Last modified 12 years ago Last modified on 10/10/08 20:33:45

dCache Tier Support phone conference

Participants : PIC(Paco), IN2P3(Lionel), BNL(Iris), dCache(Patrick,Paul,Dmitry)

  • IN2P3 : Urgent : High load on pnfs. Symptom is : CPU load on the PnfsManager is low. No postgres request takes longer than 500ms though the number of requests in one of the queues is reaching 1.6k. Lionel provided all necessary links. Paul is investigating. Dmitry will provide a script to check the actual postgres command sequence.
  • PIC : More details on pending Tickets next Tuesday.
  • BNL : Requests to stage a file can be triggered by either a read operation of a file which couldn't be found on disk or by manual stage or restore operations like the srm-bring-online. Unfortunately there is currently no way to distinguish the different sources of requests and consequently they can't be prioritized.
  • gridKa : Some gridFtp door have been not correctly configured, resulting in an unstable system. Doris will provide detailed information to support.
  • Offline request by Flavia which we may discuss next week : (Quote Flavia)
    • Are Tier-1 sites available to run a deployment test suite on their test instances before updating the production instance ? If so, this test suite will be made available.
    • On November 11th, the pre-GDB will be dedicated to Storage. The idea is to have the developers meet site representatives (Tier-1 and Tier-2) face-to-face. In this occasion, developers will give configuration and management hints to site admins and site admins will have an opportunity to express their problems and list their requests. Are sites interested in such an event ? Can developers get prepared ?
  • Next meeting : Tuesday Oct 14, 2008

Last modified by patrick @ Fri Feb 26 22:18:58 2021