Am 06.03.2014 um 20:18 schrieb Leonardo Morelli:

> El 28/02/14 18:12, Reuti escribió:
>> Hi,
>> 
>> Am 28.02.2014 um 21:33 schrieb Leonardo Morelli:
>> 
>>> Hello everyone,
>>> when I reboot the frontend (so sge restarts), it appears an old, 
>>> already-finished job in the queue (qw state), which blocks all other jobs 
>>> from running.
>>> After manually deleting that job everything goes back to normal, but with a 
>>> reboot it appears again.
>>> 
>>> How can I delete that job forever?
>> 
>> I never saw this. Do you use classic or berkeley spooling?
>> 
>> For the latter you can shut down SGE, rename the file in spooldb directory 
>> (where the qmaster's directory resides too) "sge_job" to something else and 
>> restart SGE. It should automatically being recreated during startup.
>> 
>> For classic spooling there should be a directory in qmaster's directory with 
>> the jobs.
>> 
>> 
>> Before doing such things, it's of course good to have a backup.
>> 
>> -- Reuti
>> 
>> 
>>> Thanks!
>>> Leo
> 
> I believe it's classic spooling...
> I removed the troubling job file in 
> gridengine/default/spool/qmaster/jobs/00/0000/ and it seemed to have 
> disappeared after restarting sge, was that a safe fix?

Yes. I do this occasionally on an exechost, where it's still a plain file even 
with BDB.

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

Reply via email to