Last modified 12 years ago Last modified on 11/27/08 14:18:40

dCache deployment phone conference Nov 25, 2008

Participants :,Paul,Dmitry,Timur), In2p3, Sara, Bnl, Triumf, Fermi

  • Follow up in restore discussion between BNL and IN2P3


Everything OK.


Have update to 1.9.0-6. We've noticed SRM performance is OK at 5 Hz, but problems start at 6.3 Hz. The limiting factor seems to be CPU on the SRM.

The problem is no worse than with the previous version: 1.8.0-15p-nn


No pressing issues.

The problem with PNFS mounts going away on pool nodes has not been observed recently. This has occurred in the past. The kernel is stock SL one: 2.6.9-nnn. The cause is currently not understood.

Ron is happy to script some automated (Nagios) tests to check for PNFS not being mounted and will report back if the problem reoccurs.


No problems.


IN2P3 have been experiencing two problems:

First problem is with Bring on-line. This was triggered by LHCb pre-staging tests. The file transfer requests go to state Done, but the status of the pre-staging request remains in Pending and is eventually timed-out.

Second problem is with SRM-ls gives the wrong value for occupancy.

Timur will investigate both issues, but he is waiting for some more information from Lionel before he can investigate further.


PIC has upgraded their SRM node to a 64-bit machine. They are still seeing problems, but the problems are different from before. Suffering SAM PUT test failures, but the other tests succeed.

Paco and Timur will investigate this further.


Have updated to 1.9.1 for almost all of the pools. There is an issue with pool not starting up if files are in an specific inconsistent state. We would recommend delaying upgrading to 1.9.1 until 1.9.1-2 has been released.




Same time, next week.

Last modified by patrick @ Sun Mar 7 00:42:24 2021