hi,
thanx for replying !
qsub -l nodes=2
is not required since v r not submitting a parallel job. the shell script i submit is as follows:
Ok... I must've misread something. I thought the problem was that you *were not* getting two nodes. Either way, when having problems, it's good to specify parameters explicitly. Use qsub -l nodes=1:ppn=1 if that's what you are requiring.
This installation appears to have been modified since OSCAR left it. OSCAR doesn't make the clienthost entries in the mom config file that you show below. Also, we use the Maui scheduler instead of pbs_sched for a number of reasons. One large reason is that it's supported. I recommend that you use Maui so that you have someone to go to about these problems. No one supports pbs_sched.
Your script looks fine. While you have one job running, and another job queued... try running "qstat -n" and "pbsnodes -a", and including that output here. That way, we can have a clearer picture of what the state of your cluster is at that moment.
#!/bin/sh
### Job name
#PBS -N test
### Output files
#PBS -e test.err
#PBS -o test.log
#PBS -q workq
#echo Working directory is $PBS_O_WORKDIR
#cd $PBS_O_WORKDIR
echo Node_File is $PBS_NODEFILE
echo Running on host `hostname`
#This will keep cpu busy for 20 secs
sleep 20
#echo Time is `date`
#echo Directory is `pwd`
#All done
this scrpit when submitted twice ...the second submission is queued ..although second node remains free
( i chked that using ps -ef on second node and
also pbsnodes -a shows it free and node1 as "job-sharing" )
as far as maui is concerned , we have stopped it and instead started pbs scheduler
/opt/pbs/sbin/pbs_sched -c config
(also started the scheduler without -c option , but in vain)
config file :
$clienthost node1
$clienthost node2
- Maulin
>From: Jeremy Enos <[EMAIL PROTECTED]> >To: "maulin pandya" <[EMAIL PROTECTED]>, [EMAIL PROTECTED] >Subject: Re: [Oscar-users] pbs scheduling problems in oscar 2.2.1 >Date: Wed, 07 Apr 2004 15:02:08 -0500 > >Have you tested with Maui? >I think you've already sent your qsub line... can you send your >"test" script as well? > >Have you tried specifying the number of nodes on the qsub line? >e.g. >qsub -lnodes=2:ppn=1 > > Jeremy > > >At 12:43 PM 4/7/2004, maulin pandya wrote: > >>hi >> >>i have setup a 3 node cluster(1 head node+2 client nodes) using >>oscar version 2.2.1 on a system using RH 7.3 >>i m facing problems in batch system configuration i.e. i cannot run >>multiple jobs at the same time (only 1 is run, rest r queued) >>the exact problem is as follows: >> >>-i m using only the pbs scheduler. so at first i shut down the ! maui >>scheduler(which comes with oscar). >> >>-without submitting any job, if i run the command 'pbsnodes -a' the >>state of both the nodes are "free" >> >>-when i submit a small shellscript (which just sends the cpu into >>sleep state for 20 seconds) multiple times using the qsub command, >>the first job starts running immediately and the rest are queued. >> >>-now the 'pbsnodes -a' command shows state of node1=job-sharing and >>state of node2=free. however, none of the nodes are configured as >>time-shared or temporarily-shared nodes. also, after the 1st job is >>over, the 2nd job starts running but that too on the 1st node only. >>somehow or the other, the jobs do not get executed on node2. >> >>-i checked the file sched_logs and found the error to be: >>"cannot find enough right type of nodes to run jobs" >> >>-however, i (as a root) can run jobs on node2 f! orcibly by using >>"qrun -H node2 432.cluster" >> >> PS:i m using RH 7.3 and i am not planning to upgrade (neither oscar >>nor Red Hat Linux) >> >>--thanks >>-maulin >> >> >> >> >>---------- >>Contact brides & grooms FREE! >><http://g.msn.com/8HMBENIN/2737??PS=>Only on www.shaadi.com. >>Register now! >>------------------------------------------------------- This SF.Net >>email is sponsored by: IBM Linux Tutorials Free Linux tutorial >>presented by Daniel Robbins, President and CEO of GenToo >>technologies. Learn everything from fundamentals to system >>administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click >>_______________________________________________ Oscar-users mailing >>list [EMAIL PROTECTED] >>https://lists.sourceforge.net/lists/listinfo/oscar-users
Post Classifieds on MSN classifieds. Buy and Sell on MSN Classifieds. ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click
_______________________________________________ Oscar-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/oscar-users
