Choose timezone
Your profile timezone:
PASTE meeting
→
Europe/Zurich
Description
CERN, bat 2 R-030
To join the telephone conference, click here: https://audioconf.cern.ch/call/0122882
To dial in to the conference:
a. Dial +41 22 7676000
b. Enter access code 0122882
Production and PASTE Meeting, 22nd May 2007, CERN
Chair: A.Tsaregorodtsev
Minutes: S. Paterson
CERN: M.Bargiotti, G.Castellani, P.Charpentier, J.Closier, A.Smith, S.Paterson, R.Santinelli
Phone: R.Bernet, R.Graciani, U.Marconi, R.Nandakumar, M.Sapunov, A.Tsaregorodtsev
Apologies: N.Brook
1. Production Status Report
Minutes from the production meeting are available here. Joel reported that no MC simulation was running,
new workflows are being constructed that will use the latest Gauss release ( ACTION: Joel ). The reconstruction
activity was affected last week by VOMS related issues (see the production minutes for more information). This has
now been resolved and the reconstruction is running at 5 Tier-1 sites.
Philippe raised the issue of root protocol at CERN and confirmed that the issues relating to inconsistent TURLs have
been resolved. Stuart agreed to put this in place pending tests at CERN, PIC and CNAF (which previously were
problematic). Rather than lcg-getturls, it was agreed to continue using lcg-gt directly and later make use of the
python binding to GFAL when this becomes available ( ACTION: Stuart ).
The problem with runit creating detached python processes was observed again, this time with the Agent Director.
This confirms Andrei's suspicion that it is related to external commands which are not properly cleaned. The possibility
of using an Agent in the WMS to monitor running processes and clean up as required was discussed.
Joel mentioned that PIC seemed to be back in production after the tape robot malfunctions but no update had been
sent, Ricardo agreed to investigate ( ACTION: Ricardo )
Roberto enquired about the GFAL plugin to root and Philippe gave a summary of the status. Two different backend
SEs (DPM and Castor) have the same interface but are incompatible (libshift). The GFAL plugin for Root is currently
under development. Since the possibility of using a DPM SE at NIKHEF exists, Philippe agreed to investigate the
protocols currently supported by Root and if it is possible to have an application accessing Castor and DPM files in a
consistent way ( ACTION: Philippe ).
Philippe also reported on UI deployment through the AA. Oliver Keeble has deployed native SLC4 32bit versions of the
lcg-utils, GFAL and LFC client (including CMT requirements). This should simplify the certification procedure.
2. Production Operations Issues
There was some discussion of whether LHCb should provide separately packaged versions of Root for user jobs. It was
agreed to pursue making versions of Root available through install_project.py ( ACTION: Hubert ). Stuart agreed to
contact the Ganga developers to discuss the implications for their Root handler. ( ACTION: Stuart )
Marianne gave a summary of the DC06 cleanup status. LFC support have been contacted about the first large
deletion of files, the ticket was opened on the 25th April and Marianne is waiting for a response. The DST removal
from the LFC has started and is expected to take between 1 and 2 days. Site admins can then be contacted and
Marianne is preparing the lists for this. Stuart mentioned that user analysis jobs are suffering from the old SE naming
conventions (e.g. CERN_Castor) where data is failing to enter the Stager. Marianne agreed to start cleaning these
entries. The fixes to incorrect SA paths are now underway, Marianne confirmed that the Castor SURLs are now
corrected.
Roberto agreed to escalate the issue of SRM SAM tests (e.g. srm-get-metadata with timeout) being added to site
critical tests. ( ACTION: Roberto ) It was decided that timing information from the Stager should be viewed as a
separate picture and not included in the SAM tests. In addition to the site metadata information, the time required to
stage files should also be reported ( ACTION: Stuart, Andrew ).
For the LHCb SAM tests, an srm-get-metadata with a timeout of 120s was agreed to be reasonable. This can be part
of the current SAM test and could also include a DaVinci job to test file access from the WNs ( ACTION: Joel )
3. DIRAC3 Tasks
Andrei attached a summary of the DIRAC3 workshop to the agenda (link here). The current plan is to have
another session at the end of June.
In order to test the DIRAC3 deployment mechanism, volhcb03 can be installed from scratch. Andrei confirmed that
more WMS services are in the pipeline.
Andrew raised the issue of pyOpenSSL on Windows, Ying has compiled this in the past and will be contacted for more
information.
Andrei mentioned that Gianluca had prepared a discussion summary of the job prioritization which was also circulated
to Philippe and Nick. Comments will be sent from Andrei as soon as possible. ( ACTION: Andrei )
4. AOB
The issue of a VOMS group for VO Support was discussed and Joel agreed to set this up. GGUS is not VOMS aware,
however, this should simplify the management of the list (ACTION: Joel).
There are minutes attached to this event.
Show them.