Deployment team

Tuesday 14 Aug 2007, 11:00 → 12:00 Europe/Zurich

EVO - GridPP Deployment team meeting

Jeremy Coles

- This is the weekly DTEAM meeting - The intention is to run the meeting in EVO: http://evo.caltech.edu/evoGate/. Join the meeting in the GridPP Community area

Minutes 070814_DTEAM_minutes.txt

11:00 → 11:20 Experiment problems/issues

Review of weekly issues by experiment/VO - LHCb 1. Glasgow has been banned today for possible LHCb software installation problems (being looked into by us). 2. Many UK sites passing SAM tests. Some sites have problems - stale NFS handles (Brunel) which we are in contact with the sites. 3. Some problem with a LHCb software tarball which is causing some SAM tests to fail. Being looked into by us. - CMS - ATLAS * From ops meeting: Tier-2s may be involved in the ATLAS M4 cosmic ray run (scheduled from 23 August to 3 September - See https://twiki.cern.ch/twiki/bin/view/Atlas/ComputingOperatonsPlanningM4) on request. Does this tally with the ATLAS view in the UK? * SL tests showing several problems today. IC-HEP and SHEF-HEP in particular do not look well! * RB tests all look bad * SAM shows problems for QMUL, UCL-CENTRAL, SHEF-HEP and OX-HEP. Still lots of warnings (CA). - Other * Several sites have seen CE overload problems caused by biomed and now pheno. Is the recommendation for sites to impose total queued jobs limits on each queue? - What is the status of regional VOs - g-Eclipse (what else needs to be provided?) - minos and supernemo

11:20 → 11:35 ROC update

ROC manager update ************************* The next ROC manager's meeting is on 21st August: http://indico.cern.ch/conferenceDisplay.py?confId=19703. There is still a call for input for the ROC voting on priority of issues to be addressed within EGEE. Fair response to questions on using DAG. Generally not very favoured in UK. A number of issues have been highlighted covering trust, service levels, conflict with OS rpms, supported installation methods (for recipe to avoid OS conflicts) and testing. This has been fed back to EGEE SA3. Ticket status *************** - With Philippa moving on to other work there is now an even greater need for T2Cs to follow up tickets for their region - T2Cs should be able to login to Footprints: http://helpdesk.grid-support.ac.uk/MRcgi/MRentrancePage.pl - Weekly reports are now sent out - is this working? - Any problems mail the support desk support at grid-support.ac.uk! EGEE ops meeting items of interest ****************************************** You can view the full agenda here: http://indico.cern.ch/conferenceDisplay.py?confId=19872 * (ROC DECH): The last updates of APEL rpms in production are from end of last year. The APEL rpms in certification fixed important bugs for sites in our region. What is holding them back to enter production? Oliver explained that the APEL patches are in YAIM4. This version of YAIM is being held up while smoothing of (unrelated to APEL) issues takes place. * UKI explained several sites were now limiting the number of queued jobs (e.g. 1000 for biomed). Steve and others thought this issue was already known about and that the limit can be per user not per VO. * GOCDB3 allows several settings for monitoring - Production + monitoring on - In downtime + monitoring on - In downtime + not monitored This came up as there was confusion as to why some sites in downtime were still being monitored. * There is still a request from WLCG MB/GDB for sites to move to SL4 before the end of the month. * Sites running ia64 were concerned about the dropping of LCG_2_7_0 which is the last release to support it. Oliver mentioned that 3.1 effort had 32-bit and then 64-bit AMD as priorities but not ia64 (used by very few sites).

11:35 → 11:45 Topics for DTEAM-PMB face-to-face meeting (28th Aug)

- What topics would we like covered at the meeting? - Some suggestions (focus should remain on higher-level areas) -- GridPP3 changes: reporting structure and representation -- Improving PMB-DTEAM relations. Is JC as the only point of contact a problem? Are there feelings of frustration in the DTEAM that the PMB is detached from the day-to-day problems and the reality on the ground? f so, how do we address? Do the DTEAM feel they have the mechanisms to engage the management? -- Input on site availability measurements (and how we might do this better). It may be possible to preview the August availability figures - a group discussion will help show areas of concern. -- DTEAM top 5 issues

11:45 → 11:55 Site response times

- Site readiness reviews prompted a PMB action - understand concerns around 2hr response times and layout what is possible -- What is the best practice for sites to follow -- Will most problems during working hours receive an acknowledgement on the 2hr timescale? -- How will things be dealt with when the main sysadmin is away. Can we insist on a backup contact? -- Are T2Cs now always copied on tickets to their Tier-2s. How often do tickets get reviewed and actively followed up by T2Cs? -- Is there any benefit from a UKI on-duty DTEAM memeber? To chase open tickets and raise problems? Is it duplicating effort.... do the Tier-2s already have something like this? - GridPP3 proposal http://www.gridpp.ac.uk/docs/gridpp3/ picks up on what is expected. --

11:55 → 12:00 AOB

- Actions http://www.gridpp.ac.uk/wiki/Deployment_Team_Action_items - Situation at Liverpool following unannounced SE change - why did the CloseSE name change? - ATLAS sgm use of SEs (see Greig's mail "Some sites have recently been reporting problems with their SEs, in particular, DPM sites. Cambridge and Durham have been experiencing out-of-memory errors which can be traced back to a large number of dpm.ftpd processes consuming system resources....." - The LHCOPN appears to have been disrupted by a fibre being cut in Germany.