[GE users] state 't'?

Sean Dilda agrajag at dragaera.net
Thu Apr 14 22:09:23 BST 2005


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Rayson Ho wrote:
>>Transferring.  That's after SGE has assigned the queue and before
>>the job has actually started up.
> 
> 
> I think it's after qmaster assigns the job to execd, and before qmaster
> hears the confirmation from it.

I think it covers a little more than that.  Remember there's a bug in 
6.0u3 where large parallel jobs will stay in 't' for a long time, which 
is caused by qmaster trying to talk to all of the nodes before the job 
starts.  I've had occasions where a job would be in 't' for a while and 
logged into the 'MASTER' compute node and sge_execd hadn't started 
sge_shepherd yet.  As such, I think it covers a little more than just 
waiting on sge_execd to start the process and report back.  Although I 
do seem to recall something about not going from 't' to 'r' until 
sge_execd reports back.

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