Choose timezone
Your profile timezone:
LCG Service Coordination Meeting
Description
Present
Jamie (Chairman), Gavin, Harry (Secretary), Maite, Piotr, Olof, Nicholas, Patricia, Ian, Louis, James, Maria, Thorsten, Judit, Jacek Wojcieszuk
Introduction
• No points
AAS (M.Dimou)
• VOMRS service is stopped due to Oracle cursor bug. Oracle working on it. Workaround would be to use fully qualified object names in application SQL. Maria wanted to reopen the service as the old ldap list is out of date and currently new users cannot be registered. Jamie preferred to wait as he was worried about existing and future corruption. It was agreed to wait for a discussion at the Physics Group Leaders meeting tomorrow.
DMS (G.Mccance,O.Barring for Castor)
• LFC is seeing a ROWID missing problem – happens overnight randomly. May just be related to the Oracle cursors bug. The ATLAS and CMS central LFCs share the same Oracle account and they should now be split.
• The fts and agent software needs upgrading. This will be done progressively.
• The CMS T0-DAQ tests did not reach the 500 MB/s required. Being investigated.
• Restricting the Castor port range for security exposed a bug in the stager api where users could get someone else’s reply. Fix planned for September and workaround is to increase port range.
Experiments
• Alice : (P.Mendez) Important activity is the T0 to T1 data export tests. Currently running at about 50 MB/s to IN2P3. Were using FZK and CNAF but both now have problems. Maite reported FZK have answered the Alice GGUS ticket – to be followed up. SARA have a problem searching their LFC Oracle catalog. James reported this was due to entering null fields in the comments attached to a catalog entry. There is a fix going into PPS now. Till the fix is out Alice should delete and re-enter any entries with empty comments. RAL have now provided 1.6 TB of disk space for Alice.
IS (P.Nyczyk)
• Prototype of new instrumented grid job wrapper is in testing.
• For the ops VO at CERN only jobs through ce106 pass the sft srm test as they find the default SE to be srm.cern.ch while the other ce’s find castorgrid.cern.ch.
WMS (T.Kleinwort)
• There is a new more efficient version of the CE dynamic information provider for LSF. However, more queues are being made which makes the provider more heavy.
• A bug in the accounting script has been fixed but we still only account for one CE due to the fact that APEL uses LSF directory names to distinguish new records and all CE’s see the same LSF directory structure.
Deployment (L.Poncet)
• Have found a RedHat bug that the ID command for local accounts crashes.
• Maite reported that they plan to release gLite 3.0.2 next Monday.
Gmod (J.Novak)
• Nothing special to report.
AOB
• Jamie asked about plans for migrating to the next batch of new hardware, due mid-September. James said most services can be migrated transparently by adding the new hardware to the service before stopping the old.
• Ian thought we should also re-install all grid servers due to their unknown security status. This was agreed to be done in a rolling manner. Harry thought the first two weeks of September would be quiet and was reminded the first one includes Jeunes Genevois.
• Harry mentioned that new hardware would be the time to move required services into the critical power area.
• Jamie announced that CMS are asking CERN to take over some new critical services for them. We will learn more this afternoon.
There are minutes attached to this event.
Show them.