Dear site admins
You will probably be aware of the python update issue as circulated by Rod Walker yesterday. ATLAS has now developed a patch but wish to run further tests before implementing it to production on Monday.
(See also https://twiki.cern.ch/twiki/bin/view/LCG/WLCGDailyMeetingsWeek120618#Friday).
Jeremy
Begin forwarded message:
> From: Tiziana Ferrari <[log in to unmask]>
> Date: 22 June 2012 14:10:03 GMT+01:00
> To: NGI Operations Centre managers <[log in to unmask]>
> Subject: [Noc-managers] Fwd: [Operations] [ EGI BROADCAST ] SL5 update breaks running ATLAS jobs
>
> Dear all
>
> I would like to draw your attention on the following broadcast.
> The update to python and its libs that come with the recent sl5 update cases problems to ATLAS jobs, as explained in the text below.
>
> ATLAS people are currently testing the fix in their code that should solve in the problem.
>
> In the meanwhile I would ask every NGI to make sure that sites supporting ATLAS are informed and follow the recommendation of not upgrading sl5 until further communication.
>
> Thanks Tiziana
>
>
> -------- Original Message --------
> Subject: [Operations] [ EGI BROADCAST ] SL5 update breaks running ATLAS jobs
> Date: Thu, 21 Jun 2012 15:59:41 +0200
> From: EGI BROADCAST <[log in to unmask]>
> To: Site administrators/global_email_cc <[log in to unmask]>
>
> ---------------------------------------------------------------------------------------------------------------
> EGI BROADCAST TOOL : https://operations-portal.egi.eu/broadcast
>
> ---------------------------------------------------------------------------------------------------------------
> Publication from : Rodney Walker <[log in to unmask]>
> Targets : Site administrators/global_email_cc <[log in to unmask]>
> ----------------------------------------------------------------------------------------------------------------
>
>
>
> Hi,
> The recent sl5 update, now in fnal and cern repositories, breaks running
> ATLAS jobs. The update is to python and its libs. The ATLAS jobs import some
> xml module at the end of the job, and this fails since the python binary and
> libs in memory are pre-update ones, and apparently they are not compatible
> with the new libs. The update is to
> python-2.4.3-46.el5_8.2
> due to
> http://rhn.redhat.com/errata/RHSA-2012-0745.html
>
> The error looks like
>
> File \"/usr/lib64/python2.4/xml/parsers/expat.py\", line 4, in ?
> from pyexpat import *
> ImportError: /usr/lib64/python2.4/lib-dynload/pyexpat.so: undefined symbol:
> _Py_HashSecret
>
> It is likely that ATLAS can patch the pilot code to do the import at the
> beginning or the jobs, which is probably best practice anyway(?). In the
> meantime it would be best if sites either do not update, or do so by first
> draining ATLAS jobs. Otherwise, running ATLAS jobs will fail at the end,
> which is very expensive.
>
> Cheers,
> Rod.
>
>
>
>
> ----------------------------------------------------------------------------------------------------------------
> link to this broadcast :
> https://operations-portal.egi.eu/broadcast/archive/id/687
> ----------------------------------------------------------------------------------------------------------------
>
> _______________________________________________
> Operations mailing list
> [log in to unmask]
> https://mailman.egi.eu/mailman/listinfo/operations
>
>
> _______________________________________________
> Noc-managers mailing list
> [log in to unmask]
> https://mailman.egi.eu/mailman/listinfo/noc-managers
|