[GE globus] Grid Engine and GT4

Diego Bello dbello at gmail.com
Fri Jun 30 04:35:08 BST 2006


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

On 6/29/06, Melvin Koh <melvin at apstc.sun.com.sg> wrote:
> Hi,
>
> This may not be it, but I just want to make sure. Have you set both
> reporting=true and joblog=true using GE's "qconf -mconf" command?
>
>
> On Thu, 29 Jun 2006, Otheus (aka Timothy J. Shelling) wrote:
>
> > Greetings from Innsbruck, Austria,
> >
> > I've also been assigned to the task of getting SGE to work with Globus GT4,
> > and like Diego Bello, I have run into troubles in GT 4.0.2.  My problems
> > appear to be almost identical, yet not quite.
> >
> > First, the symptoms. I'm trying to submit a basic job on the host which is
> > the SGE submission host and the host running the GT4 container.
> >
> > $ globusrun-ws -submit -factory $HOSTNAME -Ft SGE -c -- /usr/bin/printenv
> > Submitting job...Done.
> > Job ID: uuid:718d3368-07a9-11db-b935-000423c068ae
> > Termination time: 06/30/2006 19:57 GMT
> > <after waiting for 5 minutes>
> > Current job state: Unsubmitted
> > <hit ctrl-c>
> > Canceling...Canceled.
> > Destroying job...Done.
> > globusrun-ws: Operation was canceled
> > $
> >
> > >From what I can tell, the job is being created AND submitted (and run, etc).
> > HOWEVER, the reporting appears to not be properly parsed or something.
> > Here's the tail of the reporting file:
> >
> > 1151613380:queue_consumable:all.q:k10:1151613380::slots=1.000000=8.000000
> > 1151613383:acct:
> > all.q:k10:dps:otheus:sge_job_script.8006:672:sge:0:1151613373:11
> > 51613382:1151613382:0:0:0:0:0:0.000000:0:0:0:0:9402:0:0:0.000000
> > :0:0:0:0:524:89:
> > NONE:defaultdepartment:NONE:1:0:0.000000:0.000000:0.000000:-U dps_users:
> > 0.000000
> > :NONE:0.000000
> > 1151613383:queue_consumable:all.q:k10:1151613383::slots=0.000000=8.000000
> >
> >
> > Now, I only got this far once I made sure that the sge.pm file was setting
> > the SGE_* environment variables properly.  I also followed another piece of
> > advice in altering the JobManager.pm module to output the log to /tmp.
> > Here's what I found in the other logs:
> >
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   script location:
> > /home/otheus/.globus/ef2c0c96-07aa-11db-9dc2-000423c068ae/sge_job_script.750
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   email(s) will not be sent
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT: Checking project details
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   Project not specified
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT: Determining job type
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   Job is of type multiple
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   forking multiple requests
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT: SGE job script successfully built! :-)
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT: Submitting a job
> > Thu Jun 29 22:07:55 2006 JM_SCRIPT:   successfully submitted
> >
> > And that's it until I interrupt globusrun-ws.
> >
> > Finally, there's container.log. And *This* contains something the previous
> > thread discussed but did not follow up on: missing
> > "globus-scheduler-provider-sge". Indeed I don't have one. But to my
> > knowledge, I am not trying to submit via MDS. So why is this even appearing?
> >
> > 2006-06-29 22:12:49,124 WARN  usefulrp.GLUEResourceProperty [GLUE refresher
> > 0,runScript:315] Script Execution error when executing shell
> > /software/globus/gt4.0.2/libexec/globus-scheduler-provider-sge
> > java.io.IOException: java.io.IOException:
> > /software/globus/gt4.0.2/libexec/globus-scheduler-provider-sge: not found
> >         at java.lang.UNIXProcess.<init>(UNIXProcess.java:148)
> >         at java.lang.ProcessImpl.start(ProcessImpl.java:65)
> >         at java.lang.ProcessBuilder.start(ProcessBuilder.java:451)
> >         at java.lang.Runtime.exec(Runtime.java:591)
> >         at java.lang.Runtime.exec(Runtime.java:429)
> >         at java.lang.Runtime.exec(Runtime.java:326)
> >         at
> > org.globus.mds.usefulrp.GLUEResourceProperty$PeriodicExecutor.runScript(
> > GLUEResourceProperty.java:299)
> >         at org.globus.mds.usefulrp.GLUEResourceProperty$PeriodicExecutor.run
> > (GLUEResourceProperty.java:260)
> >         at java.lang.Thread.run(Thread.java:595)
> >
> > That's all I have. Ideas? Suggestions?
> >
> >
>


Hello.
I finally got it working but with another sge.pm file. Now I can send
ws and non-ws jobs to a SGE queue trough Globus.

First, check if you can send jobs using qsub, i.e., just using Grid
Engine without globus.

If that works ok, change the reporting parameters according to Melvin
Koh's suggestion and try with globus.

If that doesn't work, try this sge.pm file (attached).

I hope this helps you.

Regards.


-- 
Diego Bello Carre?o
Estudiante Memorista de Ingeniería Civil Informática
UTFSM, Valparaíso, Chile
Usuario #294897 counter.li.org


    [ Part 2, Application/X-PERL (Name: "sge.pm") 21 KB. ]
    [ Unable to print this part. ]


    [ Part 3: "Attached Text" ]

    [ The following text is in the "iso-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

---------------------------------------------------------------------
To unsubscribe, e-mail: globus-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: globus-help at gridengine.sunsource.net



More information about the gridengine-users mailing list