[torqueusers] question about creating checkpoint

Al Taufer ataufer at adaptivecomputing.com
Mon Mar 7 10:38:21 MST 2011


----- Original Message -----
> Hi,
> Is it possible to manually create a checkpoint despite the fact that
> no "-c" was
> used in the submission? A user forgot to use "-c" when he used "qsub"
> and now we
> want to create a checkpoint and sutdown the server.
> 

I do not know if it you can checkpoint the job.  The checkpointing is done by the script listed in your mom config file as $checkpoint_script.  This script, if not modified, will be using the cr_checkpoint command installed by BLCR.  The man page for this command lists several ways that the job can be started so it is checkpointable.  If the -c had been used on the qsub then Torque would have run the job using the cr_run command from BLCR, but since -c was not specified Torque just ran the job normally.  The only way that the $checkpoint_script and/or cr_checkpoint might work is if the job was originally linked with one of the libraries listed in the man page for cr_checkpoint.

Al Taufer

> 
> Thanks for any feedback.
> // Naderan *Mahmood;
> 
> 
> 
> _______________________________________________
> torqueusers mailing list
> torqueusers at supercluster.org
> http://www.supercluster.org/mailman/listinfo/torqueusers


More information about the torqueusers mailing list