Last modified 13 years ago Last modified on 09/28/07 17:01:37

dCache, SRM 2.2 deployment meeting Friday

Participants : Gene, Timur, Flavia, Greig, Patrick

  • Dimitry couldn't make it. This is his summary :
    I fixed these two items from Flavia' list:
            2.  srmPrepareToPut returns SRM_FAILURE instead of SRM_NOT_SUPPORTED at
            request level in case the protocol specified is not supported. SRM_FAILURE
            should be returned only at file level in this case  : - URGENT
            3. srmPrepareToPut returning SRM_AUTHORIZATION_FAILURE instead of
               SRM_AUTHENTICATION_FAILURE in the following  :
    and almost done on this:
             Put status changing with time  - URGENT
  • BNL : Pnfs seems not to be mounted on the SRM node, so that we observe the same srmls effects as at IN2P3 in the past.
  • UCSD detects performance degradation if pnfs directories become 'full' (several thousands)
  • Atlas and LHCb insist in having up to 11000 entries within one directory. Flavia will find out if there is a usecase for doing a 'srmls *' on those directories. She will furthermore test large directories against DESY and Endinburgh. CMS is fine with about 1000 entries per directory.

The workshop at Edinburgh will take place 13 noon to 14 evening.

  • The number of participants is limited to 108
  • dCache Tier I's have been invited. SARA, PIC and IN2P3 already agreed.
  • Greig will let us know who else he will invite. (may be using user-forum at
  • Financial support may be shared between gridPP, WLCG deployment and participants.

Last modified by patrick @ Sat Feb 27 16:05:09 2021