Production and PASTE Meeting, 17th June 2008, CERN
Chair: A.Tsaregorodtsev
Minutes: A.Tsaregorodtsev
CERN: V.Romanovsky, J.Closier, P.Charpentier
Phone: R.Bernet, R.Nandakumar, A.Tsaregorodtsev
0. Ongoing Action List
(07) - 27/11/07 - Zoltan, Elisa - Prototype bookkeeping system on volhcb07
(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
(59) - 22/07/08 - Ricardo - Add ldap python modules to DIRAC python distribution
(62) - 22/07/08 - Roberto - Check whether glite 3.1 comes with requested features/patches & coordinate upgrades
(63) - 22/07/08 - Roberto,Joel - Develop DIRAC3 script for adding new users to the VO + CS + LFC registration
(64) - 29/07/08 - Philippe, Andrew - Prepare SRMv1 retirement migration plan
(65) - 05/08/08 - Roberto - Define our SAM tests in the dashboard
(66) - 05/08/08 - Vladimir - Investigate interfaces to GOC DB for site downtime (c.f. ATLAS Dashboard)
(67) - 05/08/08 - Andrei - Test failover request forwarding mechanism
(68) - 05/08/08 - Zoltan - Migrate new BK from volhcb07 -> volhcb04
(69) - 12/08/08 - Joel - All the sites to provide single user account for the sgm user
(70) - 12/08/08 - Joel - DIRAC3 installation in the LHCb release area at CERN
(00) - 05/08/08 - DIRAC developers - DIRAC3 (ongoing)
Update on ongoing actions:
(57,59) - Vladimir will check if the ldap python module shipped together with the LCG middleware
can be used. This would avoid to ship the ldap module with DIRAC.
(63) - Joel will join the work on DIRAC3 user management scripts
(66) - Vladimir has implemented a Google calendar with the information about the site
scheduled downtimes. Matvey has provided the link from the Web portal (Web menu)
(67) - several fixes were made, more testing necessary
1. CCRC low rate production
Production has started over the weekend as described in the eLog record:
http://lblogbook.cern.ch/Operations/258
More attention of the shifters/operators is necessary to follow the results
of the jobs. Joel will restart doing that.
Philippe pointed out that many monitoring views used to be in DIRAC2 are no
more available in DIRAC3. The information is there but pages with predefined
views should be elaborated.
There were massive transfers into the CERN-RAW SE over the weekend stopped at ~22:00
Sunday 10th August. Was not clear the origine of these transfers. Radu was requested
to comment.
2. SW shared areas
There are still problems with applications software areas on sites with pool sgm accounts.
There is no a clear and definite solution to define correct file access permissions. It
was decided to systematically contact all the sites still maintaining pool sgm accounts
to convince them to change this to a single sgm account ( ACTION: Joel )
3. DIRAC2 production
Philippe pointed out that many fixes were made recently to the logic of the Workflow
modules. These fixes should be forwarded to the DIRAC3 workflow modules where appropriate.
4. volhcb09 flickering behaviour
The machine rebooted several times last day. IT support will update the Linux kernel
as this problem is known.
4. Aliases for DIRAC service hosts
An updated list of aliases names is atached to the agenda
5. DIRAC3 tasks
It was decided to install DIRAC3 v0r4p0 in the LHCb release area. Joel will provide
DIRAC environment setting scripts conforming with the LHCb rules (ACTION).
Ricardo reported on the new Task Queue implementation in the DIRAC WMS system.
This is necessary to increase matching efficiency and to allow flexible use of
job prorities, especially in the mixed production/user environment. Andrei stressed
that the current TQ implementation should be still usable to have a fallback solution
before the new system proves its intended goals.
Andrei reported that the new API to manage Production Requests is available. Since
Matvey is on vacations for 3 weeks, the web interface will be delayed. Andrei will
provide the command line interface in the meanwhile
6. AOB
Joel raised a question of the possibility to use ORACLE backend for at least a subset
of DIRAC databases, in particular databases hosted at PIC ( accounting ). Andrei
commented that in order to do that and standard DIRAC Oracle python binding should
be implemented. Zoltan's work with Oracle access from the Python environment suggests
that this is feasible. This is not urgent but should be kept as an opportunity.
The strategy of the MySQL backups was discussed. It was agreed that a possible solution
can be setting up a slave MySQL instance automatically synchronized with the main
database running as a master. The backup of the MySQL databases can be then done on the
slave instance without interference with production activities.
There are minutes attached to this event.
Show them.