[Neglected to ask Richard about hsi action item, perhaps he could follow
up with the status? I think I missed at least one name of those at SLAC
too]
ATLAS SCCS Planning 17Jan2007
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Stephen, Booker, Charlie, Chuck, Gordon, Wei, Richard
Agenda:
1. DQ2 Status/Web Proxy
Running fine but BNL had problems. They cannot get the data out of
Tier-2s. The current rate is very low and data is accumulating. Will
run out of space soon. Alexei wants to do a test of replicating
data to SLAC. Not sure which space we have for the 10% (which would
be about 400GB). Things that we produce data at 200GB/day. Could we
provide the space for the test by stopping production if that is
needed? Could also remove the 700GB of data is that we could move
that elsewhere. Have quite a lot of space on one of the work.
2. Tier-2 Hardware
One of the next things on the list to get ordered. Also need to
find power for it. Evidence from the last week is that we are
fairly certain we can find power but not sure exactly where it
will be till it arrives. Did have the experience of a breaker
blowing due to a mistake of not taking account of dual power
supplies failing over from one site to the other. Looks like the
logic of where we connect things has not been shown to be wrong but
it is close to the limit. Work is going on to install more power.
Actively working on unplugging and unracking the oldest servers to
save space and power. Will probably turn off the shire
machines. Next are the toris and nomas, would still run them if had
the power and cooling.
Also still looking for external solutions, like Sun's Black Box
(200kW of stuff in standard racks in a shipping container). Also
looking at water cooled racks. It has 2 inch water pipes and normal
voltage. THere will be 6509 Cisco switch in the container with a
10G connect in. Racks are front-to-back. The heat exchangers are
right up to the top of the racks so safety is a bit worried. The
cost isn't very different than installations in building 50. Would
only put CPUs there not storage. It is not space designed for
people. Will need to write a plan for access to it. Sun are making
four for their Beta program and we can only have one. Have also
officially accepted the low-bid for water cooled racks in the
computing building but not sure how long it will take to get them
in. The Sun Black Boxes should start to become available in
July. This would be a drain on Substation 7 which has 3MW
spare. What is more critical is that it would be a drain on the
chilled water. We cannot tap into the pipes as they go into the
building (next to where the container would be) but after the water
has gone up to the fourth floor and come back down to the first
floor where existing access is there (which was there for adding
auxiliary chilled water, not getting it out). A lot of worry about
the thin water pipes underground. Should talk about options for if
the pipes fail (rent a chiller or use fires hoses?).
Need to renew quotes as they existing ones are quite old. Might
align more with the Kavli purchase which could save some money. Two
weeks delivery is what is quoted just now. Not sure if this can be
sole source or not.
3. HPSS/gsiftp/xrootd
Agreed that gsiftp is the easier way to go. So should try to get
that working to see if we really need to go to a full SRM
solution. With this solution we can put all the xrootd servers
behind the firewall. Would introduce a single point of failure
though. We need to transfer 1TB/day and a machine should be able to
handle that.
Machines 18,19,20 have been used to setup a test xrootd
cluster. Need to write some code to get the gsiftp module
working. Expect it will be weeks before we can test this. Booker is
working on this with Andy's help. May have a better idea of how
much work is involved.
4. AOB
- Networking Meetings
Not sure we would have heard about them if Charlie hadn't forwarded
the mail.
- wLCG Meeting at CERN Next Week
http://indico.cern.ch/conferenceDisplay.py?confId=3738
Action Items:
-------------
070117 Stephen Email Wei archived dataset at SLAC for moving
061213 Booker Talk to AndyH (and others) about moving SRM interface
forward
070103 Can plug in a gsiftp module for HPSS 6.2, DPM and
dCache. Wei told AndyH but sure if it is useful. We are
running HPSS 5.1 though.
070110 Will have a meeting tomorrow afternoon
070117 Done.
061108 Richard Discuss with SLAC Security longterm approach to ATLAS VO
061115 No information.
061213 Nothing happened yet.
070103 No information.
070110 Richard & BobC in Denver, Stephen will email them.
|