Deployment team

Tuesday 17 Jul 2007, 09:45 → 11:00 Europe/Zurich

VRVS - Plane virtual room

Jeremy Coles

09:45 → 10:05 Experiment problems/issues

Review of weekly issues by experiment/VO - LHCb - CMS - ATLAS - Other

ATLAS - no major site problems. Migrated to DQ2 2.3 - awaiting rampup. Using CASTOR 2.1.3 for T0 transfers. Restarting production soon. Need to migrate some CASTOR 2.1.2 data to 2.1.3 or dCache. -- Durham Q closed as reboot due on Wednesday -- UCL HEP / Central - Duncan aware and chasing up. LHCb - Nobody present or any issues to be raised CMS - Castor work (migration to 2.1.3 in progress) - Some stuck jobs at QMU

10:05 → 10:20 Status of SL4 testing - UK rollout

- Summary of findings from last week of testing - What are site plans to upgrade over next 6 weeks - Any experiment objections/new findings to be reported?

CMS explicitly stated wish for SL4 - general move to try and implement by end August. Chris Brew - as been working through their RPM list and found no clashes with base install etc. ATLAS list also OK according to Derek. RPM list from installed SL4 WN - http://www.gridpp.ac.uk/wiki/RAL_Tier1_Batch_Worker_Nodes_SL4 . Derek to provide kickstart summary if poss. Duncan added that some minor hassles (glite rather than SL4) such as looking for EDG-get-broker which is now glite-get-broker. Need to get critical mass for experiments to properly test. Rollout plans: Scotgrid:: Glasgow - Pushed back to w/c 20 Aug (CHEP Papers / Hols) Edinburgh - New site will be SL4 from start. Plan to go live end Aug anyway Durham - Phil on leave but August likely. Northgrid:: Waiting for Alessandra. Southgrid:: Yves testing at Birmingham. RAL testing now. Others to follow LT2:: Should be OK for end Aug Brunel - Testing now - 40 cluster running SL4, 44 cluster running SL3 still. Summary - Should have >50 % sites on SL4 by end Aug Tier1 - Will be getting new robust hardware for a new CE, then migrating WN / etc into it as requested by experiments

10:20 → 10:35 Sites transfers and data management update

- Report from Andrew on status of work in this area - Where are we vs GridPP milestones - What other work should be being done here?

(um - poor minutes as I was talking) Need some storage at T1 that we can use for testing - CASTOR instance? then we can methodically assure that all T2 sites can hit target figure of 300Mb/s Use of Gridmon for historic site checks - Are they suffering from rate-limiting (discovered packet loss at birmingham. now seeing same at cambs - investigate to see if it's caused by bandwith limiting) Action AE - Update matrix for current round of tests and produce summary to dteam list with recommendations. Need to follow up ralt2+glasgow to oxford tests. (higer load on oxford end with dcache than dpm feeding)

10:35 → 10:50 Revisit - killing jobs discussion!

- Taking forward proposals to the TCG. How should jobs be killed such that useful information is sent to users. - Results from other sites examining CPU/WALL time tails

Skipped - Short on Time

10:50 → 10:55 AOB

1) We now have 5 votes to use on the top issues page (as per Philipas email) - See https://twiki.cern.ch/twiki/bin/view/EGEE/SA1_TCG 2) use EVO for next weeks meeting