Dear All,
Please find attached the weekly GridPP Project Management
Board Meeting minutes. The latest minutes can be found each week in:
http://www.gridpp.ac.uk/pmb/minutes/this-week.txt
(as well as being listed with other minutes e.g. 020612.txt).
Cheers, Tony
________________________________________________________________________
Tony Doyle, GridPP Project Leader Telephone: +44-141-330 5899
Rm 482, Kelvin Building Telefax: +44-141-330 5881
Dept of Physics and Astronomy EMail: [log in to unmask]
University of Glasgow Web: http://ppewww.ph.gla.ac.uk/~doyle
G12 8QQ, UK Video - IP: 194.36.1.234
________________________________________________________________________
GridPP PMB Minutes 35 - 12th June 2002
======================================
"Monthly Meeting" held on 12th June 2002 at UCL.
Present: Dave Britton, Nick Brook, Tony Cass, Peter Clarke, Tony Doyle,
Neil Geddes, John Gordon, Steve Lloyd,. Robin Middleton
Apologies received from Roger Barlow.
Recent minutes from OC had been circulated. These were referred to several
times during the meeting. Actions from these minutes are denoted by OCn
below.
1. METRICS FOR SUCCESS
OC3 Produce document showing metrics for success of the project within a month.
DB described structure of his project map which pointed to existing
documents from experiments and defined new metrics for some areas.
Attribute map is colour-coded according to status of metrics. This gives
an overview of the project. Some input to it will come automatically from
existing mechanisms (eg TB) but others will require DB to seek their
status.
NEW ACTION 35.1 everyone to finalise their sections of the map by
Friday. This should produce a self- credible version but it can still be
changed before the next OC meeting in the light of wider consultation.
1.1 Specific sections
CERN looks OK - main deliverable is to prepare list of deliverables.
EDG already lists delivery of staff effort Needs EDG deliverables which
are UK responsibility.
Applications -complete enough for now but more will be added over the
summer as experiment documents are revised.
Infrastructure - refine metrics
Interoperability - needs filling out by TD/PC. GridPP interaction with
DataTAG doesn’t merit metrics. What about Globus? Is it important enough
to appear at the top level, or does it appear under standards?
Dissemination - SL will define list and monitor it. He will need input
from others (e.g. talks to other organisation)
Resources - DB list of metrics look reasonable.
2. OTHER OVERSIGHT COMMITTEE MATTERS
OC4 Future resource reports should include details of resources deployed
by/for GridPP in addition to the £17M e-Science funding.
Two areas were considered: hardware and staff effort
Hardware -
NEW ACTION 35.2 JG to value additional kit used at RAL and other
sites (eg JREI, JIF) but only when it is connected to the Grid) BaBar
equipment can be counted now.
Staff -
NEW ACTION 35.3 DB to include additional effort in resource reporting.
Plus management cost of people not paid by GridPP.
OC5 Continue to document deliverables from the project but extend approach
to cover all areas of the project not just the experiments. The
deliverables should be numbered and clearly associated with appropriate
parts of the Gantt charts.
OC6 The project should finalise its planning for Tier1A, stable test beds,
middleware and dissemination; to be reported at the next review.
OC liked the rollout plan and would like similar documents in the other
areas. The project map has defined much of this since the OC.
OC7 Identify risks and issues associated with engaging the system
administrator community. Subsequently, engage with this community to gain
‘buy-in’ for deployment of Grid infrastructure. Comment on initial
findings to be made at next Oversight Committee (OSC) meeting.
HEPSYSMAN deals with this. It is addressed via dissemination but we also
need to keep sysadmins informed/committed. Site visits will help. DB
explained his proposal to give all sites a PC equipped to act as a minimal
grid node. This could be a useful way to include the sites which did not
have the experience or interest to participate. It should be left until
the interested sites are involved.
NEW ACTION 35.4 TD to contact sites which do not yet have a "green dot"
in order to determine what problems/limitations exist and discuss their
plans for their involvement/integration.
OC8 The OSC recommends that the open source policy be more clearly stated
in the project vision statement.
Need a web page to state that all GridPP software is open source and
propose licensing through EDG licence. Do we understand the EDG licence
implications for individual recognition?
NEW ACTION 35.5 SL Web page on GridPP software licence position
3. PROJECT MANAGER ROLE
OC9 The OSC recommends that GridPP give dedicated additional effort to the
mechanics of the project management.
GridPP Project Manager role was discussed. List of duties specifies people
who should be responsible for providing information to the PM (with a
backup). Their formal roles should be updated to include this
responsibility. PM duties include those of existing Resource Manager.
Group duties under 4 primary roles:- setup tools, gather information,
manage and report.
NEW ACTION 35.6 TD to update draft document describing PM role.
NEW ACTION 35.7 PMB to update other role descriptions.
Resources to fund PM are those released by SL not taking funding from
GridPP. TD proposed that DB was appointed to this role and the PMB
approved. Exact funding to be agreed between PPARC and ICSTM.
This will result in DB stepping down from EB.
OC10 Executive summary should only give guidance to OSC on critical issues
over the reporting period to focus the Committee on key issues to be
covered during the OSC review meeting.
TD will take this into account when preparing summary for next OC.
OC11 Daron Green to meet with Tony Doyle in one month once a metrics
document is complete to discuss future reporting. Recommendations from
this meeting to be documented.
The meeting happened and a summary was circulated. NG insisted that formal
recommendations be documented so that questions were not reopened at the
next OC
NEW ACTION 35.8 TD to document recommendations from meeting with Daron
Green
4. EXPERIMENTS BOARD (NB)
Next meeting planned for 18th June with TB.
The following issues will be discussed then:-
Tape access at RAL. CMS are questioning whether the existing interface is
sufficient. Use of Tier1/A by other experiments. Disk Allocations -
proposal to be discussed
EB also expressed an interest in the next round of procurement.
EB should be reviewing quarterly reports from experiments but they have
not all been delivered yet. There was a discussion on whether this should
be done before meeting or at the meeting by all participants. This will be
reviewed after this next meeting. Successful LHCb/Atlas meeting at
Coseners House in May sponsored by GridPP. It drew attendees from Europe
and USA and was a good example of GridPP fostering wider collaboration.
5. CERN (TC)
LCG draft High Level Plan is in circulation for comments. Should be
finalised by July. Structure within this is of different levels of
detail. Applications section is well-defined. June workshop on persistency
shows that real work has started post-RTAG.
Tier Centre RTAG has reported but results don’t help GridPP define a Tier2
centre. GridPP should make its own working definition using this document
as input. Detector Description RTAG has been set up.
6. DISSEMINATION & COLLABORATION BOARD(SL)
No success in fixing a CB meeting in May/June. SL circulated status report
to CB. No negative comments so assume that CB is relatively happy.
Next news item will be on the recent OGSA workshop at ANL, followed by
TierA/1. We have fallen down on giving EDG information for their
dissemination. The recent NeSC meeting is an example. We should be
providing copy for the EDG newsletter.
7. PROCUREMENT (JG)
Evaluation document has been written about last year’s procurement.
Conclusions were that it was successful and produced good kit but there
were still lessons to be learned for next time. A timescale has been
proposed for the next procurement. If T0 is 1st July 2002 then delivery
of equipment would be in January 2003, paying the bill in the second half
of March. NG reported that there might be cash flow problems for the
project this year. Deferring hardware purchase until the next financial
year might be one solution. NG should confirm exact figures in July
(Action 35.9) so that DB can consider possible impact.
NEW ACTION 35.9 NG to confirm exact cash figures for this year
The timetable proposed could easily be slipped so that payment would be
next financial year. With this proviso, PMB approved the start of the
process for the second year’s procurement.
NEW ACTION 35.10 JG to form HAG2
JG proposed to keep the Hardware Advisory Group the same if the same
people were willing to continue
8. ROLLOUT DOCUMENT (PC)
PC asked for weekly meetings of sites to discuss the rollout. The aim
being to give people a chance to ask questions. An appropriate milestone
to start such meetings might be the start of the Testbed1.2 rollout
although there were arguments to start meetings sooner than this. There
was a general agreement on the need for such meetings as a general benefit
to the project.
NEW ACTION 35.11 JG to set up meetings on UK rollout
9. MEETING WITH CORE E-SCIENCE PROGRAMME’S ENGINEERING TASK FORCE
(ETF)
ETF were impressed by what GridPP had achieved so far. They will look at
taking some of our technology such as gridmapdir, RB, slashgrid, spitfire.
VO LDAP server.
10. EDG TB1.2 (JG)
There are several outstanding issues holding up EDG Testbed1.2 (TB1.2)
a) A bug in globus gass_cache is causing ResourceBroker problems under
load. Condor are working on a fix. A workround is being used.
b) MDS instability. Sites dropping ouf of the testbed cause the whole MDS
hierarchy to fail. WP3 are investigating.
c) RB stability. The RB is failing too frequently. WP1 are debugging.
It looks likely that progress can be made soon on all three so TB1.2 could
be released within a couple of weeks.
11. TIERA/1 (JG)
To clarify an issue that was not well-minuted at the meeting 31. The new
cpus have been absorbed into the production CSF farm where their use by
BaBar/LHC can be controlled by queue priorities. For the next three months
all experiments will have access so that they get a good test.
Restrictions limiting access to grid work will be introduced later this
summer. If there are requirements for substantial cpu resources on the EDG
testbed then machines will be moved to it.
The first of the new disks have been made available to BaBar and CMS for
testing. Further allocations will wait for plans from EB.
The queues on CSF are being changed so that a queue for BaBar and LHC has
priority on the new kit but other work will run there when this queue is
empty.
12. NEXT STEPS
Next meeting v/c Monday 24th.
Next physical meeting July 15th at UCL
Next OC meeting scheduled for 1 Oct but DB will be out of contact for 10
days before this. This implies that input will be required by 5th
Collaboration Meeting at IC in September. A physical meeting around then
would be useful.
Next but one physical meeting Sept 11th at RAL
13. REPORTS ON ACTIONS
ACTION 17.14 : JCG revise Tier 1/A staffing plans
ongoing
ACTION 17.19 : DB process spreadsheet from PC and use in reporting.
done
ACTION 29.3 : TD to e-mail PIs with info on experiments reporting mechanism
WP8 at UCL was the only outstanding one. This has now been done.
ACTION 33.1 : ALL to provide input to draft Metrics for Success document.
Done
ACTION 33.2 : DB to edit/author Metrics for Success document, based on PMB
input.
Should be complete by Friday
ACTION 33.3 : TD to define position/role of Project Manager.
Done
ACTION 33.4 : TD to appoint Project Manager.
Done - this meeting appointed DB
ACTION 33.5 : TD to arrange a meeting with OC Chair (Daron Green).
Done
ACTION 33.6 : RM to define and implement budget codes for this FY.
Ongoing
ACTION 34.1 : SL to report on ATLAS/LHCb mtg. as a news item
Done
14. OUTSTANDING ACTIONS
ACTION 17.14 : JCG revise Tier 1/A staffing plans
ACTION 33.2 : DB to edit/author Metrics for Success document, based on PMB
input.
ACTION 33.6 RM to define and implement budget codes for this FY
ACTION 35.1 Everyone to finalise their sections of the map by Friday
ACTION 35.2 JG to value additional kit used at RAL and other sites
ACTION 35.3 DB to include additional effort in resource reporting.
ACTION 35.4 TD to contact sites which do not yet have a "green dot"
ACTION 35.5 SL Web page on GridPP software licence position
ACTION 35.6 TD to update draft document describing PM role
ACTION 35.7 TD to update other role descriptions
ACTION 35.8 TD to document recommendations from meeting with Daron Green
ACTION 35.9 NG to confirm exact cash figures for this year
ACTION 35.10 JG to form HAG2
ACTION 35.11 JG to set up meetings on UK rollout
|
