Tech Meeting Minutes 20111201

From ScotGrid

ScotGrid Tech Meeting 1 December 2011
Agenda: http://indico.cern.ch/conferenceDisplay.py?confId=148172
Present:
Mark (chair)
Sam (minutes)
Mike
David
Andy
Wahid

- The main issue is Durham. We have had responses from multiple directions concerning response times on tickets against the site.

Mark: it has been asked that tickets for Durham be treated slightly differently than for other sites, to allow longer time to respond. There are certain things we can't do remotely‚ we can try an approach where Mike updates tickets, and *requests* work from Glasgow/Edinburgh in the back end for tech support.

-

Durham tickets

75488 - compchem We suspect this is a VOMS server issue, linked to the enmr.eu issue at Glasgow. We need to talk to Brunel, who are the other site who support compchem, to check our configuration against them. We *think* we and Durham are configured correctly, but‚ cvmfs ticket - Alessandra understands that this is on hold. We do need to pick a date to update the ticket with, though. Mike is happy to say, say, 31st Janurary for a due date.

76288 - ATLAS nodes being problematical. Stuart fixed - this was poorly configured nodes due to site configuration management system issue.
76487 - LHCb ticket. No space left on device. Needed to go to Alessandro de Salvo to get him to clean up *ATLAS* software to make space on the experiment software area for LHCb (and everyone else). (Mike has done this).
biomed ticket - Sam fixed by deleting dark data on Durham SE. Sam to mark it solved to get Biomed to actually respond.

Chris and Ewan are looking into pheno proxy renewal testing to explore the issues pheno have been having. Mike will get back to Chris on this.
Mike: someone got back to us about how the procurement process worked for the networking bid. What I've asked for is a complete refresh of all the switches behind our network infrastructure.
-

Edinburgh

76815 - Chris's test jobs. It was a problem with our elderly lcg-CE, basically not an issue. Just ensuring it's not listed as active still.
cvmfs ticket: got the green light from ECDF now, deploying today! Should be able to make it before Christmas pending ATLAS testing. Should be fine.
spacetokens 75542: this is in progress. Wahid: waiting on Victor to free up space. Will also get better in general when we have more disk after procurement.

- Glasgow

enmr.eu ticket: started off being lcg-tags, but actually is a user mapping issue for enmr.eu. It looks like we're configuring the VO correctly according to available data. They use the INFN VOMS server, like compchem. Our plan is linked to the compchem issue at Durham (same VOMS server).
ticket: we had a problem since updating to torque 2.5.7, where we had to also update our MPI install and related code (CASTEP). This has proven less trivial than we hoped, so we need to decide if we want to just drop support for MPI and then resolve stuff more deliberately.
DPM Head node ticket: "resolved" (was against svr025, which is an EMI node)
Dave: the other issue is that there is an acknowledged memory leak in Torque 2.5.7, which causes it to die fairly frequently (especially when rolling over logs at midnight). This is being followed up.

AOB
ScotGrid New Year Meeting: 9 February 2012, Glasgow. There may be an "team bonding" trip to Auchentochen in the afternoon.
Christmas Cover: Generally, our approach is best efforts for Tier 2 sites over the Christmas period.

---

11:02:11] David Crooks joined
[11:02:40] Andrew Washbrook hiu
[11:02:41] Andrew Washbrook hi
[11:02:57] Mike Johnson joined
[11:03:12] Mike Johnson hello, just sorting out sound as usual...
[11:04:38] Andrew Washbrook only on simpsons questions
[11:06:06] Wahid Bhimji joined
[11:10:17] Andrew Washbrook sure
[11:42:48] David Crooks They do have a cafe
[11:43:18] Wahid Bhimji I think so - indeed far future
[11:44:13] Wahid Bhimji To be honest they don't look after ecdf much
[11:44:39] Wahid Bhimji (but last times we have run loads of jobs as all local users go off) . Put at-risk in Goc for all the period
[11:44:47] David Crooks Yep
[11:45:10] David Crooks We put in an at risk as well.
[11:45:30] Wahid Bhimji bye
[11:45:32] Wahid Bhimji left
[11:45:33] Andrew Washbrook left
[11:45:33] David Crooks left
[11:45:34] Mike Johnson left
[11:45:35] Mark Mitchell left