Understanding LED 581 in AIX Version 4.x and 5.x

Answer

This document describes what happens during the time that LED 581 is shown during boot up and what may cause an indefinite hang on LED 581. This document applies to AIX Versions 4 and 5.

What happens during LED 581
Problem determination for LED 581 hang

What happens during LED 581

LED 581 is shown during the time that the Configuration Manager (/usr/sbin/cfgmgr -s) configures TCP/IP and runs /etc/rc.net to do specific adapter/interface/host name configuration.

Much of what happens during this time depends on factors such as access to a DNS, network traffic, valid network addresses, network access, network hardware/microcode, and the number and type of adapter/interfaces being configured. TCP/IP, over some interfaces (token-ring, for example), will wait for replies. If there are no replies, it will eventually timeout on the attempt and mark the interface as down. Nevertheless, this will take some time, possibly between three to ten minutes, or may hang indefinitely.

Problem determination for LED 581 hang

Most of the time LED 581 is caused by the DNS client not being able to resolve loopback correctly. This is often caused by a misconfigured or unreachable DNS server. This can be easily solved by resolving loopback locally using the /etc/netsvc.conf file, which dictates the order of name resolution.

1. In maintenance mode, enter the following:

         mv /etc/netsvc.conf /etc/netsvc.conf.saveme
   

   NOTE: This file might not exist.

2. Enter:

         cat /etc/hosts
   

   Search for the loopback line and make sure its correct, it should look like the following:

         127.0.0.1      loopback       localhost
   

3. Using your favorite editor, create a new file as follows:

         /etc/netsvc.conf
   

   Add the following according to your AIX level and save the file.

   • At AIX level 4.2.1 and below add:

           hosts=local,bind
     

   • At AIX level 4.3.x and above add:

           hosts=local,bind4
     

4. Enter:

         mv /etc/resolv.conf /etc/resolv.conf.saveme
   

   NOTE: This file should exist only if you are using DNS for name resolution.

5. Enter the following:

         mv /etc/rc.net /etc/rc.net.saveme
   

         touch /etc/rc.net
   

         chmod 755 /etc/rc.net
   

6. Using your favorite editor, edit the following file:

         /etc/inittab
   

   Comment out the following 3 lines in /etc/inittab:

         rctcpip:2:wait:/etc/rc.tcpip > /dev/console 2>&1 # Start TCP/IP daemons
    
         rcnfs:2:wait:/etc/rc.nfs > /dev/console 2>&1 # Start NFS Daemons
    
         dt:2:wait:/etc/rc.dt
   

   Also comment out all other lines which depend on TCP/IP being started.

   NOTE: Commenting out in /etc/inittab is done with a colon (:) in front of the line.

   Boot back up in multiuser mode. After the system is back up in multiuser, enter the following commands:

         mv /etc/rc.net.saveme /etc/rc.net
   

         cfgmgr
   

         /etc/rc.tcpip
   

7. Check to see if the TCP/IP interface is up by trying to ping your default gateway and DNS server. Enter:

         netstat -rn
   

   You should see output similar to the following:

         Destination      Gateway           Flags   Refs     Use  If   PMTU  Exp 
   Groups
         default          9.53.38.1         UG        3    79448  tr0     -   -  
   

   From this we see that this machine's default gateway is 9.53.38.1. Make sure you can ping your own default gateway, if it exists.

8. Test ping and telnet. Enter the following:

         ping <gateway_ip>
   

         mv /etc/resolv.conf.saveme /etc/resolv.conf
   

         cat /etc/resolv.conf
   

   You should see information similar to the following:

         domain austin.ibm.com
         nameserver 9.53.183.2
   

   Try to ping your nameserver, if you use one.

        ping <nameserver_ip>
   

   If all ping tests pass, then see if you can telnet to loopback.

   Enter the following:

         tn 127.0.0.1
   

9. If the telnet passes, put the files back. Enter the following:

         mv /etc/resolv.conf.saveme /etc/resolv.conf
   

         /etc/rc.nfs
   

   If everything starts properly when run manually, uncomment the lines (rcnfs, rctcpip, rc.dt) in /etc/inittab that start TCP/IP, NFS, and the Common Desktop Environment (CDE). You should now have the problem resolved and be able to boot properly in manual mode.