[torqueusers] Some questions on elipogue/prologue
lolitushka at gmail.com
Mon Jun 4 13:08:44 MDT 2007
On 6/4/07, Garrick Staples <garrick at clusterresources.com> wrote:
> On Mon, Jun 04, 2007 at 11:13:02AM +0300, Alexander Piavlo alleged:
> > Dear all,
> > I have some questions on elipogue/prolouge execution:
> > - Which script is executed first prologue.user or prologue if both exist?
> > - Which script is executed first epilogue.user or epilogue if both exist?
> pro/epilogue execute before the .user scripts.
Intuitively i would expect the order to be:
prologue and then prologue.user
and vice versa for epilogue:
epilogue.user and then epilogue
And what happens then
(a) a job fails to stagein files before start
and is put W state to later be resubmited if it is rerunable.
(b) if job fails during execution
Are any of pro/epilogue scripts get executed in case (a) or (b)?
If not how can i make run some cleanup scripts in these cases?
> > - Then a job is cancelled the epilogue.precancel is executed, but do the
> > epilogue, epilogue.user & epilogue.parallel are also executed and in
> > which order?
> Yes, all epilogue scripts run. epilogue.parallel executes on sisters
> before epilogue and epilogue.user.
> > - Are the scripts epilogue.parallel & prologue.parallel executed an all
> > pbs_mom sister nodes or on only defined at /var/spool/pbs/aux/$jobid ?
> All sister nodes, which is the same list as /var/spool/pbs/aux/$jobid
> (except for the execution node.)
> > - Are the epilogue.parallel script executed on sister nodes in parrallel
> > with epilogue on mom node?
> > - Are the prologue.parallel script executed on sister nodes in parrallel
> > with prologue on mom node?
> Nope. parallel executes on sisters first.
> torqueusers mailing list
> torqueusers at supercluster.org
More information about the torqueusers