: Query operation for slot 
        resource has failed on partition 2. 
        
        The partition command is:
        drmgr -Q -c slot -s U5294.001.23B2424-CB1-C01 -w 0 -d 3
        
        The partition standard output is:
        drslot_chrp_slot: slot_type=1 all slots=0x20042d68
        
        The partition standard error is:
        0931-029 The specified slot contains a device or devices that are currently 
        configured. Unconfigure the following device or devices and try again.
        
        pci11
        
        
        The RMC return code is 0. The AIX return code is 1.

        Reporting partition type: RPA
        
        Return code type: Failure (hard stop or user intervention required).
        
        Message:
        com.ibm.hsc.client.rmc.HSCRMCException: Query operation for slot resource 
        has failed on partition 2. 
        
        The partition command is:
        drmgr -Q -c slot -s U0595.001.435DF30-CB1-C02 -w 0 -d 3
        
        The partition standard output is:
        drslot_chrp_slot: slot_type=1 all slots=0x20042ca8
        
        The partition standard error is:
        0931-029 The specified slot contains a device or devices that are currently 
        configured. Unconfigure the following device or devices and try again.
        
        ent0
        pci6
        The RMC return code is 0. The AIX return code is 1.

      • The following AIX commands are helpful in determining which logical device resources use an expansion unit. In the following examples, U0595.001.435DF30 is the location code of the expansion unit being powered off.

        Displays information about logical device resource pci5:
        lscfg -l pci5 
        
        Displays all logical resources in U0595.001.435DF30:
        lscfg | grep U0595.001.435DF30 
        * pci5      U0595.001.435DF30-CB1                PCI Bus
        + sisioa0   U0595.001.435DF30-CB1-C01            PCI-X Dual Channel U320 SCSI RAID Adapter
        + scsi0     U0595.001.435DF30-CB1-C01-T1         PCI-X U320 SCSI RAID Adapter Physical bus
        + hdisk0    U0595.001.435DF30-CB1-C01-T1-L6-L0   16 Bit LVD SCSI Disk Drive (36400 MB)
        * ses0      U0595.001.435DF30-CB1-C01-T1-L15-L0 SCSI Enclosure Services Device
        + scsi1     U0595.001.435DF30-CB1-C01-T2         PCI-X U320 SCSI RAID Adapter Physical bus
        + sisraid0  U0595.001.435DF30-CB1-C01-C1-T1      PCI-X U320 SCSI RAID Adapter Logical bus
        
        Displays the rootvg hdisk(s):
        lsvg - o  

      • The resources representing the partition's LAN connection to the HMC, such as ent0 and en0, must not be unconfigured or the procedure will not work. If these resources use the expansion unit, the partition must be powered off to continue with the procedure.
    Attention: If resources in the expansion unit are assigned to I/O server partitions and are being used by I/O server partitions (also known as virtual I/O servers), work with the customer to determine whether powering off the expansion unit will negatively impact the client partitions. If so, the client partitions should be powered off before continuing.
13. Power off all external DVD-RAM storage devices (containing their own power supplies) connected to resources in the expansion unit, and then go to step 14.
14. Use one of the following procedures to power off the expansion unit. Carefully follow the instructions provided by the utility to complete the power off.
    Attention: If the server is not managed by a management console and there is no IBM i service partition, the expansion unit power cannot be turned off with the server power turned on.

    To turn off the power in an expansion unit with the server power on, do one of the following procedures:

    • If the server is HMC-managed, the power off and power on functions must be performed by using the Power On/Off Unit utility on the HMC.
      • For HMC V6R1.3 and earlier, do the following steps:
        1. In the navigation area, open the Service Applications folder.
        2. Select Service Focal Point.
        3. In the contents area, select Service Utilities.
        4. In the Service Utilities window, select the system on which you are working.
        5. Select Selected > Power On/Off Unit
      • For HMC V7 and later, do the following steps:
        1. Select the managed system, then select Serviceability -> Hardware -> Power on/off IO Unit.
        2. In the Power on/off IO Unit window, expand the Managed system I/O and select the appropriate unit.
        3. Click the Power off button.
      • For the IBM Systems Director Management Console (SDMC), complete the following steps:
        1. On the Welcome page, under the Resources tab , select the server.
        2. Click Actions > Service and Support > Hardware > Power on/off IO Unit.
    • If the server is not managed by a management console, the power off and power on functions must be performed by using the Packaging Hardware Resources utility from the IBM i service partition, do the following steps:
      1. Access the system service tools or the dedicated service tools on the IBM i service partition.
      2. Select Start a service tool.
      3. Select Hardware service manager.
      4. Select Packaging hardware resources.
      5. Select Concurrent maintenance.
    Notes:

    1. If the procedure is being done from an IBM i partition, the system prevents the procedure from being done from a console or terminal session connected through the expansion unit being powered off. A different console or terminal session not connected through the expansion unit being powered off must be found, or the server must be powered off. See Powering on and powering off a system for instruction on powering off the server, then return to the procedure that sent you here.
    2. If I/O activity to resources in the expansion unit was not ended as described in step 12, the power off will not continue and messages will be presented indicating which partition resources need to be varied off or unconfigured. Similarly for other conditions described previously in this page that prevent the power off.
    3. Carefully follow all instructions given by the utility. For instance, you may be informed that critical disk resources exist in the expansion unit and that panel function 69 must be used to power the expansion unit back on.
    4. Wait for the expansion unit to power off. This might take up to 15 minutes depending on the configuration.
    5. When the utility gives notification that the power off has completed, ensure that the expansion unit is actually powered off by verifying that one of the two air moving devices (AMDs) in the expansion unit has stopped and that the power LED/indicator on the expansion unit control panel is off.
    6. If the expansion unit does not power off after more than 15 minutes, the system may be hung. If this occurs, contact your service provider for assistance.
15. Were you directed here to power off the expansion unit to set the expansion unit configuration ID or the machine type-model-serial numbers (MTMS)?
    • No: Go to step 16.
    • Yes: Return to the procedure that sent you here.
16. Does the procedure that sent you here instruct you to remove the ac power from the expansion unit?
    • No: Return to the procedure that sent you here.
    • Yes: Go to step 17.
17. Is this an expansion unit with ac power cords connected from this unit's ac input/charger to an SPCN-controlled expansion unit (for example, a 5088 expansion I/O unit)?
    • No: This expansion unit can be serviced without affecting an SPCN controlled expansion unit. Return to the procedure that sent you here.
    • Yes: Servicing this expansion unit may affect an SPCN-controlled expansion unit that is receiving ac power from this expansion unit. Go to step 18.
18. You may be able to maintain ac power to the expansion unit that is receiving ac power from this expansion unit and remove power from this expansion unit so that you can service it. The following steps will help you to determine what you can do. Were you directed to this procedure as part of a procedure to exchange the ac input/charger, any of the device boards, or the power board?
    • No: You can maintain ac power to the expansion unit receiving ac power from this expansion unit while you are servicing this expansion unit. When the expansion unit is powered off, open the rear cover and trace the ac jumper cords from the three power supplies to the ac input charger. Then disconnect them at the ac charger. Power is now removed from the components that you will be working with or near. Return to the procedure that sent you here.
    • Yes: You must search for an alternate ac power source for the expansion unit receiving ac power from this expansion unit. Go to step 19.
19. Can you find an alternate ac power source for the expansion unit that is receiving ac power from this expansion unit?
    • No: You must power off the system to service this expansion unit. See Powering on and powering off a system for instructions on powering off the system, then return to the procedure that sent you here.
    • Yes: Move each of the expansion unit's ac power cords one at a time to the alternate ac power source to prevent the expansion unit from losing ac power. Return to the procedure that sent you here.