[GE users] qrsh vs. qsub and exit 99 behavior

sophana jobarjo78 at yahoo.fr
Mon May 17 15:37:22 BST 2004


I suppose that the problem is that qrsh does have to reconnect stdin and 
stdout after a requeuing.
this has no sense since the exit code is made after the connection has 
been made.

qsub does not have the problem because there is no interactive connection

I don't see how would be a correct behaviour of qrsh re-queing a job 
that has already finished.

Kirk Patton wrote:

>Hello all,
>
>I have been working to incorporate SGE at my work to replace LSF.
>This are going quite well, but I have noticed that the behavior of
>qrsh is different than that of qsub when it come to requeueing on
>an exit value of 99 from the prolog script.
>
>qsub will not run a job if it gets an exit status of 99 and will instead 
>requeue and run only if the exit status of the prolog is 0. :-)
>
>qrsh will not run a job if its prolog exits with 99 but it will *not*
>requeue. :-(
>
>I am trying to approximate the familiar behavior of LSF as that is what
>users at my site are expecting.  LSF has a consistent behavior between its
>interactive and batch submission command.  They both honor the exit status
>and pend until it is 0.
>
>Can this difference in behavior be change?
>
>Thanks,
>Kirk
>
>  
>


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