ScotGrid Technical Meeting

Wednesday 18 Jun 2014, 10:55 → 12:00 Europe/London

Other Institutes

Gareth Roy (University of Glasgow)

ScotGrid Technical Minutes 18 June 2014

 

--

Gareth Roy

Sam Skipsey

Gang Qin

Ewan Steele

Oliver Smith

Wahid Bhimji

--

 

-

Durham

-

 

Ewan: whole segment of WNs dropped out of the batch queue due to

"network congestion" thanks to ATLAS data flow. (SLURM removes

uncontactable nodes from it's batch queue to maintain performance.)

 

[Gareth asked about how the ATLAS data transfers are configured. Due

to legacy issues, Durham is currently using rfcp - as they need 1.8.x

releases of DPM everywhere to use root direct access. The data flows

were filling the 4x1GB core access.]

 

Ollie working on DPM, upgrading the old nodes to SL6!

 

Now running almost entirely with SELinux=enforcing (!) with no problems to date.

 

Need to email APEL-support and sort out accounting migration. (Gareth

recommended ticketing them, as it also lets us raise issues in Ops

etc.) (Ollie now has a GGUS account for ticket support.)

 

Planning 2 wks downtime in July for migration to the new (HPC) machine

room (with water-cooled doors).

 

-

ECDF

-

 

Wahid: Andy sorted the APEL problems last week. We installed our new

storage, joining the "Petabyte Club" in GridPP.

 

Had some ATLAS jobs which made ECDF annoyed at us (by untarring files

with lots of inodes into disk, causing metadata thrashing).

 

A ticket about Heartbleed vuln (but mistaken).

 

-

Glasgow

-

 

Gareth: things going fairly well. We just finished updating cvmfs to

2.1.19 (the new baseline for LHCb). Additionally, just updated

bouncycastle-mail to avoid SHA-2 problems.

(re: GDB presentation by Alessandra)

 

Gang + Gareth started restructuring the test cluster for new role

testing "cloud on containers", possibly overlay grid site on top.

 

Finishing pulling together things for tenders.

 

Gang: (Test machines will be moved to SL6.5)

 

Sam: (disk074 now has no files that anyone cares about on it, so we're

going to reinitialise the RAIDset and read it to the DPM) (Also need

to actually clean up non-rucio paths in the DPNS.)

 

 

-

AOB

-

 

(Cloudsoft?)

 

The Quarterly Report is coming - please tell Gareth about interesting

things for the Narrative section.

 

-

 

Chat log

 

-

 

 

Gareth Douglas Roy: (18/06/2014 11:22)

SLides

https://indico.cern.ch/event/272622/session/0/contribution/3/material/slides/0.pdf

11:00 → 11:15 Tickets

11:15 → 11:25 Research Updates

11:25 → 11:35 AOB