[GE users] changing job script

reuti reuti at staff.uni-marburg.de
Wed Feb 11 11:30:39 GMT 2009


Am 11.02.2009 um 12:09 schrieb craffi:

> The "-o" and "-e" switches are supposed to point at files, not  
> folders:
>
>   #$ -o /SGEoutput/job-out.txt
>   #$ -e /SGEoutput/job-err.txt

It can also point to a directory, which will then use the standard  
name for these files in the specified directory.

>
> Also, "/SGEoutput" path looks like it would be on your root
> filesystem, not on a NFS path so it is likely that this folder does
> not exist on your compute nodes unless your cluster is diskless and
> the entire "/" filesystem is NFS resident.
>
>
>
> -Chris
>
>
> On Feb 11, 2009, at 5:45 AM, tamara wrote:
>
>> hi,
>>
>> i used job "simple.sh" to test the system and it's working then i
>> added some things to the script of this job:
>>
>> #$ -o /SGEoutput
>> #$ -e /SGEoutput
>>
>> that assign the standard and error output file (and i already made
>> this folder "SGEoutput" sharing through nfs"
>>
>> but now i have a problem the job instead of going from pending job
>> to running to finished, the job acctually going from pending to
>> running to pending again, why is this happening and why it's not
>> finished?

Is one node after the put into error state? Is there any output in  
$SGE_ROOT/default/spool/qmaster/messages or the one of the affected  
nodes?

You can write to this directory?

-- Reuti

------------------------------------------------------
http://gridengine.sunsource.net/ds/viewMessage.do?dsForumId=38&dsMessageId=103403

To unsubscribe from this discussion, e-mail: [users-unsubscribe at gridengine.sunsource.net].



More information about the gridengine-users mailing list