Opened 11 years ago

Last modified 9 years ago

#607 new defect

IZ2814: Removing a job causes DRMAA to behave in a non-compliant way

Reported by: szalik Owned by:
Priority: normal Milestone:
Component: sge Version: 6.2
Severity: Keywords: drmaa
Cc:

Description

[Imported from gridengine issuezilla http://gridengine.sunsource.net/issues/show_bug.cgi?id=2814]

        Issue #:      2814             Platform:     All      Reporter: szalik (szalik)
       Component:     gridengine          OS:        All
     Subcomponent:    drmaa            Version:      6.2         CC:    None defined
        Status:       NEW              Priority:     P3
      Resolution:                     Issue type:    DEFECT
                                   Target milestone: ---
      Assigned to:    dagru (dagru)
      QA Contact:     templedf
          URL:
       * Summary:     Removing a job causes DRMAA to behave in a non-compliant way
   Status whiteboard:
      Attachments:

     Issue 2814 blocks:
   Votes for issue 2814:


   Opened: Mon Dec 1 10:25:00 -0700 2008 
------------------------


Hi,

According to the latest DRMAA 1.0 specification [1] and also discussions on the
mailing list [2] drmaa_wifaborted() should be true only if the job was cancelled
*before* entering the running state.

But in SGE drmaa_wifaborted() is true also when the job is deleted (qdel) or
terminated with DRMAA itself (DRMAA_CONTROL_TERMINATE).

According to my discussions with DRMAA authors the return status from
drmaa_wait() in case of removing an already running job should reflect the fact
that the job is signalled (drmaa_wifsignaled() = true).

[1] From http://www.ogf.org/documents/GFD.133.pdf:

drmaa_wifaborted(...)
Evaluates into 'aborted', a non-zero value if 'stat'
was returned for a job that ended before entering the running state.

[2] http://www.ogf.org/pipermail/drmaa-wg/2008-August/thread.html

   ------- Additional comments from szalik Thu Dec 4 12:17:06 -0700 2008 -------
"But in SGE drmaa_wifaborted() is true also when the job is deleted (qdel) or
terminated with DRMAA itself (DRMAA_CONTROL_TERMINATE)."

should read:

"But in SGE drmaa_wifaborted() is true also when the job is deleted (qdel) or
terminated with DRMAA itself (DRMAA_CONTROL_TERMINATE) while it is already in
running state"

Change History (0)

Note: See TracTickets for help on using tickets.