[GE users] SGE-6.0u7.1: Tuning of scheduler interval?

Erik Soyez E.Soyez at science-computing.de
Mon Jan 14 08:50:25 GMT 2008

Good morning list,

a user complained, because he qdel'ed a 32-cpu job in order
to get the next 32-cpu job started (which was first in qstat,
means highest priority).  But instead of his own job two 16-cpu
jobs (lowest priority, last in qstat) of another user started.

The most plausibel cause is that because of the load_threshold
of at least one of the nodes no 32-cpu job could be startet in
the next schedule_interval (0:0:15).

Reservations are not an option (because no reservation user max
can be defined in 6.0).

A longer schedule_interval seems to be the best option to minimize
the risk with minimal effort.  But what happens with

   schedule_interval                 0:0:30
   flush_finish_sec                  0:1:0 ?

Does flush_finish_sec have any effect at all?  The man page is
not quite clear with this detail....


Erik Soyez.


Vorstand/Board of Management:
Dr. Bernd Finkbeiner, Dr. Florian Geyer,
Dr. Roland Niemeier, Dr. Arno Steitz, Dr. Ingrid Zech
Vorsitzender des Aufsichtsrats/
Chairman of the Supervisory Board:
Prof. Dr. Hanns Ruder
Sitz/Registered Office: Tuebingen
Registergericht/Registration Court: Stuttgart
Registernummer/Commercial Register No.: HRB 382196 

To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net

More information about the gridengine-users mailing list