Hi,
why are the SAM tests still using the RBs for monitoring?
If I use a WMS there is no problem any more.
Regards
Klaere
Markus Schulz schrieb:
> Hi Maarten,
> I don't think that we should investigate RB problems any longer.
> There is a fix for all RB problems and this is the upgrade to the WMS.
>
> Maybe Oliver can push the message that the LCG-RB is no longer supported and should be phased out.
> When we last asked there were only some old Dirac instances that depended on the RB and it was promised to
> be ending very soon (whenever this will be...)
>
> markus
>
> -----Original Message-----
> From: LHC Computer Grid - Rollout on behalf of Maarten Litmaath
> Sent: Fri 9/19/2008 3:50 PM
> To: [log in to unmask]
> Subject: Re: [LCG-ROLLOUT] Jobs ok via WMS, but fail via RB
>
> Hallo Kläre,
>
>>> the torque server is now on another machine, it also serves queues for
>>> other clusters.
>>
>> Apparently stage-out does not work with your new setup.
>>
>> The job wrapper has "#PBS -o" and "#PBS -e" directives which I see
>> being ignored by your installation.
>>
>> Please look at this Wiki page:
>>
>> http://goc.grid.sinica.edu.tw/gocwiki/Unspecified_gridmanager_error
>>
>> When I run the script shown on that page, the job ends up in 'W' state
>> after
>> having been in 'R' state for a second or so:
>>
>> ---------------------------------------------------------------------------------------
>>
>>
>> Req'd Req'd Elap
>> Job ID Username Queue Jobname SessID NDS TSK
>> Memory Time S Time
>> -------------------- -------- -------- ---------- ------ ----- ---
>> ------ ----- - -----
>> 41975.tonia egops015 egops STDIN -- 1 1
>> -- 12:00 W --
>> ---------------------------------------------------------------------------------------
>>
>>
>> This means also the stage-in does not work.
>> You need to fix both.
>
> It seems stage-in only fails for absolute paths on the WN, as used by
> the test script. The lcgpbs job manager uses a relative path, which
> explains why RB/WMS jobs can start.
>
> Still, on other PBS/Torque installations the test script works...
--
Klaere Cassirer
Fraunhofer-Institute for Algorithms and Scientific Computing (SCAI)
Department of Simulation Engineering
Schloss Birlinghoven
D-53754 Sankt Augustin
Tel: +49 - 2241 - 14 - 2758
Fax: +49 - 2241 - 14 - 42758
E-mail: [log in to unmask]
Internet: http://www.scai.fraunhofer.de
|