LHC Computing
Grid Project
Project Execution Board
Notes
of the meeting of Thursday February 12, 2004
DRAFT 2 19/FEB/2004
Present:
Nick
Brook, Federico
Carminati, Philippe Charpentier,
Bob Jones, Jürgen Knobloch (secretary), Alberto Masoni,
Bernd Panzer, Gilbert Poulard, David Quarrie (from 9:50),
Phone: Matthias Kasemann, Mirco Mazzucato,
Actions: Actions are identified by bold blue italics.
The minutes of the meeting of the 12th January were accepted. The conclusion on Arda has been superseded by the new proposal of Les.
Mirco has summarized the main conclusions of the last GDB meeting in a document. Federico expressed concerns on the general use of Quattor as installation tool. Zdenek confirmed that a high level of expertise (good knowledge of both LCFng and Quattor) and a lot of work would be required for a full implementation. He also mentioned that certain centres have requested that the installation should not require any of the two systems. Les stated that it was not suggested to have Quattor as generally recommended tool for the installation of the LCG distribution of middleware. At some large centres such as CERN an installation tool was found necessary but the LCG software would not be tied to it. Mirco added that several large centres have expressed interest in a common automatic installation tool. He is looking forward to a collaboration of these centres on common requirements and on providing resources for implementation and support.
Alberto added that at the GDB the need for resource reporting was reiterated. It is expected that for the March GDB, Bernd should receive an update on the resources for 2004 from the regional centres.
Bernd asked what was behind
point 4 in Mirco’s report concerning experiment’s
software in cases where no shared file system was available. Flavia will collect requirements to develop a script which
would mirror the files installed in a single place to all the worker nodes. Further
details should be discussed in the weekly deployment meetings.
Les sent out the second version of his note aiming to launch the project by proposing its scope and style. After agreement on this, a project manager could be appointed who would then produce a workplan rather rapidly.
There was general agreement on the draft formulation. Federico felt that the project had a clear reporting line and would not need to report to the Architects Forum. After some discussion initiated by David Stickland, on where to best locate the project it was agreed to have it not assigned to any specific area but have it at the same level as the areas with its own reporting in the quarterly reports.
As this coordination and integration project has the name Arda, a new name for the middleware needs to be defined. Bob will do this.
The proposal of Les to nominate
During
the discussion other points came up:
Matthias presented transparencies containing the proposed SC2
membership and meeting schedule for 2004. The eight meetings per year will be
split into four meetings evaluating the quarterly reports and four meetings
where each will be dedicated to an in-depth evaluation of a given area. For the
quarterly progress reports it is proposed to go back to the normal definition
of quarter (removing the two-week shift) and to ask for additional
contributions from the experiments and from the regional centres
on their progress. Initial feedback was given on the recent progress report
covering Q4/2003. The full feedback will be given to the POB on February 16.
Zdenek kindly provided the following text for the Grid Deployment:
Every LCG release had a monitoring
system included (the gridICE) and we have enhanced it
for the LCG-2. It has been widely used by GOC. The inclusion of R-GMA as a
monitoring (not info-provider) has always been on our plans, however, due to
other much higher priorities we were unable to work on its integration. If
experiments want/need to use it for monitoring of their production, LCG should
work with them to understand their requirements and be able to adequately test
them to ensure there is no unexpected interaction between R-GMA and LCG-2
middleware. The contacts between relevant people from EDG WP3, WP7 and LCG have
been established long ago and "roadmap" was agreed upon. Briefly, if
the whole R-GMA software can be packaged separately from LCG-2, and deploy with
agreements on sites (our understanding is that some R-GMA software has to go
onto WN to make the monitoring work), this is OK for LCG. The complete
integration into the LCG-2 with full testing and certification will be
scheduled for later time when all experiments problems for running Data
Challenges are solved.
With the common understanding between
Alice, CMS and LCG, experiments will start the DC with "classical
SE", SRM is not available yet. Due to the incoherencies between various
parts of software (Replica Manager, Pool) it is possible that we will have to
go through data management migration because of the structure of data catalogs may need to change. LCG is doing utmost to avoid
that. However, if that appears to be inevitable, LCG will develop necessary
scripts and know-how, and assist experiments help in the conversion. It will
work very closely with affected experiment on the scheduling of such an event
in order to minimize disruptions in experiments activities to the absolute
minimum.
LCG was very happy with the input from
experiments obtained on the now regular Monday's deployment meetings. Some of
the problems were already corrected; we are working on others that need more effort.
Experiments are invited to test al the fixes on LCG EIS testbed
before the corrected software is deployed. Problems found by experiments have
the absolute highest priority for LCG.
In this context it is fair to say LCG is
currently working with the top priority on making the LCG-2 work for Alice and
CMS so they can successfully proceed with scheduled
Data Challenges.
Federico said that
David Stickland
was also pleased with the deployment meetings and felt that things are going
very well. He would like to make sure that the people available to actively
help
Nick said that LHCb is performing basic tests, starting this week stress tests of LCG-2 with Dirac. A concern is the support for RLS and mass storage in outside centres that do not have Castor. Zdenek replied that dCache is now being installed at CERN to try it out with LCG-2. The SRM integration is seen as a solution for Tier-2 centres.
On a question of Federico, Bernd reminded that the Castor disk pool manager is not supported for outside centres.
Gilbert reported that the ongoing tests in Atlas are not yet at a production level. They plan to start the data challenges beginning of May.
Philippe had expressed concern about the
inflation of grid-related meetings and workshops. It is proposed that there be
some control and visibility of such events and that the LCG office maintains a
list of meetings.
Les said that during a MoU
task force meeting the need for updated requirements for Phase 2 in Tier-1 and
Tier-2 centres came up. The last estimate was probably collected in the
Hoffmann report. Someone from each experiment is
needed to come up with an estimate within the next few weeks.
Bob mentioned that a Data Grid project
review would take place next week.
A request from Atlas to support MySQL will be discussed at the next PEB.
Bernd will need to rearrange resource
allocation for this year in order to avoid clashes arising from the delay in
the start of data challenges.
Actions |
||||
# |
Date opened |
Description |
Responsible |
Date closed |
1 |
16dec03 |
ALICE, CMS and LHCb to name someone
responsible for coordinating deployment on LCG-2 |
Federico, David S., Philippe |
|
2 |
16dec03 |
Understand why the substantial resources
in 6jan04- visit to RAL organised for 24jan04 |
Les |
|
3 |
16dec03 |
Confirm that the absence of BNL in the
LCG-2 deployment list is due to manpower shortage |
Les |
|
4 |
16dec03 |
Experiments to request through their
national contacts that their resources in the core LCG-2 centres are
integrated in LCG-2 |
Federico, Dario, David S., Philippe |
|
5 |
16dec03 |
Regional centres to be asked to clarify
their mass storage plans. Presented by RCs
in GDB of 13jan04 |
Les |
13jan04 |
6 |
12jan04 |
Revised proposed GAG mandate |
Federico |
27jan0 |
7 |
27jan04 |
Revised ARDA note |
Les |
12feb04 |
8 |
27jan04 |
Establish a weekly “Deployment
Meeting” |
Ian |
2feb04 |
9 |
27jan04 |
Note on new project proposal from |
Federico |
|
10 |
12feb04 |
Define new name for middleware |
Bob |
|
11 |
12feb04 |
Nominate Arda
contact persons |
Experiments |
|
12 |
12feb04 |
Nominate people for Phase 2 requirements
of the experiments |
Experiments |
|