Claudiu Demian wrote on 23/11/10 20:35:
> Hi,
>
> JOB_MANAGER=pbs
> JOB_MANAGER=pbs
did you mean here CE_BATCH_SYS?
on my MPI enabled CREAM CE I have
JOB_MANAGER=pbs
CE_BATCH_SYS=torque
But I don't know if it is critical and could cause a problem you have.
You can try to change CE_BATCH_SYS to torque and reconfigure CE.
By the way, with what options did you invoke yaim?
Can it be that specifying CPUNumber=16 scheduler tries to fit all jobs
on one WN? Did you try to specify e.g. CPUNumber = 20? What happens then?
> Every WN has 16 logical CPU's
> np = 16
>
> Regarding the time at the WNs, I've deployed the same working ntp.conf
> on all of them, but they did not update their time. Because normal jobs
> where submitted without problems I did not look further into the issue.
> Could this have anything to do with the problem?
I think it's better to eliminate all issues what can cause a potential
problem...
Nikolay.
>
> Cheers,
> Claudiu
>
> În Mar, Noiembrie 23, 2010 6:54 pm, Nikolay Kutovskiy a scris:
>> What values for JOB_MANAGER and CE_BATCH_SYS do you have?
>>
>>
>> Nikolay.
>>
>>
>>
|