[GE globus] [GE users] Problem sending jobs with globusrun-ws: Current job state: Unsubmitted

Esteban Freire Garcia esfreire at cesga.es
Wed Dec 5 17:04:40 GMT 2007


    [ The following text is in the "ISO-8859-1" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

Hi Jeff,

Thanks for you answer. Ok, I have the file 
$SGE_ROOT/default/common/reporting.
No,  we are not using ARCO.  The only  thing  that  I think  maybe is 
happening, it's  that globus cannot read this file, but I tested to read 
this file as user "globus" and as user who sent the job, and I could 
read without any problem. Is there any place where I can indicate to 
globus to read this file?

I put below the output to the file 'reporting', after send a job using 
globus and send a job with qsub.

tail -f $SGE_ROOT/default/common/reporting
------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
1196873426:new_job:1196873426:1417619:-1:NONE:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:1024
1196873426:job_log:1196873426:pending:1417619:-1:NONE::cyteduser:svgd.cesga.es:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:new 
job
1196873437:job_log:1196873437:sent:1417619:0:NONE:t:master:svgd.cesga.es:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:sent 
to execd
1196873437:host_consumable:compute-1-12.local:1196873437:X:num_proc=1.000000=1.000000,s_vmem=524288000.000000=1.300G
1196873437:queue_consumable:pro_cytedgrid:compute-1-12.local:1196873437::num_proc=1.000000=1.000000,s_vmem=524288000.000000=1.000G,slots=1.000000=1.000000
1196873437:job_log:1196873437:delivered:1417619:0:NONE:r:master:svgd.cesga.es:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:job 
received by execd
1196873437:acct:pro_cytedgrid:compute-1-12.local:cesga:cyteduser:sge_job_script.28406:1417619:sge:0:1196873426:1196873390:1196873391:0:0:1:0:0:0.000000:0:0:0:0:5330:0:0:0.000000:0:0:0:0:258:45:NONE:defaultdepartment:NONE:1:0:0.000000:0.000000:0.000000:-U 
pro_cytedgrid -l 
arch=i386,h_fsize=1G,h_stack=16M,num_proc=1,s_rt=3600,s_vmem=500M:0.000000:NONE:0.000000
1196873437:job_log:1196873437:finished:1417619:0:NONE:r:execution 
daemon:compute-1-12.local:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:job 
exited
1196873437:job_log:1196873437:finished:1417619:0:NONE:r:master:svgd.cesga.es:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:job 
waits for schedds deletion
1196873437:host_consumable:compute-1-12.local:1196873437:X:num_proc=0.000000=1.000000,s_vmem=0.000000=1.300G
1196873437:queue_consumable:pro_cytedgrid:compute-1-12.local:1196873437::num_proc=0.000000=1.000000,s_vmem=0.000000=1.000G,slots=0.000000=1.000000
1196873448:job_log:1196873448:deleted:1417619:0:NONE:T:scheduler:svgd.cesga.es:0:1024:1196873426:sge_job_script.28406:cyteduser:cesga::defaultdepartment:sge:job 
deleted by schedd

1196873742:new_job:1196873742:1417621:-1:NONE:test.sh:esfreire:cesga::defaultdepartment:sge:1024
1196873742:job_log:1196873742:pending:1417621:-1:NONE::esfreire:svgd.cesga.es:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:new 
job
1196873753:job_log:1196873753:sent:1417621:0:NONE:t:master:svgd.cesga.es:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:sent 
to execd
1196873753:host_consumable:compute-1-14.local:1196873753:X:num_proc=1.000000=1.000000,s_vmem=1073741824.000000=1.300G
1196873753:queue_consumable:GRID:compute-1-14.local:1196873753::num_proc=1.000000=1.000000,s_vmem=1073741824.000000=2.000G,slots=1.000000=1.000000
1196873753:job_log:1196873753:delivered:1417621:0:NONE:r:master:svgd.cesga.es:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:job 
received by execd
1196873754:acct:GRID:compute-1-14.local:cesga:esfreire:test.sh:1417621:sge:0:1196873742:1196873658:1196873658:0:0:0:0:0:0.000000:0:0:0:0:689:0:0:0.000000:0:0:0:0:202:2:NONE:defaultdepartment:NONE:1:0:0.000000:0.000000:0.000000:-U 
paralelo-gigabit,jmourino,esfreire,blades_dell -l 
arch=i386,h_fsize=1G,h_stack=16M,network=gigabit,num_proc=1,s_rt=3600,s_vmem=1G:0.000000:NONE:0.000000
1196873754:job_log:1196873754:finished:1417621:0:NONE:r:execution 
daemon:compute-1-14.local:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:job 
exited
1196873754:job_log:1196873754:finished:1417621:0:NONE:r:master:svgd.cesga.es:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:job 
waits for schedds deletion
1196873754:host_consumable:compute-1-14.local:1196873754:X:num_proc=0.000000=1.000000,s_vmem=0.000000=1.300G
1196873754:queue_consumable:GRID:compute-1-14.local:1196873754::num_proc=0.000000=1.000000,s_vmem=0.000000=2.000G,slots=0.000000=1.000000
1196873764:job_log:1196873764:deleted:1417621:0:NONE:T:scheduler:svgd.cesga.es:0:1024:1196873742:test.sh:esfreire:cesga::defaultdepartment:sge:job 
deleted by schedd

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

On the other hand, we are using SGE 6.0u6

Thanks,
Esteban

Jeff Porter escribió:
> Hi Esteban,
>
> the logfile noted in the docs is the 'reporting' file: $SGE_ROOT/default/common/reporting.  The gt4 c-code reads that file for jobs state information instead of the calling qsub from sge.pm as is done for gt2.  I wouldn't spend much time on the sge.pm file as its use in gt4 is essentially just for submission.  And the patch you say you applied before is directed at fixing gt2-specific details that break gt4 submissions.
>
> One other issue is if you are running ARCO you may have this problem. I understand the dbwriter code deletes the reporting file with each read as its mechanism for checkpointing. Thus gt4 will never see the change in state through this file. 
>
> Thanks, Jeff
>
>   
>> Hi Melvin,
>>
>> Thanks for you answer. I have "reporting=true" but I had 
>> "joblog=false", 
>> at these moments I already have changed this and now I have 
>> "joblog=true", after this, I have reinstalled the packages of 
>> "London 
>> e-Science Centre" y I have ran the gpt-postinstall again, but 
>> unfortunately,  it keeps without pass of the state "unsubmitted":
>> --------------------------------------------------------------------
>> --------------------------------------------------------------------
>> ----------------------------------------------
>> [esfreire at svgd ~]$ globusrun-ws -submit -pft -T 10000 -s -S -
>> factory 
>> svgd.cesga.es -Ft SGE -c /bin/hostname
>> Delegating user credentials...Done.
>> Submitting job...Done.
>> Job ID: uuid:1fe5c0d2-a31d-11dc-a78b-000423ac0723
>> Termination time: 12/06/2007 10:30 GMT
>> Current job state: Unsubmitted
>> --------------------------------------------------------------------
>> --------------------------------------------------------------------
>> ----------------------------------------------
>> One thing that I don't understand is that in the link to "London 
>> e-Science Centre" say, "Your SGE installation must also be 
>> configured 
>> with support for the reporting logfile enabled, and that logfile 
>> must be 
>> accessible from the server on which you are installing GT4", I 
>> don't 
>> know which is this "logfile"? I suppose that is 
>> "$SGE_ROOT/default/spool/qmaster/messages"
>>
>> Other thing that it's indicating that something go wrong,  I think 
>> is 
>> that the job only run about 1 second.
>>
>> --------------------------------------------------------------------
>> --------------------------------------------------------------------
>> ----------------------------------------------
>> [globus at svgd JobManager]$ qacct -j 1417415
>> ==============================================================
>> qname        pro_cytedgrid      
>> hostname     compute-1-12.local 
>> group        cesga              
>> owner        cyteduser          
>> project      NONE               
>> department   defaultdepartment  
>> jobname      sge_job_script.1784
>> jobnumber    1417415            
>> taskid       undefined
>> account      sge                
>> priority     0                  
>> qsub_time    Wed Dec  5 11:18:41 2007
>> start_time   Wed Dec  5 11:18:05 2007
>> end_time     Wed Dec  5 11:18:06 2007
>> granted_pe   NONE               
>> slots        1                  
>> failed       0   
>> exit_status  0                  
>> ru_wallclock 1           
>> ru_utime     0           
>> ru_stime     0           
>> ru_maxrss    0                  
>> ru_ixrss     0                  
>> ru_ismrss    0                  
>> ru_idrss     0                  
>> ru_isrss     0                  
>> ru_minflt    5328               
>> ru_majflt    0                  
>> ru_nswap     0                  
>> ru_inblock   0                  
>> ru_oublock   0                  
>> ru_msgsnd    0                  
>> ru_msgrcv    0                  
>> ru_nsignals  0                  
>> ru_nvcsw     262                
>> ru_nivcsw    44                 
>> cpu          0           
>> mem          0.000            
>> io           0.000            
>> iow          0.000            
>> maxvmem      0.000
>> --------------------------------------------------------------------
>> --------------------------------------------------------------------
>> ----------------------------------------------
>> I don't know what else change.
>>
>>
>> Thank you very much,
>> Esteban
>>
>> Melvin Koh escribió:
>>     
>>> Have you enabled "reporting=true" and "joblog=true" in "qconf -
>>>       
>> mconf"?>
>>     
>>> On Fri, 23 Nov 2007, Esteban Freire Garcia wrote:
>>>
>>>   
>>>       
>>>> Hi,
>>>>
>>>> First of all, thanks for answer me. We installed the patch 
>>>>         
>> yesterday, 
>>     
>>>> unfortunately, we continue with the same problem, we will try 
>>>>         
>> look the 
>>     
>>>> jobmanager, because I think for some reason, the 
>>>>         
>> jobmanager(sge.pm) is 
>>     
>>>> not seeing the status for the job correctly, and it doesn't know 
>>>>         
>> when 
>>     
>>>> the job have finished.
>>>>
>>>> -----------------------------------------------------------------
>>>>         
>> -------------------------------------------------------------------
>>     
>>>> [esfreire at svgd ~]$  globusrun-ws -submit -pft -s -S -F  
>>>>
>>>>         
>> https://svgd.cesga.es:8443/wsrf/services/ManagedJobFactoryService -
>> Ft 
>>     
>>>> SGE -c /bin/hostname
>>>> Delegating user credentials...Done.
>>>> Submitting job...Done.
>>>> Job ID: uuid:580a49d2-9923-11dc-9646-000423ac0723
>>>> Termination time: 11/23/2007 17:49 GMT
>>>> Current job state: Unsubmitted
>>>>
>>>> globusrun-ws: Error querying job state
>>>> -----------------------------------------------------------------
>>>>         
>> -------------------------------------------------------------------
>>     
>>>> Thank you very much,
>>>> Esteban
>>>>
>>>> Otheus (aka Timothy J. Shelling) escribi?:
>>>> Hi,
>>>>     
>>>>         
>>>>> On Nov 20, 2007 9:13 AM, Esteban Freire Garcia 
>>>>>           
>> <esfreire at cesga.es 
>>     
>>>>> <mailto:esfreire at cesga.es>> wrote:
>>>>>
>>>>>     Hi,
>>>>>
>>>>>     We have installed 'gt4.0.5-x86_64_rhas_4-installer' on "Red 
>>>>>           
>> Hat>>>     Enterprise Linux ES release 4 (Nahant)".  ...
>>     
>>>>>     Now, we are trying to integrate Globus with SGE 6.0u6, 
>>>>>
>>>>>
>>>>> I don't know if this will help or not. I had to patch gt4.0.2 
>>>>>           
>> to work 
>>     
>>>>> with SGE 6.0u4 as follows:
>>>>>
>>>>>       
>>>>>           
>>>   
>>>       
>>
>>     
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: globus-unsubscribe at gridengine.sunsource.net
> For additional commands, e-mail: globus-help at gridengine.sunsource.net
>
>
>   





More information about the gridengine-users mailing list