Boot problems due to wrong run-level with patch kit 7 on Tru64 V4.0f

Boot problems due to wrong run-level with patch kit 7 on Tru64 V4.0f

Post by Uwe Krutschinn » Wed, 21 Aug 2002 17:35:09



Hello from Berlin,

I upgraded our Tru64 V4.0f cluster to patch kit 7 last week. No problem
on the frist member, the machine boots without problems. On the second
member I observed a problem during boot to run-levels 2 and 3:

The scripts for the switch to rc2 and rc3 are entered but the subscripts
S* to start the different services do not launch everything required
because the run-level reported by 'who -r' (field 9) is set to '3'
instead of 'S' as expected:

There is a difference between both nodes even if booted in the same way:

Member 1 (ok)

     'who -r' on top of /sbin/rc2:
        .    run-level 3 Aug 20 07:46     3  0  S
     'who -r' on top of /sbin/rc3:
        .    run-level 3 Aug 20 07:46     3  0  S

Member 2 (nok)

     'who -r' on top of /sbin/rc2:
        .    run-level 0 Aug 20 07:41     0  0  3
     'who -r' on top of /sbin/rc3:
        .    run-level 0 Aug 20 07:41     0  0  3

Up to now I have no idea, what went wrong. Everything in the system
environment (/sbin/init, /etc/inittab, boot environment) is identical on

both systems. Even manual run-level changes using 'init <x>' on member 2
report crazy states afterwards if evaluated by 'who -r'.

Any hint is welcome.

Thanks in advance and regards,
Uwe Krutschinna

 
 
 

Boot problems due to wrong run-level with patch kit 7 on Tru64 V4.0f

Post by Uwe Krutschinn » Wed, 21 Aug 2002 21:20:03


Thanks to hints I got the problem could be fixed:

There was an invalid /var/adm/utmp located in the root partition and used
during boot process. After mount of /var the correct file replaced the wrong
one. This led to the inconsistency seen during boot.

Way to recover:

   * boot to single user mode
   * remove file in root partition
   * reboot to multi user mode

Thanks for support.
Uwe


> Hello from Berlin,

> I upgraded our Tru64 V4.0f cluster to patch kit 7 last week. No problem
> on the frist member, the machine boots without problems. On the second
> member I observed a problem during boot to run-levels 2 and 3:

> The scripts for the switch to rc2 and rc3 are entered but the subscripts
> S* to start the different services do not launch everything required
> because the run-level reported by 'who -r' (field 9) is set to '3'
> instead of 'S' as expected:

> There is a difference between both nodes even if booted in the same way:

> Member 1 (ok)

>      'who -r' on top of /sbin/rc2:
>         .    run-level 3 Aug 20 07:46     3  0  S
>      'who -r' on top of /sbin/rc3:
>         .    run-level 3 Aug 20 07:46     3  0  S

> Member 2 (nok)

>      'who -r' on top of /sbin/rc2:
>         .    run-level 0 Aug 20 07:41     0  0  3
>      'who -r' on top of /sbin/rc3:
>         .    run-level 0 Aug 20 07:41     0  0  3

> Up to now I have no idea, what went wrong. Everything in the system
> environment (/sbin/init, /etc/inittab, boot environment) is identical on

> both systems. Even manual run-level changes using 'init <x>' on member 2
> report crazy states afterwards if evaluated by 'who -r'.

> Any hint is welcome.

> Thanks in advance and regards,
> Uwe Krutschinna

--
+-------V-------+ Uwe Krutschinna, Alcatel SEL AG Berlin, TBE/ES-D4
| A L C A T E L | Colditzstrasse 34-36, D-12099 Berlin, Germany
+---------------+ Phone: +49-30-7002-3260, Fax: +49-30-7002-3851