[gridengine users] resource management and over-subscription?

Julie Ashworth julie.ashworth at berkeley.edu
Wed Nov 23 02:58:39 UTC 2011

On 22-11-2011 23.26 +0100, Reuti wrote:
> sorry for the confusion: the "no" was meant for memory + consumable resource + diskspace (i.e. the $TMPDIR). None of them will be freed.

Ah, this is my fault - a bad combination of optimism and reading too quickly ;).
> What you can do if your jobs support it: use the checkpointing feature in SGE. A suspension of a job (or queue) will then reschedule the job and free the used resources this way. But:

Checkpointing may be too situational, given that I can't control the applications used. Slotwise preemption (subordinate queues) seems more robust and simple... and/or writing a script to suspend/oversubscribe nodes.

Thanks again for your help and advice.

Julie Ashworth <julie.ashworth at berkeley.edu>
PGP Key ID: 0x17F013D2

More information about the users mailing list