21 June 2006
IT amphitheatre
Europe/Zurich timezone

Location: IT amphitheatre, B31, CERN; VRVS room SUBWAY

Slides

Agenda

Morning (09:00 - 12:30)

  • Understanding Disk - Disk and Disk - Tape Results (Maarten)
  • Why is it so hard to setup basic services? (Gavin)
  • What features are missing in core services that are required for operations? (James)
  • Moving from here to full production services and data rates (based on experiment and DTEAM challenges/tests) (Harry)

Each Tier1 should prepare a few slides addressing specific issues regarding:

  1. Problems seen during the disk-disk and disk-tape transfers and steps taken/planned to address them
  2. Problems seen in implementing the agreed services, including a timeline
  3. Problems encountered in the gLite 3.0 upgrade (maybe this has been covered to death elsewhere...)
  4. Features seen as missing in core services / middleware required for operations

We will appoint a rapporteur to cover each topic, using the slides provided by the Tier1s and Tier0.

Afternoon (14:00 - )

Production Activities and Requirements by Experiment

  • ATLAS - Dario Barberis(?)
  • CMS - Ian Fisk
  • ALICE - Patricia Mendez, Latchezar Betev
  • LHCb - Umberto Marconi

Specifically, each experiment should address:

  • What they want to achieve over the next few months with details of the specific tests and production runs.
  • Specific actions, timeline, sites involved.
  • If they have had bad experiences with specific sites then this should be discussed and resolved.

Wrap-up / AOB

Original Announcement

Its been quite a while since we have had a dedicated technical meeting on Service Challenge issues (where we "get down and dirty").

One possibility would be to hold one at CERN on Wednesday June 21st - immediately following the joint OSG/EGEE operations workshop.

(There is also the Tier2 event the week before, but from the people registered there would be rather little overlap).

There is clearly a lot to discuss - as a first draft I would propose the following:

  • Detailed technical discussions of the throughput results in April and continuing in May, both disk-disk and disk-tape, including the resolution of rate problems following actions over the coming weeks (1/2 day)
  • Discussion and timetable for adding "controlled complexity" to approach progressively the real conditions that we will have to support during LHC data taking (1/2 day)

This latter would include say:

  • disk-disk re-runs, starting as for "SC4", hopefully ramping up to full nominal rates within a few hours;
  • then adding write streams at CERN, as was done over the last night of the disk-disk transfers;
  • Once these are understood and working smoothly, move to fill / transfer / clean (when?) of the Tier0 output buffers and then fill (from network) / archive / clean Tier1 buffers etc.

Whilst there are also quite a few operational issues to discuss too, these should surely be held during the - ahem - operations workshop.

(My view on the latter is that, with June - September as the "pilot WLCG" service phase, this is also our opportunity to shake-down any operational issues, moving to full production operations from October in conjunction with the production WLCG service).

As I happen to have the IT amphitheatre booked for that day, VRVS would be available. However, I think that it would be really good to get the main technical contacts from all Tier1 sites together for this purpose.