Deployment team

Tuesday 18 Sept 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

Experiment Reports ------------------------ LHCb (Raja) not much to report no simulation right now, reconstruction and analysis at ral stalled jobs - not sure why. dcache OK - investigating finally seeing SLC4 running OK - they now suport SLC3 and 4 ATLAS (stephen) - was on hols unsuire oif status CMS / ALICE - no-one present other exps - Pete Gronbech and AE noted that there are v few LHC vo's using the clusters now - Pheno and biomed are using them Report from ROC Meeting ------------------------------ CE will be ready for SL4 'soon' bdii next still doing 32 bit 1st, 64 later few complains about the change to glite wms being 'sneaked' in no other real issues arose Ticket status ---------------- LondonT2 (Dave) ====== only a few outstanding yesterday - Mostly with QM in hand caused by software installation - prob with the node the sgm jobs run on UCL - central should be happy and active but few jobs there - initial burst then dwindled off = passing tests don't appear on blacklists Networking problens at RHUL ansd brunel out of memory error as reported on storage list SouthGrid ====== few problenms at ox - new CE on sl4 some instability with 4444 error - gridICE eating cpu high load hoping to get new kit up n running by next week - will migrate service notes to new kit NorthGrid ====== * noone present ScotGrid ===== Glasgow - Fairly quiet, pheno / biomed using 75% of ~200 slots, Atlas ~16% --rest idle - plan to upgrade to SL4 this week Edinburgh - Issue with GPFS on SL3 (suggest to try Jon Wakelin at Bristol) Transfer Tests ----------------- castor slow for CMS - ral to imperial phedex? Apparently should be OK to do CASTOR/dteam tests out glexec report - Mingchao ------------------------------ sent out report last week some discussions with developers - link to chep talk - most recent info. any Q? - wlcg workshop we will have to use it, but can be used in diff ways ie, suid root or as regular user, or off. Dave Colling - your recommendatopm? MC - no doc on framework job - seems to be per VO setup if you trust VO can be set suid root, but still need audit trail up to sites how they want to run he couldn't test. Devels clainm it doesn't change inteacrtion with batch system. Reccomend that sites test before deployment check per Q with and without glexec - doesn't have environemt to test 'is it a secure product' - ' if we deploy at loindon do we open to poss abuse' - - discussion ensued - lack of documentation about the model - partially teted by MM - can't say for certain. based on curr tests - it should be up to site to make own decisions is the risk caused by the poss to grab proxy etc the 'fault' of pilot job policy. there are presently ~10 ways that VOs can submit pilou jobs we assume that pilot job framework is 'secure' to begin with can the docs be revised with risks of running in each mode (dave colling only skim read) - together with details about testing done in each part - or lack thereof .... continued discussion about glexec continued... Poss compromise of UK root CA ------------------------------------- - will require new keys to be installed on machines and users into their browser Follow up will happen by email later (Action Mingchao) Review of Actions --------------------- http://www.gridpp.ac.uk/wiki/Deployment_Team_Action_items D-060630-12 Dave Colling -- will mail to list. no activity to 48h it gets cut off - CLOSED D-061017-2 - Dave C - to take on - Gridview includes failures when sites fail caused by external problems - CLOSED D-061103-19 - ongoing work by Gidon - extended funtionality D-070220-3 - AE- update wiki D-070410-1 - Dave C - on indico - will close once mailed D-070807-2 - glue schema - has happened - CLOSED D-070807-3 - interruped jobs - posted to italian list - no meaningful reply - will follow up - ONGOING D-070807-21 - ALL - circulate JC-070822-1 - risky sites may get compensation if they take risks with software - ongoing D-070807-1 - Greig - CLOSED -- Action to update wiki and move off the closed actions Security challenge - not happened yet - Allesandara. ----------------------- contribs to EGEE07 - Dave, Mingchao, Graeme, Stephen, Mona, Pete all going ------------------------ AOB - ----- 1) Mingchao - T2 tech meetings - please add item to agenda to provide security contact info to MM. GOCDB not helpful - 'security officer' - needs a name as well - Should it be the site security or grid site admin? Concern over widely disseminating info - more likely to be a small 'campus' list - have to be careful with calling 'off duty' people out of hours - but can sites check GOC DB info 2) Greigs SRM workshop - unanimously supported - poss around 12th Nov. (concern at close to hepsysman - buit diff to choose a non-clashing date) interest from beyond GridPP - good / bad to open up to non-uk people - invites to non-uk and all UK Meeting closed 12:30

11:00 → 11:20 Experiment problems/issues

Review of weekly issues by experiment/VO

11:20 → 11:35 ROC update

ROC manager update ************************* Ticket status *************** EGEE ops meeting items of interest ******************************************

11:35 → 11:45 Update on networking tests/direction

- Review of Andrew's milestones - Current status and problems - What links are being exercised now by CSA07/FDRs? - Planning

11:45 → 11:55 Update on glexec

- Status from Mingchao on glexec - Comments based on David Grope's CHEP talk/paper - Any other security related discussion items

11:55 → 12:00 AOB

- Actions http://www.gridpp.ac.uk/wiki/Deployment_Team_Action_items - Contributions to EGEE07 - Status of ROC SSC2 - storage