Hi,

submitting a job in 6.2u5 with:

$ qsub -l exclusive,h_rt=120 -q parallel@@infiniband -pe mpich_mslave 16 -w e 
-R y test.sh
Your job 2249423 ("test.sh") has been submitted

works fine, and I can do it even a couple of times. But then it gets:

$ qsub -l exclusive,h_rt=120 -q parallel@@infiniband -pe mpich_mslave 16 -w e 
-R y test.sh
Unable to run job: error: no suitable queues.

This behavior shouldn't change at submisson time at all (unless there would be 
max_u_job/max_job limits in SGE's configuration, and the error message would be 
different then). There is also no JSV defined.

Using "-w v" instead of "-w e" leads to the assumption, that it checks the 
acutal usage in the cluster (like "-w p") at submission time. But right now 
only my jobs request the "exclusive" resource, and "-w v" should assume an 
empty cluster instead of:

Job 2249428 (-l exclusive=TRUE,h_rt=120) cannot run in queue "node141" because 
exclusive resource (exclusive) is already in use.

This is a line from "-w v" - there is no queue "node141", it's a name of an 
exechost, and the exclusive resource is attached to the exechosts. Did anyone 
encounter a similar behavior? Maybe April Fool's joke was activated one day in 
advance.

-- Reuti
_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to