dCache Tier I meeting June 14, 2011

[part of a series of meetings]

Present

dCache.org(Tigran, Paul, Tanja), IN2P3(), Sara(), Triumf(), BNL(), NDGF(Gerd), PIC(Gerard), GridKa(), Fermi(), CERN()

Agenda

(see box on the other side)

Site reports

NDGF

Gerd reported that NDGF have upgraded their PostgreSQL instance from v8.4 to v9.0. The upgrade went quite well and took 1.5 hours.

NDGF have updated all their head-nodes to use the FHS-compliant version of the packages.

Things are working fine.

PIC

Gerard reported that everything is running fine. PIC has 1.9.12-3 running in production for their Tier-3 facility and the latest 1.9.5-36 running in their Tier-1 facility.

Both instances are running fine and without any issues.

Migrate VO between two instances

Gerard wishes to migrate support for a VO from their Tier-1 facility to their Tier-3 facility. Is this possible?

Yes. There are four steps:

1. create the Chimera SQL needed to create the namespace in Chimera using pnfsDump.
2. inject the SQL into Chimera namespace
3. move pools containing the data from Tier-1 facility to Tier-3 facility
4. do pnfs register on each of the newly moved pools.

Is it possible to move the namespace into a different directory?

Yes. The pnfsDump tool takes two options: one is the PNFS directory to start in (expressed as a PNFS-ID); another option is the (corresponding) Chimera directory (also as an ID). The namespace is copied from PNFS to Chimera so that each file's path, relative to the PNFS start, is that file's path, relative to the Chimera directory.

GridKa

Doris reported that everything is fine so-far.

GridKa have upgraded their CMS instance to 1.9.5-36. Everything is fine so far.

A similar upgrade for the ATLAS instance is planned.

Support tickets for discussion

[Items are added here automagically]

DTNM

Proposed: same time, next week.