[GE users] how long exists jobs files?

reuti reuti at staff.uni-marburg.de
Wed Oct 20 15:02:42 BST 2010

    [ The following text is in the "utf-8" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some characters may be displayed incorrectly. ]

Am 20.10.2010 um 02:38 schrieb sonsh99:

>>> I was wondering whether jobs files exists even though job was done.
>>> I think that jobs files($SGE_ROOT/$S?GE_CELL/spool/qmaste?r/jobs/)
>>> related with qstat information. right?
>> so you are using classic spooling I assume.
> Is there another way spooling?

Yes, BDB (Berkeley DB). Then you even don't see the script file in qmaster's spool directoty any longer, but only on the node when it gets executed.

>>> I was developing monitoring tools, which needs qstat information. 
>> What in particular do you need, what is not available elsewhere?
> I need information which using 'qstat -j`
> (stderr_path_list , stdout_path_list , script_file)
> but qacct was not include this information.

This is even not recorded in ARCo AFAIK.

>>> but qstat information is disappear when job was done.
>>> Is possible keep jobs files?
>> Why are you not using the accounting file for finished jobs?
> because accounting files not include information (stderr_path_list , stdout_path_list , script_file )

When you need this only for running and past jobs, you could write a global prolog, which will write this information into a file per job (to avoid race conditions when you would write to one file only).

exec 3> /usr/sge/default/common/history/$JOB_ID.${SGE_TASK_ID/undefined/1}
echo $JOB_SCRIPT >&3
exec 3>&-

We do a similar thing to have the ability to put additonal information about the job (from the job's "context") in the sent email after the job, as of time of sending the email the `qstat` doesn't provide the information any longer. Diffence is, that for us I create such a file in a local spool directory, as it needs only to be known on this particluar machine which sends the email. In your case you need it global instead.

-- Reuti

> -- Reuti
> thank you
> ------------------------------------------------------
> http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=288506
> To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].


To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].

More information about the gridengine-users mailing list