LED 231 or E1F7 Hang During Netboot of SP Nodes


Contents

About this document
Procedure

About this document

LED 231 or E1F7 occur during a netboot of sp nodes for an installation, migration, maintenance or diagnostic. LED 231 will occur in MCA type nodes and LED E1F7 will occur in PCI type nodes.

It is recommended that you have some knowledge of SP before attempting to follow this document.

This document applies to AIX versions 4.2.x and 4.3.x, also to PSSP 2.x and 3.x.


Procedure

This procedure assumed that you did not receive any errors when executing setup_server on any of the nodes. If you received errors when running setup_server, please correct the errors or call Supportline before you can follow this procedure. setup_server will get executed when you run the setup_server command or the spbootins command below.

   spbootins -r {install | customize | disk | maintenance | diag | migrate } -l
   <node_number>

For this example, install node #4 called spsn4. Now, run the spbootins -r install -l 4 command. It should have completed without any errors, then proceed to netboot the node. It should have hung with LED E1F7.

  1. Check to make sure node #4 is setup correctly for the installation. Enter:
             splstdata -b
    

    The following is the output specifically for node #4:

             4 spsn4            10005AFA2F25    0      install          hdisk0 
             bos.obj.ssp.421 Sun_Jan__2_15:14:51     bos.obj.ssp.421        
    aix421     PSSP-2.4 
    

    NOTE: Make sure the preceding settings are correct.

  2. If setup_server ran cleanly, check the following:

  3. To check the MAC address of the integrated Ethernet card, refer to one of documents on manually booting an RS/6000 SP node, currently available at AIX Techdocs. Ensure it is the same as that listed in the lsnim command output and the /etc/bootptab file. If the MAC address is not the same, complete the following steps.

    1. If the MAC address on the node is not the same as the output of the splstdata -b command, then update the System Data Repository (SDR), set the node back to disk and enter the following command:
                sphrdwrad -l <node_number>
      

      WARNING: Any nodes specified will be powered off to acquire the Ethernet addresses.

    2. If the MAC address is the same on the node as in the splstdata -b command output but different than in the lsnim -l command output and the /etc/bootptab file entries, then enter:
                delnimclient -l <node_number>
      

    If the MAC address is the same, run a ping test. If the ping test fails, make sure the Ethernet on the Control Work Station (CWS) is fine. If no problems are detected and you cannot ping the en0 interface on the nodes, there may be a hardware issue. If the ping test is successful, continue with the procedure.

  4. Check in /etc/inetd.conf file to ensure the following entries exist and are not commented out.

    For AIX 4.2 system

    bootps  dgram   udp     wait    root    /usr/sbin/bootpd       bootpd
    /etc/bootptab 
    tftp     dgram  udp    wait     nobody  /usr/sbin/tftpd         tftpd -nisrv 
    

    For AIX 4.3 system

    bootps  dgram   udp     wait    root    /usr/sbin/bootpd       bootpd
    /etc/bootptab 
    tftp     dgram  udp6    SRC     nobody  /usr/sbin/tftpd         tftpd -n 
    

    If the entries appear fine, run the refresh -s inetd command. Return to the MANUAL NODE CONDITIONING SCREEN and continue with the procedure. Ensure that bootp packets are sent and received. If packets are not being received, go to step 5. If bootp does send and receive packets, check to see if tftpd is running. You should be able to see te bootp packets sent and received incrementing up. If it only sent and the receiving is not incrementing, then proceed with step 5.

  5. Go back to the CWS and complete the following steps.

    1. Enter the following command to list the bootp server port:
         grep bootps /etc/services 
      

      Sample output:

         bootps          67/udp                          # bootp server port
      
    2. Enter the following command:
         netstat -an |grep 67
      
      Make sure this line displays in the output:
         udp        0      0  *.67                   *.*
      
    3. Next, put bootpd into debug mode.
      • Open the /etc/inetd.conf file and comment out bootpd.
      • Refresh inetd. Enter:
           refresh -s inetd
        
      • Verify that a bootp process is not running. Enter:
           ps -ef |grep bootp
        
        If a bootp process displays, kill it, open a new terminal on the CWS, and run the /tmp/bootpd.log script.
      • Start bootpd in debug mode. Enter:
           bootpd -s -d -d -d -d -d
        

        By completing this step on the s1term where the manual node conditioning was performed, bootp should receive packets and start with the installation. If it still does not receive packets, review the debug output.




    [ Doc Ref: 95194856320550     Publish Date: Oct. 23, 2001]