[GE users] Checkpointing and task

reuti reuti at staff.uni-marburg.de
Tue Nov 3 19:21:21 GMT 2009


Am 03.11.2009 um 16:05 schrieb aom:

> I was able to set up a checkpoint
> environment:
> --------------------------
> ckpt_name          check_user_scripts
> interface          APPLICATION-LEVEL
> ckpt_command       /opt/sge/ckpt/userckpt_app.sh $ckpt_dir $job_id
> migr_command       /opt/sge/ckpt/usermigrate_app.sh $job_pid  
> $ckpt_dir $job_id
> restart_command    NONE
> clean_command      /opt/sge/ckpt/userclean_app.sh $ckpt_dir $job_id
> ckpt_dir           /scratch/checkpoint
> signal             NONE
> when               xsmr
> -------------------------
> which is working as expected.
> But I'd like to know what is the best way to deal with tasks in job  
> arrays. It seems to be possible to suspend a specific task using  
> the syntax 'qmod -sj <jobid>.<taskid>' but I was unable to find a  
> way to pass the task id to the scripts (I tried to use:
> /ckpt/usermigrate_app.sh $job_pid $ckpt_dir $job_id $task_id
> but it was refused).
> I there a way to get the task number in the environment scripts ?

on the one hand you are right, as it's not available as a pseudo  
variable (defined in source/libs/uti/config_file.c).

But it seems to be exported to the script as usual as SGE_TASK_ID.

-- Reuti

> we are using GE 6.1.
> Thanks
> ------------------------------------------------------
> http://gridengine.sunsource.net/ds/viewMessage.do? 
> dsForumId=38&dsMessageId=224834
> 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