Last modified 12 years ago Last modified on 02/03/09 16:59:28

dCache Tier I meeting Feb 3, 2009

Expected :,Tigran,Paul), In2p3(), Sara(Onno), Triumph(Simon), BNL(Jane), NDGF(), PIC(Gerard),gridKa(Doris), Fermi()

Topics :


  • Ticket 4292 PNFS Filesystem loop detected at PIC (Gerard is still investigation).
  • Plan is to have a second PNFS replica which is ROnly. Tigran is suggesting to run slony instead of applying postgres diff files.
  • Common space in the WIKI space. Will be setup by Patrick.


  • Network outages but not dCache related.


  • Question on whether to use gPlazma module or cell. Timur is suggesting to use the gPlazma cell as long as there are not performance problems. Timur will consult Ted for further information. Jane is proposing to use gPlazma cell with prio I and module with prio II. Timur agrees that that can be cone. Note : all gPlazma related file have to be copied to all doors if the module option is used for gPlazma. Timur will follow up on the using this mailing list.
  • dCache upgrade Feb 16,17,18. will send around suggested release.
  • Question : did any site split the billing table into small table. Suggestion is to index the 'date' inside the table. This should speed up queries significantly. Vladimir will send more information on this around. Jane/Iris? will provide the queries BNL is mostly using in conjunction with the billing database.


  • FZK is still observing high load on the pnfs system. The system is stable but under high load. Once the pnfs host rebooted unexpectedly. Mostly this is triggered by sudden high request peaks to pnfs. We would suggest to use available mechanisms (like pnfs dump threads) to find out what's going on during those peaks.
  • fzk wants to check consistency of the tape repository. We do this offline.


  • Load has dropped. System is doing fine.

Next meeting :

Same time next week.

Last Modified by patrick @ Sun Mar 7 00:53:24 2021