[torquedev] Bug in post_epilogue()
jacksond at clusterresources.com
Tue Aug 28 17:47:37 MDT 2007
I was putting in a simple job pointer caching system within
scan_for_terminated to retry if mompost failed to send an obit. This
would work and retry until the mom was recycled. However, the retry is
lost at the point. Clearly, it would be better to save the fact that
the obit need to be sent into the job state record so the retry
continues after a mom recycle. Had you already designed an effective
way of accomplishing this?
On Tue, 2007-08-28 at 16:18 -0700, Garrick Staples wrote:
> On Tue, Aug 28, 2007 at 10:38:38AM -0600, Dave Jackson alleged:
> > Garrick,
> > I don't see a fix in trunk. In fact, the bug was first detected and
> > reported on a recent trunk based distribution. Is there a possibility a
> > fix did not get committed? Does the fix involve mompost return code
> > checks and retry from within scan_for_terminated()?
> Yes, I had thought I fixed up those problems when I rewrote all that code.
> torquedev mailing list
> torquedev at supercluster.org
More information about the torquedev