[torqueusers] tiny jobs starve everyone else

Garrick Staples garrick at usc.edu
Tue Feb 8 17:43:35 MST 2005


We're currently having a situation where 1 user has submitted a few tens of
thousands of tiny single node jobs that run for about a minute.  This has
completely starved out all of the other jobs.  The tiny jobs have been routed
to an isolated queue with 16 dedicated nodes.  

But despite the isolation, and the existance of dozens of other jobs that should be
runnable, pbs_server never seems to get around to telling Maui about them.
'checkjob' can't find any of the other user's jobs.

This seems more like a limitation in Torque rather than a configuration/policy
decision.  Any thoughts?


-- 
Garrick Staples, Linux/HPCC Administrator
University of Southern California
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
Url : http://www.supercluster.org/pipermail/torqueusers/attachments/20050208/083c9652/attachment.bin


More information about the torqueusers mailing list