On Thu, 25 Aug 2011 00:26:28 +0000
손성환 <[email protected]> wrote:

> Hi~
> 
> I have some question
> 
> 1)
> I have problem about Eqw job state.
> 
> I think almost Eqw job is working directory access problem.
> 
> For example, delete working directory when job has qw status.
> 
> So, I want to clear this Eqw job automatically (not manually 'qmod' command)
> 
> How to solve this problem?


You've described this mostly in terms of what you want to do rather than what 
you want to accomplish.  It might be easier to help if you told us what the 
goal is rather than presenting a partial solution.  You might be able to get 
the prolog to check for the existence of the working directory and bounce the 
job out by returning 99.  In such circumstances the job shouldn't enter Eqw. 
However this might fail if the working directory was only intermittently 
unavailable and where the working directory was permanently deleted would just 
keep bouncing the job in and out of the queue.

Alternately if its the manual part that bothers you a cron job could check for 
jobs in Eqw and clear said status. 

  


> 
> 2)
> Sometimes despite process is done, job still has run status.
> 
Hard to say without more information . Are you sure that the whole job has 
finished not just the process you consider interesting?  Check if the job's 
shepherd is still present.  Does it have any descendants?  Is the job's spool 
directory still present.  Is there anything in the logs.    

> I have been delete this job manually.
> 
> Is bug in SGE? or configuration problem?
> 
> How to solve this problem?
> 
> thank you


-- 
William Hay <[email protected]>

Attachment: pgpBVYWQUtARp.pgp
Description: PGP signature

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

Reply via email to