[torqueusers] Unexpected qsub error

Garrick Staples garrick at clusterresources.com
Mon Nov 20 13:57:26 MST 2006


On Mon, Nov 20, 2006 at 10:43:48AM +0100, David McGiven alleged:
> 
> Dear Garrick,
> 
> Unfortunately, things haven't cleared themselves. I deleted the whole
> /var/spool/PBS (with rm -rf) on the server and reinstalled
> torque2.0.0.p11
> 
> Then in one of the nodes (with 10 cpu's) I deleted all the temp files
> under /var/spool/PBS/mom_priv/jobs and basically anyother thing that
> might cause problems, like aux, spool, and undelivered directories.

That was probably a bit too drastic :)

 
> Still, the same problem.
> 
> "qsub: Job exceeds queue resource limits"
> 
> Now it allows me to send with 2 processors but not with 4. But this
> behavious changes, so there's times in which I can only send with 1
> processor.
> 
> I've tried everything, but nothing seems to work, and I don't know if
> this problem is still related with the qdel -p.
> 
> Could you please give me some more advice ?

As long as you are basicly starting from scratch at this point, move to
2.1.6.  2.0 has subnode counting bugs that could be causing your
problem.



More information about the torqueusers mailing list