On Wed, 13 Jun 2007, Pann McCuaig wrote:

> Greetings, all:
>
> I've already contacted SAS technical support, but thought that someone
> on this list might have and idea.
>
> Last weekend someone tripped a breaker (or something) and the dedicated
> 220VAC line to our cluster rack went bye-bye. Monday morning I rebooted
> the cluster nodes one at a time and they all came up OK, although one
> required manual fsck intervention.
>
> After the reboot all seemed to be well until someone tried to run SAS.
>
> Here is my brief description of the problem for SAS technical support:
> ---------------------------------------------------------------------
> Scientific Linux SL release 4.4 (Beryllium)
> 2.6.9-42.0.10.ELsmp
> Sun V40z, 4-Opteron CPUs, 32G RAM
>
> SAS hangs and generates a number of non-kill-able tasks when executed
> either with the GUI (which never shows up) or with -nodms from the
> command line.

You could attach a "strace" to the "hung" process to determine if it is 
really hung or doing something "else".

   strace -o /tmp/trace.out -p <processid>

-Connie Sieh
>
> Installed SAS on a similar platform. Runs fine.
>
> Did a second parallel installation of SAS on the platform that is
> failing. Same symptoms with the clean installation as with the one that
> has been running fine (until recently).
> ---------------------------------------------------------------------
>
> I'm flailing so any and all suggestions will be greatly appreciated.
>
> Cheers,
> Pann
>