PASTE meeting

Tuesday 25 Sept 2007, 10:00 → 12:00 Europe/Zurich

Andrei Tsaregorodtsev

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, 25th September 2007, CERN Chair: A.Tsaregorodtsev Minutes: S. Paterson CERN: M.Bargiotti, G.Castellani, P.Charpentier, J.Closier, E.Lanciotti, S.Paterson, R.Santinelli Phone: R.Bernet, A.Carbone, M.A.S.Miguelez, M.Sapunov, A.Tsaregorodtsev Apologies: N.Brook, R.Nandakumar 1. Production Status Report Minutes from the production meeting are available here. Joel summarized the production status: - Simulation Nothing to report, Roland was asked to submit remaining prod requests from Olivier - Reconstruction GridKa No staging requests issued (1 file successfully staged in 24hrs) IN2P3 Several dozen Lyon b-inclusive no progress with staging, maximum bias went through OK since files on disk CERN / CNAF Several jobs have exceeded max staging retry, files either of zero size, not existing or not migrated seem to be the cause. These jobs should be investigated using Joel's script to extract problematic replicas. This should also become a daily operation and Joel agreed to add this script to CVS. ( ACTION: Marianne, Joel ) NIKHEF An incorrect account mapping problem lhcbpr -> lhcbprd is now solved after interactions yesterday with Jeff. The remaining reconstruction jobs are running smoothly. RAL HOME directory problem, GGUS ticket 27224. The apparent cause of the problem is a ROOT call 'getpwuid(getuid())' that doesn't work in the WN environment. Philippe agreed to contact the ROOT developers about this issue since the call may be redundant. ( ACTION: Philippe ) Staging is slow at RAL, Joel will contact Raja to investigate. ( ACTION: Joel ) PIC Progress is slow but steady after the SLC4 migration. Around ~200 jobs / day were executed yesterday. - Other items Only 15 out of 25 M maximum bias events have been reconstructed. A few productions were missing, Joel has added them to the Processing DB but the discrepancy still exists. Directories were added by hand and no new entries found in the LFC so Processing DB appears consistent. This should be understood. ( ACTION: Joel ) Philippe requested that the storage usage script should be upgraded to report the number of files as well as the total space consumed. Joel and Stuart discovered that SLC4 versions of the software have been installed at sites and SLC3 versions are installing local to the jobs running there. Furthermore, recent modifications to use the SLC4 compiled software have revealed that the compatibility libraries are not in place for platforms other than SLC3. To resolve the situation, the Software Distribution module should be equipped to install multiple architectures and the compatibility libraries should be investigated with Hubert. ( ACTION: Joel, Stuart ) In the future, guessPlatform can return the full list of available platforms such that Agents can present the possibilities to the Matcher. Any jobs with compiled libraries will require to specify the desired platform and this can be taken into account during Matching. Elisa and Roberto raised the issue of PPS tests with Storm. Since the original tests have been performed using local scripts at CNAF, the original suggestion was to duplicate the tests via DIRAC. After some discussions it was agreed that these tests should be postponed since the results will not change significantly by adding the DIRAC / Grid machinery. Roberto agreed to circulate the link to the forthcoming paper on the Storm testing. 2. Production Operations Issues - CNAF CE configuration Angelo proposed the following configuration: LCG.CNAF-T2.it: ce_t2_01.cr.cnaf.infn.it ce_t2_02.cr.cnaf.infn.it and LCG.CNAF-T1.it: ce_t1_01.cr.cnaf.infn.it ce_t1_02.cr.cnaf.infn.it DIRAC needs to decipher from the WN whether it is on CNAF T1 or CNAF T2. From the site gatekeeper endpoint information DIRAC can determine the site according to LHCb conventions. After some discussion it was agreed that a distinct CE for the Tier-2 that only sees the Tier-2 queue would be the preferred solution for LHCb. ( ACTION: Angelo ) - NIKHEF grid-monitor jobs Roberto explained that the LCG RB keeps submitting grid-monitoring jobs as long as jobs are in the CE. Since a continuous stream of pilots were submitted to NIKHEF, eventually the CE became flooded. Around 50 jobs were in the waiting state for NIKHEF continuously submitting pilots which Aborted. Since there were so few jobs, they went unnoticed during CHEP. Philippe commented on Roland's mail regarding new event types. Joel confirmed one of the workflows is for a new event type and a script must be run to update the bookkeeping. ( ACTION: Marianne, Joel ) Roberto reported CERN gsiklog testing progress but unfortunately a critical VOMS mapping problem has been identified. It is difficult to proceed but the behaviour on the remaining Tier-1 sites should be investigated. Through VO ID card on the CIC portal, desired information can be presented from the VO for the site admins. YAIM then uses these configurations but other mechanisms e.g. QUATTOR do not use same procedure. Due to the lack of consistency in configuration mechanisms the propagation of LHCb policies is non-trivial. Andrei remarked this is an important issue which should be escalated. ( ACTION: Andrei ) Roberto mentioned that the ATLAS approach is to use only the default VOMS mapping everywhere such that nothing is required from the sites. Whether or not the default mapping can satisfy all our use cases should be investigated with a very high priority. ( ACTION: Roberto, Joel ) Current storage usage should be investigated and a cleanup should be performed at the Tier-1's. A rough estimate can only be made since the retention for each signal is not known. Assuming ~30% average around 25TB will be required for one copy of the forthcoming data. Storage needs for the rest of the year should be calculated and requests to the sites should be renewed. ( ACTION: Nick, Philippe ) 3. DIRAC3 Tasks Ongoing after CHEP and vacations. 4. AOB The CHEP contributions page is available here. The presenters were asked to prepare their papers as quickly as possible. It was suggeted to start the meeting earlier to avoid frequently late finishes. Andrei agreed to propose this to the PASTE list. ( ACTION: Andrei )

10:00 → 10:15 Production status report

- Report from the Production meeting:

10:15 → 10:45 Production operations issues

- CNAF CE configuration - NIKHEF grid-monitor jobs

11:05 → 11:35 DIRAC3 tasks

11:35 → 11:45 AOB