[GE users] Jobs can not run because queue instance is not contained in its hard queue list

FL lengyel at gmail.com
Fri Jun 1 13:00:20 BST 2007

    [ 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. ]

Without analyzing the problem further, I reinstalled the sge qmaster
and rebuilt the queues from scratch. For some reason the in place
upgrade from 6.0u7 to 6.0u10 wasn't so smooth.  My jobs were
able to run after rebuilding (most) of what I had,

I've been using host groups to control which usersets have
access to various machines. but those usersets and hostgroups
have to be attached to the configuration of a queue. Some
users get access to everything, some don't.

@all.q still exists, but is accessible only to deadlineusers.
It would be nice to write  /opt/$arch/prog
but the users who want 64 bit systems aren't indifferent
to $arch. Possibly the users who run on P3's or P4's wou;d
be indifferent.Since I'm redoing things,  I should  implement
your suggestion; I have
users who would really need  64 bits systems but
would put up with with p4s if no 64 bit systems were available.

On 6/1/07, Reuti <reuti at staff.uni-marburg.de> wrote:
> Am 01.06.2007 um 01:22 schrieb FL:
> > I've added 64 bit nodes to a cluster running SGE 6.0u10.
> > I have a queue for 64 bit nodes called x86_64.q which was
> > cloned from another queue, where I have no trouble submitting jobs.
> >
> > My trouble is that trivial scripts such as
> >
> > #!/bin/bash
> > #$ -S /bin/bash
> > #$ -q x86_64.q
> > hostname
> >
> > don't run, and it's not obvious why not.
> >
> > The error message is:
> > Jobs can not run because queue instance is not contained in its
> > hard queue list
> What is the output of " qstat -j <jobid>" for this job, in detail the
> hard queue list there?
> >
> > I don't see why this should be so. I cloned the queue from one that
> > had no problem--the only
> > difference is in the hostlist.
> It might be a matter of personal taste, but why did you create an
> identical queue for just another platform, which is in some way a
> Torque setup. You can have all machines in one queue and select the
> architecture if necessary with "-l arch=lx24-amd64" as resource
> request. This would even allow to run programs, when you don't care
> where it's running by having 32- and 64-bit versions of a program in
> a similar directory structure:
> /opt/lx24-amd64/myprog
> /opt/lx24-x86/myprog
> and in the job script you run:
> /opt/$ARC/myprog
> But back to your question: you checked with "qstat -f" the existence
> of the queue instances, and also "qhost" shows the new hosts as
> available?
> -- Reuti
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net

More information about the gridengine-users mailing list