Date and Location

  1. July 2005
  2. :00 - 17:30 B28-S-15

Attendees

ASCC: ...

BNL: Bruce Gibbard, Razvan Popescu

FNAL: Yujun wu

FZK: Doris

IN2P3: Lionel

Paris: Michel

INFN: Tiziana

SARA: Ron

NGDF:

TRIUMF: Reda

PIC: Gonzalo

RAL: Andrew Sansum

CERN: Jamie Shiers, James Casey, Patrica Mendez Lorenzo, David, Simone, Roberto, Sophie

DESY: Michael

Expts:

Tape Throughput Tests

IN2P3 - tape tests without FTS due to a problem with dCache. 30MB/s. HPSS is doing fine. Some stability problems with dCache to be understood.

BNL - periods of 24Hr at 60MB/s. Some period were a bit lower. No special issues. 10 tape drives

SARA - Problem with tape robot over the weekend. Running at 45-50MB/s otherwise. 3 drives

DESY - 30-40MB/s continuous. 4 drives

PIC - working in best effort mode right now, since take drives as available. 3 9940B drives - writing at 30MB/s for last 24 hours.

RAL - working at 40-50MB/s right now

INFN - 40MB/s. Up to 4 drives.

TRIUMF - 3-4 drives. Up to 40-50MB/s. Can make 60 with the 4th drive.

Mark: Currently we see 40-45MB/s on 3 9940B drives, where the max achievable bandwidth is 90MB/s (3x30MB/s). What do other sites see?

Michael: We see the same, but it's understood due to data starvation in the data server in front of the tape drive. We see the transmit buffer of the pool buffer is always full, but the drive is idling waiting for the data to be delivered.

Tier-2 tests

Paris
Problems with srmpcp & DPM. Lionel reports some transfers are working when they pull

FNAL
(Yujun) 5 site working using phedex.

RAL
At Lancaster, were running at 11/12 files per hour. Imperial 30-40MB/s. Edinburgh 40MB/s. Have FTS stats for this which we'll distribute.

CNAF
starting tomorrow until Friday or early next week. Testing with 4 sites.

ASCC
Already transferring data using phedex to the Tier-2

Plans for August

Jamie: We need to work now on tuning the system, and try and overcome the problems we've seen during July.

Doris: Will it be all sites, or individually.

James: I think there are different problems for different sites- We'll try and work with several sites in parallel (3 or 4) looking at specific problems. One important thing is

dCache Workshop

Michael: We'd be happy to run the workshop August 30th (noon) unti September 1st. In order to come up with a useful agenda, please could people send their ideas/questions/problems to the dcache-support list.

Preparations for service phase

Jamie: Clearly this assumes all works well in the debugging.

At the PDB, the experiments agreed to provide sample jobs, and also to fill in spreadsheet describing capacity and cpu usage at T0/1/2

AOB

Jamie: At the PEB it was discussed that quarterly technical meetings outside CERN would be useful.
Edit | Attach | Watch | Print version | History: r1 | Backlinks | Raw View | WYSIWYG | More topic actions
Topic revision: r1 - 2005-07-27 - unknown
 
    • Cern Search Icon Cern Search
    • TWiki Search Icon TWiki Search
    • Google Search Icon Google Search

    LCG All webs login

This site is powered by the TWiki collaboration platform Powered by PerlCopyright & 2008-2019 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback