Choose timezone
Your profile timezone:
PASTE meeting
Description
CERN, bat 2 R-030
To join the telephone conference, click here: https://audioconf.cern.ch/call/0165202
To dial in to the conference:
a. Dial +41 22 7676000
b. Enter access code 0165202
Production and PASTE Meeting, 9th September 2008, CERN
Chair: A.Tsaregorodtsev
Minutes: S. Paterson
CERN: J.Closier, Z.Mathe, M.A.S.Miguelez, S.Paterson, V.Romanovsky, R.Santinelli, A.Tsaregorodtsev
Phone: N.Brook, R.Nandakumar, M.Sapunov
Apologies: P.Charpentier, A.C.Smith
0. Ongoing Action List
(07) - 27/11/07 - Zoltan, Elisa - Prototype bookkeeping system, roadmap for production
(09) - 04/12/07 - Roberto, Marcos - Develop a general service monitoring framework leading to a virtual control room
(57) - 24/06/08 - Vladimir - Automatic publishing of BDII information in the CS using GOCDB site names
(63) - 22/07/08 - Roberto,Joel - Develop DIRAC3 script for adding new users to the VO + CS + LFC registration
(70) - 12/08/08 - Joel - DIRAC3 installation in the LHCb release area at CERN
(71) - 19/08/08 - Andrei,Joel - Add comment field to mask operations and formulate procedure
(72) - 19/08/08 - Roberto,Vladimir - Check BDII / GOCDB inconsistencies in site published informations
(79) - 26/08/08 - Andrei, Vladimir - Update CS to use new host aliases
(80) - 26/08/08 - Stuart - Update DIRAC2 Stager for SRMv1->v2 migration
(82) - 01/09/08 - Ricardo/Adri - to provide information from accounting to allow GridMap for LHCb VO T1 activity
(83) - 01/09/08 - Stuart - To perform glexec tests
(85) - 09/09/08 - Zoltan - Define authorization rules for BK interface methods
(86) - 09/09/08 - Andrei - Document the job finalization output data upload policy
(87) - 09/09/08 - Stuart - Proceed with testing xrootd at NIKHEF
(88) - 09/09/08 - Matvey - Provide improved RequestMonitoring interface in lhcbweb
(89) - 09/09/08 - Stuart - Add production LFNs to production job JDL on client side
(90) - 09/09/08 - Joel - Use new ProductionOutputData JDL parameter in modules instead of creating LFNs
(91) - 09/09/08 - Andrei - Create LHCbSystem agent to populate logSE with standard output / error
Completed Actions:
Done - (73) - 19/08/08 - Andrei - Resolve BK catalogue status problems in the Replica Manager
Done - (76) - 26/08/08 - Hubert - Modifications to install_project to support SGM pool accounts
Done - (77) - 26/08/08 - Philippe - Prepare summary of SGM pool account saga for MB
Done - (78) - 26/08/08 - Zoltan - Update BK twiki page with current machines, services & setups
Done - (84) - 01/09/08 - Raja - To check PIC VO box for problem with re-sending BK info
1. Action Updates
- Action(07) - Prototype Bookkeeping
Zoltan is working on the GUI for the Bookkeeping and should have a prototype available by the end of the week. The
processing pass implementation has also started. There was some discussion regarding the access control to the new
BK instance. Andrei suggested creating a new DIRAC group (not VOMS) to allow secure write operations to the BK.
Zoltan was asked to go through the methods in the BK interface and propose the restrictions for security. ( ACTION:
Zoltan )
- Action(09) and Action(65) - Virtual Control Room
Roberto reported some interactions with the developers of the new SAM webpage and intends to convey the DIRAC
site naming convention to them for introduction in the page. Defining the availability of the sites based on critical
services for LHCb in a way that automatically collates results over time is possible with the new interface. Service
monitoring tests are under development for checking access to the CondDB instances. DIRAC3 Tier-1 VO-box
monitoring is being added as an SLS sensor. There was some discussion of making use of the error messages from
the Stager in the Logging server. Action(65) is merged with (09).
- Action(57) - Automatic publishing of BDII information in the CS
Some scripts are being prepared to check BDII <-> CS information, this is ongoing, the CS information is being
updated. An agent is also in the pipeline that will automatically publish information from the BDII to the CS.
- Action(63) - DIRAC3 scripts for adding new users to the VO + CS + LFC registration
DIRAC3 support should be confirmed and script should be added to CVS.
- Action(69) - Single user account for SGM user
After discussion at the MB level there are some security concerns with static accounts so it was agreed to continue to
put protection in place on our side (install_project) for the pool accounts. ( ACTION: Hubert ) Philippe will prepare a
summary of the difficulties for the MB ( ACTION: Philippe ).
- Action(70) - DIRAC3 installation in the LHCb release area at CERN
Joel has prepared a DIRACEnv type script to set the environment. Reminder: last week it was agreed to enable write
permission to the DIRAC configuration file for the DIRAC developers in case of emergencies. Some documentation
about the final procedure should be prepared.
- Action(71) - Comments for site mask
Andrei will add this to the DB and Stuart will update the client side.
- Action(72) - GOCDB / BDII Inconsistencies
Ongoing. It is important to make a first pass through this information and ensure that the list of available LHCb sites
is automatically kept up to date.
- Action(73) - BK catalogue status problem in Replica Manager
Andrei agreed to document the policies for output data policy. ( ACTION: Andrei )
- Action(79) - Ongoing
- Action(84) - PIC VO-box request forwarding
Problematic requests were found to be old and were not complying to the schema.
2. Production Status Report
Stuart reported that recent stalled jobs could be due to the CS hanging. The functionality to specify banned sites is
not available and was lost from the old system, this is preventing ramping up with background MC activity and should
be investigated after the meeting. The DC06 stripping is not smoothly running due to crashes in DaVinci (~18%), a
patch for this is being pursued by Philippe. CCRC Stripping is pending the preparation of options that will not crash
the workflow. The MC stripping workflow is yet to be commissioned, Joel is working on this. The recently developed
file rescheduling limit should be deployed in production. The regular failures at NIKHEF triggered some discussions
regarding how to proceed with frequent instabilities in data access. It was agreed to test with xrootd and the CS will
be updated to reflect this. ( ACTION: Stuart )
Several topics were discussed during the actions list that should be raised with the application providers:
- Tests provided as part of software distribution (as discussed during the SAM meeting)
- Analysis of log files provided by application managers, how to better formalize and anticipate errors on our side
- Exit codes, if there is a way to formalize the exit codes of Gaudi Applications this would be extremely beneficial for our shifters.
3. DIRAC3 Tasks
- Requests monitoring
Matvey is starting to work on an improved interface for monitoring requests starting from the jobs. Also work is
progressing on a Request Monitoring page (allowing selections etc. in a similar manner to the Job Monitoring page). (
ACTION: Matvey )
- Output data LFN creation before the job submission
It was agreed to use existing methods to dynamically create the output LFNs in the API on the client side prior to
submission, at least so they appear in the JDL. ( ACTION: Stuart ) The modules must eventually be updated to cope
with the full LFN in the JDL. ( ACTION: Joel )
- Items in the LogSE
Andrei will create an agent in the LHCbSystem that will asynchronously populate the LogSE with workload
management information e.g. standard output, parameters, logging info. ( ACTION: Andrei )
4. AOB
None
There are minutes attached to this event.
Show them.