Minutes 19 October 06

• Phone:
  • TRIUMF: Denise
  • GridKA: Andreas
  • FNAL: Eric, Lee, Dave
  • ATLAS: Shasha
  • CNAF: Barbara
• CERN:
  • 3D: Dirk, Zbigniew, Luis, Eva
  • ATLAS: Gancho, Florbela
  • CORAL: Ioannis,Radovan
  • Oxford: Stefan
  • AMGA: Birger

• Tier1 site status:
  • Triumf (Denise): running performance tests on the SATA disks, hw mirroring, ASM.
  • CNAF (Barbara): working on the LFC LHCb work plan
  • GridKA (Andreas): I/O problem observed: one node is very busy. Doris has sent an email to Dirk. We will need an account on the database and contact the dbas for help.

• TCP tuning for Streams (Dirk D.):
  • Slides uploaded to the agenda
  • Motivation: slow transfers to BNL and Taiwan due to the network latency
  • Luca has run tests on tuning the application but they are not valid for Streams
  • Proposal: increase the TCP buffer size, apply this change consistently to all the sources and destinations using Streams
  • Florbela has run some tests using Oracle Net parameters
  • Eva will change the TCP buffer size on INTR and Floberla will repeat the tests changing the Oracle Net buffer size.

• ATLAS COOL tests:
  • Capture process on integration got stuck. It was recreated but without using the last SCN already applied at the destinations so now there is a gap on the data
  • Eva propose to create a new capture process to re-capture the changes in the gap, it is necessary to determine the SCNs to start and stop the new capture
  • Wait event observed on the apply site: it is related to poor performance on RAC environments. To be investigated!
  • Problem with OEM, mail already sent to Dirk G.

• CMS Frontier tests:
  • Segmentation fault problem: correlated to the second thread, Seal sw
  • Workaround: disable second thread. This will be better than request a bug fix on the Seal sw (it will take longer)
  • There is another issue logged in Savannah
  • Basically all Tier2 sites are already in the test