ATLAS SCCS Planning 16May2007
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Randy, Charlie, Wei, Chuck, JohnB, Richard, Booker, Andy
Agenda:
1. DQ2 Status/Web Proxy
No issues this week, has been working quite well. There was a quite
a good comment about our xrootd setup from Kaushik.
For a user area to be controlled by DQ2 need to have a difference
instance of DQ2 as the retention times are different. The other
issue is that you need to put the user instance under the
production instance file system wise.
2. Tier-2 Hardware
Last status Storage is online and cooling racks are on schedule for
CPUs. Last week an issue come up which has probably delayed the
racks by a week. Somethings to do with the control software for
them and also perhaps a water leak. The ATLAS machines are being
unboxed. The rails are being installed in the racks.
Will take John a week to get everything into the racks and all the
connections. Then need a week to get the software installed and
configured. Two weeks from when commissioning starts is
optimistic. Would very like to have this stuff ready and producing
output by 1st June, but might be tight. Shouldn't miss that by very
much.
Need to start thinking about next FY purchases. Do we have the
infrastructure planned? There is a task force being constructed by
the directorate (including David MacFarlane and folk from CEF among
others and people from the user community). This group should
collate the requirements for equipment from all areas. This is
working urgently as at the beginning of the next FY we will have
nothing new as have started doing nothing. Have observed you don't
get any new cooling or power in less than six months but generally
nearer two years. The Thumpers use about a quarter of the power per
terabyte as the existing BaBar storage, so could spend money to
liberate power. Some prioritisation has to take place as don't
believe there is a way to provide everything needed as there are
not enough dollars (to provide cooling and reduce power usage with
replacing old equipment) or time (to do the engineering). Don't
know if the next FY Tier-2 money has been timetabled yet. Expect it
will be quad-core purchases in the next round. The ATLAS purchase
for next year is in the planning (with a date that Richard can't
remember right now). May need to discuss when it is needed if we
should retire the ATLAS Trigger Cluster to provide more power.
3. xrootd/srm
BNL is working on SRM, uses LBL version which is quite
complicated. The FNAL SRM has been taken off the website. The only
other one available is StoRM. The Italians are eager to do the
integration. It is a question to get someone to manage it at
INFN. Trying to encourage Fabrizio to do that but he is working on
other things. The Italians recognise they will need this as they
want to use it in their Tier-2s also. They have already integrated
with with GTFS.
Perhaps xrootd development world needs more structure to get people
to work to timescales. We had that with BaBar and it worked well,
but now there are many more clients spread much further.
Need someone to come forward to lead this. Andy probably doesn't
have the time.
People are worried about not having a clear way of getting things
done with xrootd. Would be happier when it is clear that the burden
of support wouldn't get dumped on some other lab or
experiment. Should we try to setup a collaboration for xrootd
support? Should have another meeting to discuss this.
4. AFS Problems causing jobs to intermittently fail
Nothing happened in last week.
5. AOB
- Tier-2/3 Meeting in Indiana
Travel budget is very tight so not sure if we can send Wei to
it. There is a travel cap which hasn't changed every though we've
joined ATLAS and need much more travel. Don't believe we can spend
the Tier-2 money on travel, partly due to that. Sending one person
to such a meeting does seems like the minimum we should do. If SLAC
can't fund it we have a large amount of problems.
- Tier-2 workshop
Agreed to hold a workshop at SLAC. Need to figure out when a good
time would be.
- Americas Physics Workshop 20th-23rd August 2007
First all-Americas workshop. May be a need for local accounts for
attendees.
Action Items:
-------------
070516 Andy Setup meeting about xrootd collaboration
070509 Stephen Split up top-level AFS volume, requests to unix-admin
070516 Not done yet.
070502 Stephen Arrange meeting about ATLAS TAG data on PetaCache
070509 Not done yet.
070516 Not done yet. Did talk a little to David Quarrie.
070502 Wei Check CA certificate update mechanism
070509 Not done yet but believe VDT is the right way.
070516 Need to talk to Booked and security team to verify.
|