<HTML dir=ltr><HEAD><TITLE>Re: [Mauiusers] preemptor job holds up the rest of the system</TITLE>
<META http-equiv=Content-Type content="text/html; charset=unicode">
<META content="MSHTML 6.00.6000.16981" name=GENERATOR></HEAD>
<BODY>
<DIV id=idOWAReplyText72945 dir=ltr>
<DIV dir=ltr><FONT face=Arial color=#000000 size=2>What happens if you try changing your BACKFILLPOLICY to FIRSTFIT instead of BESTFIT (just to try one of the other backfill algorithms)?</FONT></DIV>
<DIV dir=ltr><FONT face=Arial size=2></FONT> </DIV>
<DIV dir=ltr><FONT face=Arial size=2>Does it behave differently?</FONT></DIV>
<DIV dir=ltr><FONT face=Arial size=2></FONT> </DIV>
<DIV dir=ltr><FONT face=Arial size=2>--Joe</FONT></DIV></DIV>
<DIV dir=ltr><BR>
<HR tabIndex=-1>
<FONT face=Tahoma size=2><B>From:</B> mauiusers-bounces@supercluster.org on behalf of Bill Wichser<BR><B>Sent:</B> Tue 1/26/2010 2:03 PM<BR><B>To:</B> Bill Wichser<BR><B>Cc:</B> mauiusers@supercluster.org<BR><B>Subject:</B> Re: [Mauiusers] preemptor job holds up the rest of the system<BR></FONT><BR></DIV>
<DIV>
<P><FONT size=2>Still haven't solved this problem. Again, here is the scenario:<BR><BR>A system reservation is in place for 10 days from now.<BR><BR>Users submit jobs for say 15 days and land in the top of the IDLE stack<BR>(showq)<BR><BR>Other jobs which are submited, say for 2 days or 4 hours, are sitting<BR>behind the 15 day jobs waiting to run.<BR><BR>The jobs holding everything up were submitted to a CLASS with preemptor QOS.<BR><BR>Removing the system reservation allows everything to run. Manually<BR>placing the long running jobs on HOLD allow the rest to run. My<BR>BACKFILLDEPTH more than covers all the jobs in the IDLE state as<BR>indicated by the asterisk next to job number.<BR><BR>Other maui configurations on other clusters, without this preemptor<BR>stuff, work as expected. I'm out of clues here! The relevant stanza<BR>from maui.cfg looks like this:<BR><BR>BACKFILLPOLICY BESTFIT<BR>BACKFILLMETRIC PROCSECONDS<BR>BACKFILLDEPTH 20<BR>RESERVATIONPOLICY FIRSTFIT<BR>RESERVATIONDEPTH[0] 24<BR>RESDEPTH 24<BR><BR>Any clues on what I might be missing?<BR><BR>Thanks,<BR>Bill<BR><BR><BR>Bill Wichser wrote:<BR>> We've been running Maui/Torque for quite a few years here. On our<BR>> latest cluster there has been a need to start using preemptive queues<BR>> (classes). This has done fine except for a problem we had when setting<BR>> a system reservation.<BR>><BR>> So a system reservation was set across the entire cluster. Normally, on<BR>> clusters without preemptive scheduling, jobs exceeding the wallclock<BR>> time will block, allowing other shorter jobs to backfill in.<BR>><BR>> When using a preemptive class, when a job exceeeds available wallclock<BR>> time, it remains in the IDLE state preventing other backfill jobs from<BR>> running. Yes, these other jobs are preemptee class.<BR>><BR>> Manually placing the overruns on hold allows these other jobs to get<BR>> scheduled.<BR>><BR>> Am I simply missing something in the configuration? Or is this the<BR>> expected behavior when using preemptor/preemptee classes/QOS?<BR>><BR>> Thanks,<BR>> Bill<BR>> _______________________________________________<BR>> mauiusers mailing list<BR>> mauiusers@supercluster.org<BR>> <A href="http://www.supercluster.org/mailman/listinfo/mauiusers">http://www.supercluster.org/mailman/listinfo/mauiusers</A><BR>_______________________________________________<BR>mauiusers mailing list<BR>mauiusers@supercluster.org<BR><A href="http://www.supercluster.org/mailman/listinfo/mauiusers">http://www.supercluster.org/mailman/listinfo/mauiusers</A><BR></FONT></P></DIV></BODY></HTML>