Week of 060130
Open Actions from last week: Castor2 upgrade scheduled for 08.30 - 10.30 Monday
Chair: H.Renshall
On Call: Jean-Philippe + David
Monday:
Log: Nothing
New Actions: Maarten to contact sites participating in disk to tape tests to start their ramp up.
Discussion: Jamie has scheduled this week for CERN disk to tier1 tape runs. Original objective was for each site to take data from CERN to tape at 50 MB/s using 5 drives of their current technology. In practise we expect particpation from FZK (2 drives only),
IN2P3, PIC, Triumph (2 drives only), DESY, SARA and BNL. O.Barring requested exclusive use of the c2sc3 cluster for 2-3 days after the disk-tape runs are completedI. J.v.Eldik announced that setting up of the 4 separate experiment castor2 instances should be finished before April with Alice being the first this week.
Tuesday:
Log: Tape tests underway to most sites.
New Actions: no new actions.
Discussion: Problems with remote disk pools filling up faster than the migration to tape - use FTS to reduce the rate (but the process is manual).
Wednesday
Log: Nothing
Actions: Continue disk to tape run till Monday morning.
Discussion: Disk to tape tests going quite well. BNL stable at 60-80 MB/s, FZK at 35-45 on 2 drives,
IN2P3 at 40-60 but with holes, SARA at 50 MB/s stable. ASGC is running disk to disk at 60-80 MB/s and want to continue till 11th. Planning is to continue disk to tape till Sunday then Castor team will take the c2sc3 cluster for several days. ASGC will be switched to production castor2 on Monday.
Thursday
Log: Nothing.
Actions: Continue disk to tape run.
Discussion: Sites working to around 50MB/s with short hiccups - dcache problem at Sara, HPSS problem at BNL. PIC has turned on tape migration and is taking 25-30 MB/s. DESY is taking 80-100 MB/s. Lumpy behaviour at
IN2P3 is attributed to the local disk to tape migration cycle.
Friday
Log:
Actions: Change ASGC FTS to production castor2 Monday morning.
Discussion: Disk to tape runs going well. Both ASGC and
IN2P3 had network asymmetric routing problems when the production global bdii was moved from the GPN to the LCG networks.
IN2P3 reconfigured themselves but T.Kleinwort has resurrected bdii002 on the GPN network until ASGC are ready.