Hi,
Let me know if you want to add anything.
======================================
-- CVMFS: started HC tests for RHUL and Liverpool this morning
-- QMUL: problem with panda limits on transferring and running jobs.
The limit on transferring jobs has been increased but the limit on
running jobs has never been tuned for anyone and might also be one
of the causes of the up and downs of running jobs.
-- Imperial: long standing problem with the sw full has been solved
deleting releases. Production is running again and the site is
unblacklisted.
-- New T2D requested by the cloud: RHUL LIVERPOOL SHEFFIELD RAL-PPD
-- A/B/C/D sites will be reviewed every month Manchester had big
problems in October due to bad connection to IN2P3 has been demoted.
Problem has been discussed with atlas. We are not the only one
with bad connections with specific T1s and atlas will review the way
T2D are associated to T1s. CA cloud also doesn't have enough
bandwidth to sustain T2D traffic.
Manchester in the meantime is upgrading the hardware of the DPM
head node and DPM to 1.8.2 with mysql 5.5. This should happen at the
end of the week.
-- There currently is a two day workshop about federated data stores
http://indico.in2p3.fr/conferenceTimeTable.py?confId=5527#20111122
You might be interested to browse through the talks to see what
is going on.
-- Request from Atlas to join LHCONE project. https://indico.cern.ch/getFile.py/access?contribId=3&sessionId=0&resId=minutes&materialId=minutes&confId=161077
A list of T1s and T2s are proposed to be followed up
carefully. The ATLAS approach is to evaluate LHCONE in a managed
way and gain expertese in liaison with Network providers.
Cloud Squads are requested to contact these T2 sites to validate
their participation:
- if they agree
- deploy Perfsonar-ps in the coming weeks (already
done for 4 sites)
Michael will manage the coordination with the network
providers.
For other sites:
- if connected to LHCONE, the sites can stay
connected but they are responsible for network problems. ADC
will just post a ticket to ask to solve the issue
- if not connected to LHCONE, the sites should hold
off.
Additional talks about LHCONE and perf-sonar were given at the atlas
S&C week.
https://indico.cern.ch/materialDisplay.py?contribId=34&sessionId=1&materialId=slides&confId=119171