Ops minutes

17th March 2015

Agenda: https://indico.cern.ch/event/381499/

LHCb: Raja, nothing much to report. Running MC, some problem with latest the dCache update, affecting dCache Tier-1s and RALPPD. Daniela asked whether Imperial dCache was affected. Raja: No.

CMS: Pete: Brunel having problems recently. Daniela: Possibly the DPM upgrade, seems OK now. Nothing else to report.

ATLAS: UK meeting discussed HC tests which are having some problems, related to too frequent sending of tests. Johanes solved the problem. RHUL also affected, storage problem fixed but took a log time to come back online due to too infrequent tests. Please report to UK cloud support if you are having similar issues. New Rucio client version released, many new features and bugs fixed. MC15 started. ATLAS keeping grid mostly full, 25 million events per day planned, but not yet realised. Also report on US FAX testing, if interested in this see the presentation at the last ATLAS meeting, also attached to this agenda.

Other VOs: Tom: no news from LIGO or UCLan yet. UCLan running test jobs, plan to use cvmfs. LSST?

Dirac: We reviewed the table, Bham VAC still testing. Any other sites plan to implement VAC? Glasgow intend to get VAC running. Daniela: we have nearly finished production instance of Dirac. Sent a request for 'dirac pilot' to join appropriate VOs. Could small VO managers check and approve.

Operations bulletin

- Multicore queues being enabled at Bham and will then set up multicore accounting and transition to HTcondor later. John Gordon also requests all CEs be updated. Glasgow don't see the point of enabling on CEs with no chance of running a multicore job. Ewan: multicore flag supposed to be harmless however.
- Please check Andy McNab's cloud status table.
- Wahid's leaving, we thank him for all his hard work. 

WLCG operations: WLCG HTTP task force approved. Sam to join, Duncan also.

Tier-1 update: still have network issues.

Storage and data: An annual DPM collaboration board meeting will take place in coming weeks.

Documentation: Documentation was reviewed at the Core Ops meeting.

Monitoring: please let David know plans for SL5 removal. EGI are canvassing for site opinion with a possible intention to decommission SL5 from UMD in particular by the end of the year
So not instantaneous, but long term planning.

On-duty: Daniela: strange things happening with site availability at UCL.

Rollout: Daniela has update the staged-rollout web pages. We're all up to date.

Security: meeting scheduled for this week, might not happen though.

Tickets: Matt went through the tickets. sno+ ticket to the Tier-1, user not showing up in the grid map files.

Site round table: anyone wish to mention anything. Kashif: c-groups now working for ATLAS at Oxford.
Gareth: note discussion on rebus and gstat on tb-support. 

Actions review: 

We went through the actions, most are still in progress. Argus ticket: how can we establish when all sites have updated argus? Argo testing: refactored SAM testing, in progress. Agreed no need to have an action on the UCLan situation, to be closed.

pre-GDB and GDB meeting. Pete went through the agenda. Discussion deferred until someone who was actually present (e.g Andrew McNab) is in the Ops meeting.

Tom: GridPP discussion sessions need a well-structured agenda to be circulated beforehand.

The Hepix meeting is on next week so no Ops meeting. Ewan queried the overlap. Is any of it on Vidyo? Pete: yes.

AOB: Dan: reminder please register for GridPP34 by the end of the week, especially if you need accomodation. 

Chat window:

Daniela Bauer: (17/03/2015 11:07)
My sound just dropped, will have to restart Vidyo
Tom Whyntie: (11:15 AM)
Aha! That's what that was. I will approve the request. Thank you.
Done!
Daniela Bauer: (11:17 AM)
Thanks. My audio is temperamental :-(
Daniel Peter Traynor: (11:21 AM)
set parallel = true in /etc/apel/parser.cfg on a cream ce.
Ewan Mac Mahon: (11:30 AM)
I think the Oxford status on that was that we still have some SL5 nodes at the moment, but we're completely relaxed about the idea of getting rid of them before the end of this year.
I don't think we were the only site in that position.
It's UCL.
'nuff said.
Daniela Bauer: (11:32 AM)
We have no SL5 grid nodes at IC. We'd be happy to install anything SL7 compatible thugh :-)
Ewan Mac Mahon: (11:33 AM)
Oh yes, I wouldn't mind moving some of the SL5 boxes straight to seven in principle.
Matt Doidge: (11:35 AM)
https://ggus.eu/?mode=ticket_info&ticket_id=112350
Govind: (11:36 AM)
Does anyone know when DPM will be available for SL7 ?
Federico Melaccio: (11:40 AM)
same at RALPP
(mc jobs)
Ewan Mac Mahon: (11:45 AM)
There is a note above the table saying that it's being gradually re-ordered to be most-relevant in the most obvious position.
Govind: (11:46 AM)
Sorry I have to leave now.. bye
Tom Whyntie: (11:50 AM)
Yup
Ewan Mac Mahon: (12:00 PM)
I think the questions to be answered are basically: Why are we running cloud resources? and Do we want to be running cloud at more sites? and Do we want to make the existing ones larger?
The latter ones flow somewhat from the first one though.
wahid: (12:01 PM)
Bye - see you guys somewhere sometime 
Tom Whyntie: (12:01 PM)
Bye