Production and PASTE Meeting, 15th May 2007, CERN
 
 Chair: A.Tsaregorodtsev
 Minutes: S. Paterson
 
 CERN:   M.Bargiotti, A.Casajus, G.Castellani, J.Closier, M.Dimou, R.Graciani, A.Smith, S.Paterson, R.Santinelli, A.Tsaregorodtsev
 
 Phone:  R.Bernet, N.Brook, U.Marconi, R.Nandakumar
  
 Apologies: P.Charpentier

1. LHCb VO User Support ( M.Dimou )

Maria wished to clarify the LHCb procedure for the submission of GGUS tickets.  LHCb users currently contact internal 
mailing lists for Grid support and experts then submit GGUS tickets if required.  It was requested that older tickets 
should be cleaned up but this was not possible due to not having GGUS admin status.  To facilitate this it was agreed 
to set up a VOMS group for LHCb VO support.  Maria also highlighted the VO support pages at <a href="http://indico.cern.ch/materialDisplay.py?contribId=4&amp;materialId=0&amp;confId=14711">this link</a>.

2. Production Status Report  

The production meeting was cancelled this week due to the DIRAC3 discussions.  Minutes from the last production 
meeting are available <a href="https://twiki.cern.ch/twiki/bin/view/LHCb/ProductionInProgress">here</a>.  Joel reported that no MC simulation was running. Nick mentioned that two new 
sites were available (CNAF and Russian based Tier-2's) and the usage of these new resources should be 
maximised.  The new production requests will be submitted pending the new version of Gauss (2.5M events are to be 
produced).

A summary of the reconstruction status by site is below:
- CERN &amp; IN2P3 OK
- PIC some hardware problems (tape robot)
- CNAF no jobs eligible 
- GRIDKA &amp; RAL had many stalled jobs
- NIKHEF a GGUS ticket will be submitted for file upload problems to the NIKHEF-tape SE ( ACTION: Joel ).

The DC06 cleanup is ongoing, the Tier-1 disks are still to be tidied but Marianne is preparing the lists for this ( ACTION: 
Marianne ).  Joel reported that Gennady had provided a script to target the destination site for reconstruction jobs 
which is now in use.


3. Production Operations Issues 

The issue of SRM stability at the Tier-1s was mentioned and Nick agreed to escalate this on a site by site basis after 
some definite information had been gathered.  Stuart reported that the StagerAgent is now populating the StagerDB 
with all site metadata timing information and this should be represented graphically somehow ( ACTION: Andrew, 
Stuart ). Roberto agreed to report to add a metadata request to the SRM critical tests and flag with an alarm.  In this 
way, sites would not be declared 'up' unless their SRM was responsive.  The possibility to include VO critical tests in 
the evaluation of site availability should also be discussed at the operations meeting ( ACTION: Roberto ).

The issue of some user jobs incorrectly staging data at one site and running at another site is fixed at the time of 
writing these minutes.  Now the Data Optimizer will force a single destination site for all jobs ensuring data cannot be 
staged twice unnecessarily.  There was also some discussion of the recent User system Stager activity.  It was agreed 
that jobs accessing files on tape should be restricted somehow.  Stuart will work on this and draft an email to explain 
the policy. ( ACTION: Stuart )


4. Definition of the LHCb VO Policies

A further discussion on the LHCb VOMS policy was started, it was concluded that all LHCb users will be mapped to the 
LHCb group by default and the VO will retain specialized groups such as production managers.   There was a 
comprehensive discussion of the job prioritization strategies in DIRAC and Gianluca agreed to make a summary of this 
( ACTION: Gianluca ).


5. DIRAC3 Tasks

The DIRAC3 Tasks discussion at the PASTE meeting was postponed due to the DIRAC3 workshop (being held 
concurrently).


6. AOB

None