Swiss WLCG Operations Meeting on 2010-12-09

• Date and time: Thursday 09 at 9:30
• Place: evo
• External link / EVO:http://evo.caltech.edu/evoNext/koala.jnlp?meeting=M9MIMi2I2uDIDt9t98DD99

Agenda

• Cluster status
• CVMFS for Atlas (http://indico.cern.ch/getFile.py/access?contribId=32&sessionId=6&resId=1&materialId=slides&confId=76896)
  • Quote: It’s highly recommended to have 25-50 GB of disk cache local to the nodes mounting CVMFS, although you can set this to lower values
  • Not in production for CERN.
  • x86_64 not working yet.
• Memory problem for Atlas
• Discussion about 2K jobs limit (aborted rate)

Attendants

• CSCS: Peter, Pablo
• CMS: Derek, Leo
• ATLAS: Marc, Gianfranco
• LHCb: Roland

Minutes

• The migration from Solaris to Linux Thumpers is being delayed because of issues with the firewall
• The amount of disk that we can dedicate on WNs for CVMFS is very limited. Maybe we can set the cache to 3 GB, maximum 4 GB. Also, it would be nice to know the hardware recommendations for the squid service.
• It seems that some Atlas jobs are getting killed because of memory limits. CSCS, Leo and Derek will get involved.
• The 2000 job limit per VO is causing some headaches to CMS. We will deploy the next Moab version as soon as possible, but not before Christmas.
• Also, we agreed on making a round-trip over the cloud's sites to report problems and news in the Operations meetings, not only CSCS.

Action items

• item