[GE users] Serialising jobs

Chris Dagdigian dag at sonsorol.org
Tue Sep 9 17:08:11 BST 2008


This is the first job:

  $  qsub -N firstTask /path-to-my-script.sh

This is the second job:

  $ qsub -hold_jid firstTask /path-to-my-second-script.sh


The key here is that in the first job you named the job in a way that  
you can refer to it later without knowing a job ID in advance, the  
second important bit is the qsub argument "-hold_jid" which is how you  
tell the job about the prior dependency.

Regards,
Chris






On Sep 9, 2008, at 12:02 PM, Paolo Supino wrote:

> Hi
>
> how can I configure SGE to launch a specific job only after a  
> previous specific job finished running (the latter is dependent on  
> the results of the first)?
>
>
>
> --
> Paolo
>
> On Sep 8, 2008, at 19:02, Chris Dagdigian <dag at sonsorol.org> wrote:
>
>> Hi Paolo,
>>
>> Do a "qconf -mq all.q" and look at the "hosts" line of the queue  
>> configuration. If you mistaken host is listed there just remove it  
>> and the change will be made live as soon as you exit from the  
>> editor session.
>>
>> If you just see a SGE hostgroup such as "@allhosts" mentioned in  
>> your queue config then you need to edit the hostgroup to remove the  
>> host in question that can be dine via:
>>
>> qconf -mhgrp @allhosts
>>
>> ... or similar. Again, just like all SGE interactive configuration  
>> actions, your change is "live" in the system once you exit from the  
>> editor session.
>>
>> I've personally had better luck running the SGE installer scripts  
>> in "remove" mode to delete execution hosts. If you go into  
>> $SGE_ROOT and run "./ins_sge -h" the help syntax will show you how  
>> to invoke the script such that you are removing rather than  
>> installing the exechost. I don't recall from memory what the exact  
>> command is through.
>>
>> -Chris
>>
>>
>> On Sep 8, 2008, at 11:50 AM, Paolo Supino wrote:
>>
>>> Hi
>>>
>>> I mistakenly added the master host as an execution host and now  
>>> I'm trying to remove it. But when I try to remove it using qconf - 
>>> de [hostname] it prints out the following error message: Host  
>>> object [hostname] is still referenced in cluster queue "all.q".  
>>> How do I remove the host from the all.q so I can remove it from  
>>> the execution list?
>>>
>>>
>>>
>>>
>>> --
>>> TIA
>>> Paolo
>>>
>>
>>
>> ---------------------------------------------------------------------
>> 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


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