LHC Computing Grid Project

Project Execution Board

Notes of the meeting of Tuesday March 2, 2004

DRAFT 1 05/MAR/2004

Present:

Dario Barberis, Ian Bird, Federico Carminati, Philippe Charpentier, David Foster, Frédéric Hemmer, Bob Jones, Jürgen Knobloch (chair, secretary), Massimo Lamanna, Alberto Masoni, Bernd Panzer, David Stickland, Torre Wenaus

Phone: Mirco Mazzucato, Nick Brook

David Malon, RD Schaffer, Jamie Shiers invited for the discussion on MySQL

Actions: Actions are identified by bold blue italics.

In the absence of Les, the meeting was chaired by Jürgen. Jürgen welcomed David Malon and RD Schaffer from ATLAS and Jamie Shiers for the discussion on MySQL support for ATLAS.

Minutes of last meeting and matters arising

The minutes of the meeting of the 24^th February were accepted.

MySQL support

Bernd presented a first response to the ATLAS MySQL request. The request was first presented by Gilbert in a note and more recently in a set of transparencies from David Malon. The FIO group in IT proposes to use disk servers out of the ATLAS allocation for the requested MySQL service with well defined operational procedures and responsibilities.

David Malon said that this was an acceptable starting point sufficient for the 2004 data challenges. Jamie said that he would have expected that a higher level of redundancy and availability would be required for such a service. Bernd added that the manpower required for a 24 hour/7 days service would be at least 2 FTEs. David Malon replied that ATLAS could live with the proposed service level - accepting occasional downtime and reboot of the servers.

David added that the slides address the question of why ATLAS needs MySQL altogether (instead of moving completely over to Oracle). He argued that in the context of distributed database deployment some level of MySQL is definitely required and that if a homogeneous solution for ATLAS were required it would be MySQL today. He added that the development of distributed databases was an excellent opportunity for remote sites to participate. Mirco said that Italian sites would also like to see MySQL supported because Oracle was too heavy. Federico and others made the point that the LCG software should not be tied to a specific database implementation.

In the conclusion it was said that for the current data challenge the proposal of Bernd was acceptable to ATLAS and is considered by David Malon as a starting point. The discussions on the scope and resource requirements of a wider service will continue. Bernd added that the experiments should talk to each other to avoid multiple solutions to be supported and maintained at CERN.

Level-1 milestones

Jürgen has updated the milestone table taking into account the discussion at the last PEB.

The only new discussion was on the applications area milestones. David Stickland remarked that the milestones in AA are more concerned with plans than with deliverables needed by the experiments. Torre agreed with this observation and proposed that milestone Nr. 9 should be broadened “applications data handling software” to “LCG phase 1 application area software”. This being to vague it should be accompanied by level-2 milestones with concrete deliverables and their adoption and integration. A discussion on this will take place in the AF next Thursday, March 4. This may lead to identifying a new level 1 milestone for later in 2004. Torre also said that at least one of the milestones 2&3 – namely Nr. 3 - should be maintained as LHCC level 1 milestone.

The members of the PEB will provide feedback to Jürgen on the milestones by e-mail as soon as possible. The area managers should also verify that the level-2 milestones are compatible with the level-1 milestones and if necessary apply changes or introduce new level-2 milestones.

Responses to reviews (documents)

Jürgen went through the document concerning the LHCC review. Les had extracted the major points raised by the referees and Ian, Les and Torre have provided the responses. The document was endorsed by the PEB. There was some discussion on the first point of “Grid Deployment”. It was clarified that “resources” in regional centres refer to hardware resources (as opposed to human resources). Berd has received to date only 6 responses from regional centres. He expects more answers soon – the deadline is this week. Concerning the complexity of middleware installation David Stickland noted that some of the problems came from the fact that we have effectively gone the route of a single platform/OS version. He emphasized the importance of supporting and running LCG on multiple platforms.

Torre explained the replies to the internal review of the applications area. Jürgen pointed out that at a recent LHCC meeting there was some discussion concerning a common project on conditions data base. Federico raised worries about the convergence with mathematical libraries and duplication of work with ROOT. This will be further discussed when the AA workplan will be presented at the next meeting. Torre will publish his document with the replies and that will conclude the review. The PEB agreed.

Jürgen went through the replies to the internal review on non-application items provided by Ian Frédéric and Les. Philippe remarked that it was unfortunate that so many reviews took place in a short period of time.

The PEB also approved the response to this review and took the occasion to thank the reviewers for their effort. All responses to the reviews will be forwarded to the SC2.

All reviews discussed today were found very useful for the LCG project.

Status of data challenges

Federico reported that the Alice data challenge has started well. An initial problem with the load balancing of the I/O has been overcome – first with Alice’s own solution. Now a general LCG solution is available and Alice will move to that as soon as is reasonable. Alice produces a large number of small files which is considered part of the challenge. Federico was unhappy that previously agreed resource levels are made available only after there is pressure from the experiment. He had expected that the allocated resources would be there without any further discussion. This will be discussed at the GDB next week.

David Stickland said that CMS had managed to get some jobs running. They are still solving problems in their own software. They will soon send reconstruction jobs to the “Global Distribution Buffer” (GDB) to be registered in the RLS and export buffers (SRB and SRM/Dcache) to the Tier-1s. A problem with the metadata string-length has been fixed and will be implemented this Thursday. Altogether, David considered the progress as satisfactory.

A.o.b.

The AA workplan will be discussed in the PEB sufficiently before the SC2 meeting of March 26.

Jürgen mentioned the Les and he proposed to have an agenda item at the last PEB meeting in a quarter to go through the milestones that were due in that quarter and see what the status of achievement was.

It was mentioned that the announcement of the LCG workshop was only at the level of a draft agenda and that it would be too late for people in the US to decide on their coming based on a more detailed agenda.

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	Done
2	16dec03	Understand why the substantial resources in Liverpool are not available for LCG-2. 6jan04- visit to RAL organised for 24jan04	Les	Done
3	16dec03	Confirm that the absence of BNL in the LCG-2 deployment list is due to manpower shortage	Les	Done
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	Done
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 Trento	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