[torquedev] [Bug 122] New: pbs_mom needs config option to trigger job specific temporary directory creation/removal on the Mother Superior only

bugzilla-daemon at supercluster.org bugzilla-daemon at supercluster.org
Tue Apr 19 19:56:07 MDT 2011


http://www.clusterresources.com/bugzilla/show_bug.cgi?id=122

           Summary: pbs_mom needs config option to trigger job specific
                    temporary directory creation/removal on the Mother
                    Superior only
           Product: TORQUE
           Version: 2.4.x
          Platform: All
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P5
         Component: pbs_mom
        AssignedTo: knielson at adaptivecomputing.com
        ReportedBy: chris at csamuel.org
                CC: torquedev at supercluster.org
   Estimated Hours: 0.0


Created an attachment (id=76)
 --> (http://www.clusterresources.com/bugzilla/attachment.cgi?id=76)
Only remove $tmpdir on the mother superior, otherwise drop through

/* Found on 2.4, but exists in 2.5, 3.0 and trunk too */

If you have a global scratch filesystem which is being used for your job
specific temporary directories via the pbs_mom directive $tmpdir then there
is no need for every pbs_mom in the job to create and remove the directory
at the same time.

It would be great to have a directive in the pbs_mom.conf file which would
cause the pbs_mom to only do the creation/removal on the Mother Superior
and not on the rest of the sisterhood.

At VLSCI we've had to hack this in unconditionally to work around an
apparent bug in the version of Panasas we've just upgraded to which
causes a large number of pbs_mom processes to hang if they all try
and remove the same directory tree at the same time.

I've attached the trivial patch we've used as an example (though of
course without the check for any configuration setting).

-- 
Configure bugmail: http://www.clusterresources.com/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.


More information about the torquedev mailing list