[torqueusers] How do I ask an intelligent question?

Dan Lapine dlapine at ncsa.uiuc.edu
Thu Oct 14 09:16:20 MDT 2004


What are you using to schedule jobs? torque has a native fifo scheduler, but
you have to activate it. There are other schedulers available that torque 
can tie in with. Jobs won't run if the system scheduler isn't run and passing 
job start info to torque.

Try a manual qrun of the jobid to see if pbs will run. If the job runs, then the
issue you're seeing is probably a scheduling issue. If the job won't run, then
you should get an interesting error message to help with your troubleshooting.

On Thu, 14 Oct 2004, Michael Wheatley wrote:

> I hate asking this type of question as I have little information to go
> on.  As a result my question will now be "how do I troubleshoot so I
> can work out whats going on?"
> 
> I have set up a pbs server on a debian 3 machine with one slave node
> at the moment.  NFS works fine and /home is mapped without trouble. 
> pbsnodes -a will eventually (initially when I start daemons running it
> says state unknown, is this normal?) show the state of the only node
> as 'up'.
> 
> Now when I try and run a job echo 'sleep 10'| qsub -l n=1 the job gets
> on the qstat without trouble.  However the job never runs (leaves the
> queue).
> 
> Both machines can see each other and the /etc/hosts resolves the
> network addresses without drama.
> 
> Obviously I have made a newbie mistake, but which one.  How do I delve
> into the guts to solve my own problem?
> 
> Thanks all
> 
> Mike
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://supercluster.org/mailman/listinfo/torqueusers
> 

-- 
---
Daniel LaPine, System Engineer
National Center for Supercomputing Applications (NCSA)
email: dlapine at ncsa.uiuc.edu
phone: 217-244-9294




More information about the torqueusers mailing list