Last modified 7 years ago Last modified on 10/23/14 20:06:45

dCache Tier I meeting October 23, 2014

[part of a series of meetings]

Present, Paul), IN2P3(), Sara(), Triumf(), BNL(), NDGF(), PIC(), KIT(), Fermi(Natalia), CERN()


(see box on the other side)

Site reports


Natalia gave a further update on the clumping problem. Catalin have checked and the problem coincides with when CMS files are being replicated within the dCache instances. Such replication is done for files where it is known that they will become popular in the future (e.g., configuration files that all jobs will access).

Natalia mentioned that internal retries within the xrootd door when the clumping limit has been reached may not be desirable. This is because the xrootd client may be able to get the file from elsewhere.

Fermi are undertaking lots of development work with their test-bed. It isn't clear whether they will have a single test-bed (shared between 2.2 deployment testing and 2.10 upgrade testing) or to have separate test-beds.

There was a discussion about puppet. DESY currently does not configure dCache via puppet; but this decision is largely based on DESY-local reasons.

Support tickets for discussion

[Items are added here automagically]


Same time, next week.