[GE users] qmaster memory and performance

Raphael Y. Rubin rafi at cs.caltech.edu
Fri Apr 8 20:21:26 BST 2005


We're running sge 6.0u1 and have been seeing issues with the master using
quite a bit of memory and being a little unresponsive.

example:
After submitting ~30000 jobs last night, the master was unresponsive, and
our shadow took over.  This morning, the master was using over 2.5GB
  PID USER      NI  VIRT SWAP  RES  SHR #C S %CPU %MEM   TIME COMMAND
15094 sgeadmin   0 2837m  67m 2.7g 5520  0 S  0.0 68.5   0:24 sge_qmaster

And its still in that state after deleting all of my jobs (other users
account for <10 jobs still on the cluster).  And new jobs aren't
starting.

Is this sort of behavior something that's been fixed in the more recent
maintainence releases?  Are there some configuration options we should
know about that will fix this sort of thing?

Rafi Rubin
Caltech CS

---------------------------------------------------------------------
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