[GE users] job state "dr" and cleaning up

Vladimir vlists at veus.hr
Thu Jul 22 06:28:06 BST 2004


Patrice Seyed wrote:

>Troubleshooting an nfs issue, and ended up with some jobs in the "dr" state
>and the queues in state "a" for alarm, I suppose because the load avg was
>high. What is "dr" and how do I delete these jobs; using qdel, restarting
>sge on the node, and restarting sge on the master haven't cleared these..
>  
>

dr means that deletion of the job has been requested but for some reason 
the worker node's SGE process is not responding. I often see this happen 
when a machine crashes or machine is hosed due to some network issue ie. 
NFS. Usually after rebooting the worker node the job would disappear. 
Another alternative if you want job to "disappear" is to use the -f 
(force) flag ie.

qdel -f <jobid>

Vladimir


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