NIM/SPOT Debug for RS/6000 SP Systems


Contents

About this document
Procedure

About this document

The following procedure outlines how to create a debug version of the Shared Product Object Tree (SPOT). Some commands in this procedure are specific to the IBM RS/6000 Scalable POWERparallel System. Following these steps may potentially display an error that would not otherwise be shown during a normal Network Install Management (NIM) boot/install. This display may aid in problem source identification.

This document applies to AIX Versions 4.x.


Procedure

On the control workstation (CWS), do the following:

  1. Set the nodes to disk to unallocate all NIM resources by entering the following command:

    spbootins -r disk
                             OR
    spbootins -r disk -l

    An example for frame 1 with 15 nodes looks like the following:

    spbootins -r disk 1 1 15
                                  OR
    spbootins -r disk -l 1

  2. Now, put the SPOT in debug mode.

    1. Find the SPOT name. Enter:
            lsnim -t spot 
               spot_aix###  <resource>  <spot>
      
    2. Put the SPOT in debug mode by entering the following command:
            nim -Fo check -a debug=yes spot_aix### 
      
    WARNING: Make sure you are not NIM installing when using the F flag as documented in this step. This will cause the boot image to be rebuilt even if it is allocated.

  3. When the preceding command completes, enter the following command:
         lsnim -l spot_aix###
    

    For high/wide/thin mca-based nodes, look for the following:

         enter_dbg     = "rs6k.up.ent 0xXXXXXXXX" 
         enter_dbg     = "rs6k.mp.ent 0xXXXXXXXX"
    

    For wide/thin pci-based nodes, look for the following:

         enter_dbg       = "chrp.mp.ent 0xXXXXXXXX"
    

    XXXXXXXX will be an address embedded in the text, such as 0x0013afa0. Drop the 0x and leading zeroes, and make note of the remainder. For example, the remainder in this example would be 13afa0.

  4. Now, set the node(s) to install by entering the following command:

    spbootins -r install <Frame#> <Node#> <NumberOfNodes>
                                  OR
    spbootins -r install -l <Node#>

    This will reallocate the SPOT for use as it has been placed in debug mode.

    1. If you are debugging the installation, the bosinst_data for this node must be modified. Complete the following steps:

      • Enter:
                lsnim -l <nodename>
        

        Look for the bosinst_data resource, bosinst_data = <bosinst>

      • Enter:
                lsnim -l <bosinst>
        

        Look for the location of this resource, location = <location>

      • Open the file with your favorite editor. Make the following changes:
                 vi <location> 
                 Modify the control_flow: 
                 PROMPT = yes 
                 BOSINST_DEBUG = yes
        
      • Save this file.

  5. Start manual node conditioning steps for this node. This will result in opening a write tty to the node.

    NOTE: After opening the write tty to start manual node conditioning, open a new window in the CWS and run the following command to start a logging session for this NIM debug:

         s1term <frame#> <slot#> | tee /tmp/nim.debug.log
    

    After manual node conditioning, a ">" prompt will display. Enter the following command:

    st <remainder from step 3> 2
    g

    For our example, this would be:

         st 13afa0 2
         g
    

    It may be helpful to have the 3DigitDisplay and/or the Node Front Panel displayed so you can watch the LEDs change as the netboot continues. It may also be helpful to have the IBM 9076 Scalable POWERparallel Systems SP2 Diagnosis and Messages Guide (GC23-3899-02), "SP-Specific LED Values" available to determine approximately where in the boot process you are.

  6. Debugging installation - Optional

    If you followed step 4a for debugging the installation, then the next prompt will ask you to define a system console. Press 1 followed by the Enter key. After this, you will be prompted to choose the language. Press 1 followed by the Enter key. Now, the install menu will appear with the three menu options.

    Enter 911 at this screen. This will enable the install debugging for this session. The screen will refresh, and continue with option 1.

    NOTE: At any point during the install, if the process hangs for a long period of time, press ctrl-q to attempt to move the installation to the next phase. This step should be repeated as necessary.

  7. After finishing the NIM debug, set the SPOT back to its regular state. Enter the command:
              nim -Fo check spot_aix###
    

    Again, note the F flag in the preceding command syntax.

    This is a good means of obtaining useful error messages that you would not otherwise see with NIM. Typical problems identified by this process are name resolution issues and network problems such as routing, for example.




[ Doc Ref: 90605219814828     Publish Date: Apr. 03, 2001]