Am 26.12.2011 um 02:49 schrieb wzlu:

> 於 2011/12/23 上午 03:15, Reuti 提到:
>> Am 12.12.2011 um 02:23 schrieb wzlu:
>> 
>>> 於 2011/12/8 下午 05:10, Reuti 提到:
>>>> Am 08.12.2011 um 10:00 schrieb wzlu:
>>>> 
>>>>> 於 2011/12/8 下午 04:56, Reuti 提到:
>>>>>> Am 08.12.2011 um 01:11 schrieb wzlu:
>>>>>> 
>>>>>>> I tried for all available queues and get the same messages.
>>>>>>> 
>>>>>>> $ qsub -w v c3-parallel-pgimpich2-demo.sh
>>>>>>> Unable to run job: Job 144914 cannot run in queue "cc001-t001" because 
>>>>>>> it is not contained in its hard queue list (-q)
>>>>>>> Job 144914 cannot run in queue "q0-em64t-ge" because it is not 
>>>>>>> contained in its hard queue list (-q)
>>>>>>> Job 144914 cannot run in queue "q0-em64t-ib" because it is not 
>>>>>>> contained in its hard queue list (-q)
>>>>>>> Job 144914 cannot run in PE "mpich" because it only offers 0 slots
>>>>>>> verification: no suitable queues.
>>>>>>> Exiting.
>> Was this solved?
>> 
>> The option -w v assumes an empty cluster, hence some requests (or ACL/RQS) 
>> may prevent to get acces to any queue.
>> 
>> -- Reuti
> The problem is not solved.
> I used another PE (named as mpich1) and used qalter -pe mpich1 # to run these 
> queued jobs.
> 
> I think the problem may be cause by
> 1. old version(SGE version 6.1u3)?
> 2. slots of PE is too large( slots of PE is 81920)?

Woah, you have more than 1700 machines with 48 slots each? I can't say for 
sure, whether there is any limit in any version of SGE in such a range of slots.

-- Reuti


> Is it possible?
> 
> Thanks a lot.
> 
> Best Regards,
> Lu
>> 
>>>>>> Is there any resource request in the job script itself? As you don't 
>>>>>> request any queue on the command line, I would think so. Can you provide 
>>>>>> the header please.
>>>>>> 
>>>>>> -- Reuti
>>>>> Yes, there is "-pe mpich 2" in the job script.
>>>>> 
>>>>> the job script like following:
>>>>> # Specifies the name of the shell to use for the job(recommand)
>>>>> #$ -S /bin/sh
>>>>> # Job name(optional)
>>>>> #$ -N c3-parallel-pgimpich1-demo
>>>>> # Specifies queue name(required)
>>>>> #$ -q q0-em64t-ddr
>>>> Okay, here you request a queue. Specifying a queue request on the command 
>>>> line will override the list which explains the chaning complaints about 
>>>> the command list.
>>> Yes. To easily use the same script, I used option to overwrite the option 
>>> of the script.
>>> 
>>>> Have the queues q0-em64t-ge, q0-em64t-ib and q0-em64t-ddr the same list of 
>>>> hosts and you limit the overall slot count per exechost?
>>>> 
>>>> -- Reuti
>>> There are 112 nodes in q0-em64t-ge. The 64 nodes of q0-em64t-ge are grouped 
>>> as q0-em64t-ddr.
>>> The 64 nodes of q0-em64t-ib are different with q0-em64t-ge and q0-em64t-ddr.
>>> These 3 queues are 1 slot for each node.(total nodes=112+64 and total 
>>> slots=112+64)
>>> These 3 queues can use PE mpich.
>>> 
>>> Thanks a lot.
>>> 
>>> Best Regards,
>>> Lu
>>> 
>>>>> # Specifies number of nodes(required)
>>>>> #$ -pe mpich 2
>>>>> # Specifies output files(optional)
>>>>> 
>>>>> Best Regards,
>>>>> Lu
>>>>> 
> 


_______________________________________________
users mailing list
[email protected]
https://gridengine.org/mailman/listinfo/users

Reply via email to