• Present: Maarten, Alberto, Jamie, Harry, Olof, Jan, Miguel, Gavin (notes)
• Sites called in: BNL, GSI, FZK, IN2P3, INFN, NIKHEF/SARA, FNAL, PIC
• Expts called in: LHCb
• Apologies: TRIUMF, RAL, DESY

Experience running over Christmas / New Year

• See FTS holiday log

SRM copy use

• Noted issue with dCache in 3rd party copy mode: difficult to scale since you must go through the gridFTP doors servers and there are usually less gridFTP door servers than disk pool nodes in most T1 deployments. SRM copy (in pull mode) starts a gridFTP process on dCache disk pool node, pulling from the source, so is able to utilise all of the disk pool nodes for transfers.

• Some issues still with SRM copy and FTS. Need more running experience to iron out all the problems when things go wrong on the SRMs - we get less monitoring on the FTS server side when we use SRM copy.

• Plan for re-run: see what rate and stability can be achieved using 3rd party copy mode, and then switch channels with dCache sites to use SRM copy to see if the rate and stability can be improved.

• T1 sites running Castor or DPM will stay running 3rd party copy, since SRM copy is not required (nor recommended) for Castor or DPM.

Re-run of SC3 disk-disk transfers

• Proposed ramp-up:
  • Tue: DESY, FZK, IN2P3
  • Wed: BNL, FNAL
  • Thu: RAL, CNAF
  • Fri: NIKHEF/SARA, overflow

• IN2P3 can do max 100MB/s until network upgrade, possibly next week (TBC).

• New Castor2 cluster at CERN - new SRM endpoint noted.

• New endpoints for sites should be updated in TierOneConfiguration.

• Ramp-up will use existing production FTS servers (lxshare026d). Will set up new FTS mid-range servers (fts10x) this week.