For the record, the fix for this problem is described in the wiki for
LCG Release Fixes:
http://goc.grid.sinica.edu.tw/gocwiki/LCG_Release_Fixes
(This is indeed a big problem: at IC this morning it was discovered the
site bdii had stopped late yesterday evening, and lhcb had merrily put
over 600 jobs in the queue, on the belief that the farm was still empty.
BDII now updated!)
While the wiki pages are very good and should be kept up to date
(and probably ought to be linked to by the release notes:
http://grid-deployment.web.cern.ch/grid-deployment/cgi-bin/index.cgi?var=releases)
there does need to be some mechanism by which Release Fixes are more
clearly flagged for sysadmins (eg. message header: NEW RELEASE FIX: xyz,
with the body of the message a link to the wiki).
Laurence wrote:
> Hi Rod,
>
> The "sticky bdii" problem is well understood and a fix has been out for
> some time. The bdii update thread will die under certain circumstances.
> The GStat Monitor includes a check on BDIIs with stale data. Any sites
> worried about this having this problem can check the GStat page (which
> they should do regularly anyway!). Maarten Litmaath sent an email to the
> Rollout list informing everyone that there is a new rpm that fixes the
> problem sometime ago. An update will also be in the apt repository.
>
> Laurence
>
>
>
>
>
> Rod Walker wrote:
>
>> Hi,
>> I suspect that sg01-lcg.cr.cnaf.infn.it:2170 is stuck, since it is
>> publishing zero queued jobs despite many of my jobs queuing there
>> (it's lsf so I can`t check for sure - where is qstat?).
>> I`ve mailed the admins directly but if this is due to the bad bdii
>> version distributed with 2.4.0 then I would think many sites still use
>> this. It`s a particularly nasty bug as it can attract thousands of
>> jobs to the affected site. As such I would say it`s a candidate for an
>> "urgent patch", if such a thing exists.
>>
>> Cheers,
>> Rod.
>>
>>
>>
>>
--
=======================================================
Dr O J E Maroney # London Tier 2 Technical Co-ordinator
Tel. (+44)20 759 47802
Imperial College London
High Energy Physics Department
The Blackett Laboratory
Prince Consort Road, London, SW7 2BW
====================================
|