ScotGrid Technical Meeting

Wednesday 28 May 2014, 10:55 → 12:00 Europe/London

Other Institutes

Gareth Roy (University of Glasgow)

ScotGrid Technical Meeting 28 May 2014

 

Present

 

David Crooks

Sam Skipsey (minutes)

Gang Qin

Gareth Roy (chair)

Andrew Washbrook

 

 

-

Durham

-

 

N/A

 

-

ECDF

-

 

AW: ATLAS pilot factory ticket. turned out to be a pilot factory issue

(not a local issue). Seemed to be a problem with renewing delegation

requests (from a single pilot factory). Possibly connected to the

migration to "agile" infrastructure at CERN?

Ticket is ongoing. Is not impacting on jobs on cluster.

 

EMI3 migration marked solved in ticket. (Just niggles to clear up,

finish off the APEL monitoring - it's parsing the one CE fine, just

need to add second + push to RAL)

 

Elena mentioned our high failure rate. We have some ATLAS production

that seems to be SIGKILL'd by the batch system.

(Poss. related to the QMUL issue with SGE honouring VMEM limits for

jobs, and then killing them when exceeding.)

 

Wahid is bringing the new storage online.

 

 

-

Glasgow

-

 

DC: Two power cuts on Monday night (Gareth was in to look at things until late).

Tues was bringing things up.

Currently looks "okay" - don't seem to have lost much, if anything.

Yesterday afternoon we brought back the WNs.

Overnight we've been running LHCb and user jobs, but GangaRobot jobs failed.

 

We got ticketed for the PerfSonar boxes having issues post/during power cut.

 

GQ: condor issue - fine-tuning, documenting the config files.

 

SS: Took the opportunity post(the ATLAS fails were due to the cmsd

daemons falling over on the SE, without taking down the xrootd)

 

 

-

AOB

-

AW: going to ATLAS SW&Computing week next week. Pushing HPC things

forward with ARCHER.

   There are a few solutions to various issues with ATLAS stuff on

HPC - using modified ARC setup, sshfs for job submission; mostly tying

together various things. Looking at job scheduling on ARCHER. ARCHER

people quite open to working with us.

 

SS: test Ceph instance on test cluster (S3 integration w/ DPM next on list)

 

AW & DC discussed monitoring. AW planned visit to Glasgow to discuss

this further (in 2 weeks).

 

 

-

Chat log:

 

No contents.

11:00 → 11:15 Tickets

11:15 → 11:25 Research Updates

11:25 → 11:35 AOB