[torquedev] r1770 and r1771?
garrick at usc.edu
Tue Jan 29 19:09:01 MST 2008
On Tue, Jan 29, 2008 at 01:33:57PM -0800, Garrick Staples alleged:
> On Mon, Jan 28, 2008 at 04:12:27PM -0800, Garrick Staples alleged:
> > I'm curious about the rational for r1770 and r1771. It doesn't look like it
> > would pass the job's session id anymore. Why define a new resource?
> I'm going to remove these changes, and the code that they fix. I can't find
> a reason for them.
New code has been checked into trunk and 2.2-fixes that does what is required.
The original problem is that the jobs session_id wasn't being sent to
pbs_server for very short jobs. This is important when keep_completed is
The session_id was being sent in a jobstat request from pbs_server, but not in
the jobstat sent from pbs_mom before the obit.
The incorrect code was encoding the session_id as a resource instead of an
attribute. When encoded as an attribute, pbs_server would segfault because of
a strlen computing error in server/req_jobobit.c.
Now, all new attrs (with ATR_FLAG_SEND bit) are correctly encoded into the
jobstat and pbs_server won't segfault when it sees a non-resource attribute.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Size: 189 bytes
Desc: not available
Url : http://www.supercluster.org/pipermail/torquedev/attachments/20080129/c40aea30/attachment.bin
More information about the torquedev