Minutes of the SRM v2.2 management meeting
==========================================
Date: 12 November 2007
Attendees: Flavia Donno (FD), Gene Oleynik (GO), Harry Renshall (HR)
Minutes produced by FD
- Status of the implementations
dCache: FD reported that the only 2 remaining open issues with dCache are
listed in the implementation issues twiki pages and they are not
critical. GO requested those problems to be reported to
support@dcache.org: we should move to a production mode of operations
where the GSSD twiki page is no longer used for problem reporting.
FD and HR agree: GSSD will organize support for storage management,
organize and identify second level support, and define the channels and
the procedures to be followed.
CASTOR: FD reported that ew issues were identified by LHCb using high-level
tools. The issues have being reported as savannah bugs and are being
addressed by Shaun De Witt. The fix will be available with CASTOR
SRM v 1.3.1.
StoRM: FD: The remaining SRM v2.2 issues will be cured with release 1.4.1 that
will come out before Christmas. LHCb discovered a few bugs during their
tests. The fix has being provided and tests will be defined during the
workshop in Edinburgh.
DPM: FD: Nothing to report.
- Status of the sites
CASTOR: FD: The CASTOR team at CERN is setting up an SRM v2.2 production endpoint
for LHCb. A few not yet understood problems showed up. They are due most
probably to the installation with Quattor. Problems are being
investigated. However, the endpoint is now operational and has been
released to LHCb.
CASTOR@CNAF has been working OK during the LHCb exercises. However, today
the endpoint has problems that are being investigated. The endpoint does
not pass the SAM tests because they advertise it as well as an SRM v1
endpoint in the production information system. This might cause problems
during experiment tests.
The test endpoint at RAL has also been released to LHCb for testing. It
is being validated now by FD.
dCache: FD: The SRM v2.2 endpoints are not in good shape. NDGF and BNL do not pass S2
tests. Most of the endpoints are not correctly published in the BDII.
This prevents the experiments from using them.
NDGF has upgraded to dCache 1.8.0-1 in production with space management
turned off.
FZK has upgraded to dCAche 1.8.0-1 in production with space management
turned on. So far the upgrade went smoothly for all sites. The sites
are operational.
DPM: The only site available for experiment tests, LAL, does not pass the
BDII and high-level tools tests. Michel Jouvain said that he does not
have time to look into the setup till December 2007. ATLAS would like
to test a DPM SRM v2.2 endpoint. Therefore, if not LAL, some other site
should be identified.
StoRM: The endpoint is not correctly published.
- Status of experiment tests
LHCb: The tests are ongoing following the schedule. High-level tools are now
being tested. The integration of SRM v2.2 in the DIRAC framework will
be completed at the beginning of December.
ATLAS: ATLAS presented a test detailed plan. Last week they have completed the
integration of their DM framework with SRM v2.2.
Tests will resume this coming week.
- High-level tools
FD asked GO if gfal/lcg-utils will be installed on the OSG Worker Nodes. GO
answered that Alain Roy is looking on how to distribute these tools through
VDT. Ruth is pushing to have just one version of the high-level tools
in OSG. It is not sure if this is feaseable. gfal/lcg-utils are being evaluated
by OSG and compared to the srm* clients part of the dCache distribution.
GO will let us know the result of these activities and the official position
of OSG in terms of high-level tools.
Next meeting: 19/20 are an LHCC review so the next meeting is proposed for 22 Nov at 15.30.