[gridengine users] qstat -s p +qstat -r <> qstat -s rp

William Hay w.hay at ucl.ac.uk
Wed May 18 12:48:52 UTC 2011


On 18 May 2011 11:43, Hung-ShengTsao (Lao Tsao) Ph.D. <laotsao at gmail.com> wrote:
> did you check that the userid of  these jobs showup in qstat  -s rp in qw
> may be after you rescured form Eqw userid changed
> JAT
Yes they show up there and the userid is the same as its always been.
>
>
> On 5/18/2011 5:40 AM, William Hay wrote:
>> On 18 May 2011 10:08, Reuti<reuti at staff.uni-marburg.de>  wrote:
>>> Am 18.05.2011 um 10:57 schrieb William Hay:
>>>
>>>> On 17 May 2011 16:09, Reuti<reuti at staff.uni-marburg.de>  wrote:
>>>>> Am 17.05.2011 um 15:51 schrieb William Hay:
>>>>>
>>>>>> One of our users has submitted an array job of about 5000 tasks.
>>>>>>
>>>>>> Some of these tasks are running.  Some are not.  If I run qstat -s
>>>>>> rp
>>>>>> -u<username>  then I see tasks in both  states "r" or state "qw".
>>>>>> Running qstat -s r -u<username>  shows the running tasks.  However
>>>>>> running qstat -s p -u<username>  does not show the tasks in state
>>>>>> qw.
>>>>> Correct, there is the option:
>>>> If this is the "correct behavior" then I have lots of other array jobs
>>>> that are showing up incorrectly.
>>>>> $ qstat -s p -g d -u<username>
>>>>>
>>>>> to list each task on one line on its own.
>>>> I'm not looking for each task on its own.  I do think there should be
>>>> a single line
>>>> combining all the queued tasks when I do qstat -s p.
>>> So you want to have an output like:
>>>
>>> reuti at pc15370:~>  qstat -s p
>>> job-ID  prior   name       user         state submit/start at
>>> queue                          slots ja-task-ID
>>> -----------------------------------------------------------------------------------------------------------------
>>>      467 0.00000 test.sh    reuti        qw    05/18/2011
>>> 11:01:51                                    1 17-10000:1
>>>
>>> ?
>> Yup that's what I expect to see and don't.
>>> It's just the way it's implemented right now. Can you provide examples?
>>
>> Then why do the waiting tasks show up in qstat -s p for other part
>> running/part queued array jobs?  Although these tasks have been
>> rescued from an Eqw
>> state I would still expect them to show up.
>>
>>> -- Reuti
>>>
>>>>   The same as qstat -s rp.
>>>> The same as all the other array jobs that are part queued and part
>>>> tunning.
>>>>
>>>> Essentially I think these two command lines should produce the same
>>>> output and they don't:
>>>> (qstat -s r -u '*'|tail -n +3);(qstat -s p -u '*'|tail -n +3 )
>>>> qstat -s rp -u '*'|tail -n +3
>>>>
>>>> The line aggregating all the queued tasks only shows up when I use a
>>>> single qstat with -s rp.
>>>> Other array jobs do show the queued tasks line when I use qstat -s p
>>>> The tail -n +3 is just to chop off the headers.
>>>>
>>>>> -- Reuti
>>>>>
>>>>>
>>>>>> These waiting tasks had previously been in an error state caused
>>>>>> by a
>>>>>> problem with our cluster file system but I cleared the error once we
>>>>>> had fixed
>>>>>> the problem.  Any idea what is going on and whether these tasks will
>>>>>> eventually run.
>>>>>>
>>>>>> William
>>>>>> _______________________________________________
>>>>>> users mailing list
>>>>>> users at gridengine.org
>>>>>> https://gridengine.org/mailman/listinfo/users
>>>>>
>>>
>> _______________________________________________
>> users mailing list
>> users at gridengine.org
>> https://gridengine.org/mailman/listinfo/users
>




More information about the users mailing list