Deployment team

Tuesday 16 Oct 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

Present ====== Yingqin Zheng (observer) Jeremy Coles (chair) Pepe Gronbech Duncen Rand Jens Jensen Frederic Brochu Raja Nandakumar Graeme Stewart Stephen Burke Barney Garrett Alessandra Forti Mingchao Ma (minutes) Experiment update ============== LHCb: Everything is fine. CMS: It is not going well due to problems with the cms software framework which is hampering both the Tier 0 and Tier 1. This is resulting in lower rates. Data first hit RAL ~1week ago and should reach the Teir 2's imminently. (note these are different datasets so its not a delay getting data from Tier 1 to 2 ATLAS: capacity and funtion test finished last week. DDM updated going very well and sorted out most of problems. Setuping new analysis test using ganglia framework. it has been setup on a temporary/private web page and will move to dashboard as soon as possible. Issue about getting some tier-2 sites into the current tier0-tier1 tests, it will be useful to have at least one Tire2 site involved the test. To join the test, Frederic can forward site's request to LHC. But there is no new test plan at the moment, have to wait. After ral-ppd updated to SL4, there are many problems with ATLAS jobs. It is due to configuration problem and has been sorted out this morning. ROC update ======== GDB update --------------- Please refer to agenda for detail; PIC notes: (refer to the agenda) Problems (jobs doing nothing) are due to not only log file but also input data file downloaded from outside of uk. Will take time to fix the problem, be patient! All sites must support WLCG VOs to allow them to be able to run pilot jobs. Pilot jobs & Glexec: Southgrid, Scotgrid and Northgrid will run glexec without suid root. Security policies: a few policies have been approved. Job priorities WG: refer to agenda VOMS attributes usage: at moment, different components in MW use VOMS attributes in different way, but it need to be consistent. ROC manager update --------------------------- configure BDII and implementing, none of UKI BDII is listed bad in the recent test. See http://indico.cern.ch/conferenceDisplay.py?confId=22567 for more detail. Ticket status --------------- Quite a few UKI ticket are still open. SouthGrid, NorthGrid and ScotGrid have no problem to receive tickets. VOmS for LHCb ticket on Oxford, will sort it out soon. CE failure on IC. Bringham SE failure on PPS. A few more tickets, please refer to footprint. EGEE ops meeting items of interest ------------------------------------------- GGUS ticket #27747, refer to the agenda. Site availability review ================ http://www.gridpp.ac.uk/wiki/SAM_availability:_Monthly_summary_table http://www.gridpp.ac.uk/wiki/SAM_availability:_July-September_2007 Trend seems to be improving. Particular concerns on three sites: QM, Sheffield and UCL-central. AOB === Will go through actions list again next week!

11:00 → 11:20 Experiment problems/issues

Review of weekly issues by experiment/VO - LHCb - CMS - ATLAS -- getting some tier-2 sites into the current tier0-tier1 tests -- FB: requires 0.5TB free disk - separate from prodution pool - Other

11:20 → 11:35 ROC update

GDB update ******************* 1) Take a look at the experiment talks for a good overview of where each experiment has got to in its testing. http://indico.cern.ch/conferenceDisplay.py?confId=8488 2) The MB will receive a request to notify all sites that WLCG VOs must be able to run pilot jobs to be used. 3) Common COmputing Readiness Challenge 2008 is now being planned (wlcg-ccrc0 at cern.ch). Greig - any comments? Mainly Tier-0 and Tier-1 for now. 4) Security policies: Grid Security Policy (v5.7) approved. Grid site operations policy (v1.4) approved. Still need covering documents for specific grids. VO operations policy (v1.4) - final call for comments. Pilot Jobs Policy (v0.3) in progress. 5) Job priorities WG (Dietrich Liko). Problems with early implementations, aim is still to get experience with a simple model. Complications hit with ACLs/groups/VOMS FQAN matching... looking at options to shares and use of deny tags. Tests ongoing on certification testbed. 6) Still lots of discussion needed on VOMS attributes usage! (See above!) 7) Greig - any comment on pre-GDB meeting? ROC manager update ************************* Meeting happens today at 10am so check the agenda for items discussed: http://indico.cern.ch/conferenceDisplay.py?confId=22567. Main topic is BDIIs. No UKI ones are listed as "bad". Ticket status *************** Many tickets open - especially LT2. Will go through them at the meeting. EGEE ops meeting items of interest ****************************************** => GGUS ticket #27747 raises the issue that if a WMS/LB node has its host name changed, the configuration needs to be updated on any UI that uses it. Is there a procedure for doing this change? The suggestion was to cascade tickets through the ROCs. It will be discussed at the ROC manager's meeting. PIC notes: # We discover ATLAS production jobs doing nothing (sleep 3600) after an apparent failure staging output (apparently log) files into voatlas01.cern.ch. We were not aware of local SE problems at that time. Could atlas production jobs stage out files first into the local SE and then, if needed, transfer them to CERN asynchronously through FTS? # Many CMS prod jobs detected running with very low efficiency. CMS prod manager contacted and are having a look into it. First we thought it had to do with the dCache system no cappable of serving data through dcap fast enough, but it seems that it could also be related to the merging jobs of CMS opening many files in a not very efficient way. Experts still investigating.

11:35 → 11:45 Site availability review

- Concerns and comments on the updated SAM charts http://www.gridpp.ac.uk/wiki/SAM_availability:_July-September_2007 - Current status -- UKI-LT2-UCL-HEP and UKI-SOUTHGRID-OX-HEP accounting issues -- UKI-LT2-UCL-HEP and UKI-NORTHGRID-SHEF-HEP SAM ops issues and both are excluded under FCR by many VOs.

11:45 → 11:55 TBC2

11:55 → 12:00 AOB

- Actions http://www.gridpp.ac.uk/wiki/Deployment_Team_Action_items - "To improve the chance that the work in the grid observatory is relevant, it would be good to have feedback from SA1 on what types of recurrent fault situations disturb operations. "Black hole" detection was one type of recurrent fault that was treated with EGEE trace information". Do we have any items to list?