[GE users] trapping SIGUSR2 signal

Reuti reuti at staff.uni-marburg.de
Thu Apr 13 19:39:33 BST 2006


Hi,

Am 13.04.2006 um 16:39 schrieb Sreenath Nampally:

> Hi
>
> I have a fast.q with 'hard limit wall clock time' as 3 minutes and
> 'notify time' as 2 minutes.
>
> I would like to trap the SIGUSR2 signal whenever a job in fast.q
> is about to exceed the hard limit wall clock time and either
> resubmit or reschedule that job to a different queue.
>
> I am able to trap the SIGUSR2 signal but the signal is recieved
> when the job is killed ,  not 2 minutes (notify time 2 minutes).
> before the SIGKILL is issued.   In my case I should recieve the  
> SIGUSR2
> after a minute of job execution (hard limit wall clock time
> 3 minutes - notify time, 2 minutes)  but seems like SIGUSR2 and  
> SIGKILL
> are happening at the same time.

how are you trapping the signal in the script and the program, as it  
will be send to the whole process group and not just only the script?

But it will also extend the runtime by the notify time, i.e. the  
SIGUSR2 should come after 3 minutes, and after additional 2 minutes  
it will be killed.

Another option in your case could be to use a soft limit instead on  
the notify option.

-- Reuti


> When my script traps SIGUSR2 and does a resubmit, I get an error that
> the
> job doesn't exist.
>
> Any ideas what might be going wrong here.
>
> thanks
> Sree
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: users-help at gridengine.sunsource.net

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