Swiss WLCG Operations Meeting on 2010-11-25

• Date and time: 2010-11-25 9:30 CET
• Place: EVO (password: chipp)
• External link / EVO: http://evo.caltech.edu/evoNext/koala.jnlp?meeting=MIM9Ms292iDnDu9e9tDa9I

Agenda

• Cluster Status
  • NFS problems
  • dCache problems
• Ongoing projects
  • Migration of Pools from Solaris to Linux. SolarisToLinuxMigrationDcache
  • New firewall, network simplification
  • PhaseD hardware is arriving. ETA January.
  • Closure of old UI
• Quick Restartup of CE and SE
  • DcacheFullRestart
  • grid-service script
• Seagate disks replacement approved.
• Wiki migration and changes (User registration, etc.)
• AOB

Attendants

• ATLAS: Gianfranco
• CMS: Derek
• LHCb: Roland
• CSCS: Pablo

Minutes

• There were problems last week with NFS and dCache. NFS stopped working and we had to change the driver and firmware of the disk controller. dCache has problems with the Solaris gridftp doors, so they had to be shut down also.
• The UI is to be closed now.
• User registration in Twiki has to go through Derek or Pablo from now on.

Action items

• Derek asked for a generic dsh command to restart the Storage Element to be included in the DcacheFullRestart page
  • Derek: To give an example of what I think is an adequate level of instructions. This is my Tier-3 dcache start/stop docu which also Leo already was able to follow a few times, even though he is not a dcache wizard. The docu is proven adequate if somebody else can successfully follow it. I mainly asked, because we had an issue where Pablo and Peter were on vacation, and Jason and I were puzzling for quite some time on how to do a complete restart (of CE in that case). We decided against it because we felt we had not enough understanding of the setup.
• CSCS will measure the bandwidth restriction that the new Firewall will cause, if any. The answer is: the firewall does not create any artificial limit by itself. We have measured 8 Gbit/s when routing between different networks, which is the maximum we were able to take from any infiniband network link.
• We will also inform Derek about the new twiki SLA and support contact (always with CC to Pablo)
  • Derek: As stated before, the wiki is an important service for us in CMS. We do part of our user communications through it, and it has all our documentation about Tier2 and Tier3 issues. So, this service must be reasonably available. We want a generic address to which we can send problem reports or a trouble ticket. I think it would be optimal if it was an archived CSCS mailing list with the company being subscribed to it.
• We will also put in contact Derek with the SOracle guys for him to be able to buy a couple of the Seagate disks we give back to Oracle.