ATLAS SCCS Planning 04Oct2006
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Charlie, Richard, JohnB, Wei, Chuck, Randy, BillL, Len
Agenda:
1. DQ2 Status/Web Proxy
The machine had to be reboot, not sure why but need to
watch. Couldn't access it via the console either.
2. Slots for ATLAS Production jobs and other batch related stuff
When do we expect the new fairshares discussed last week deployed?
Have some questions out to Platform which need answered
first. Chuck and Randy are away next week so it may not happen
before then. Would base the ATLAS share on what slice of the batch
system that will be bought for the Tier2. Will use a lower estimate
just now as it is easier to raise the fraction later than lower
it. Some of the money will also need to spent on a rack, part of a
network switch and console. Will try to get this order into KIPAC.
Still need to decide if we will have 1 or 2GB per core. Today's
jobs only require 1GB but ATLAS requirements are 2GB. It would be a
very unusual to open machines and put more memory in later and
could introduce support issues. The KIPAC machines have 16GB memory
but only one disk. ATLAS jobs might need more disk. These machines
have 16 slots but if you use more than 8 (or was it 4?) they clock
down the rate so everything gets slower.
3. Trigger Farm
Work tomorrow morning should allow enough power to be provided for
the farm to move up from IR2.
4. AOB
- Advisory Board Meeting
Over the last two months an election for the Chair of the Western
Tier 2 Advisory Board. There were an even number of voters and a
there was a tie. Bill Lockman and Gordon Watts were the two folk
who were voted as both winners. There have been some discussions
between the winners and the voters, it was decided to make one
the chair for a year and the other chair-elect. As Bill isn't
available immediately Gordon will be the chair for a year with
Bill taking over afterwards and there will be a new chair-elect
elected at that time.
- Workshop on DQ2 at BNL report by Wei
They had someone there on the development team. They explained
the recent problem they say with the T1-T2 tests. This seems to
be due to a central catalogue. They are thinking of have a
distributed catalogue and getting rid of some of the old LCG
stuff to fix this problem. There are also discussions on how to
package it to make it easier to install. The developers are now
asking for login access at sites. FTS is extremely
immature. There is a need for a fall back mechanism, currently
use dq2_get. Another person described how the various dq2_*
commands work. Also a discussion on disk cleaning at the various
sites. Torre Weanus gave an explanation on how PANDA works, they
main issue at the moment is how to push PANDA out to opportunistic
sites that have free CPU power (but no storage space). xrootd was
mentioned a few times, as there are so many problems with dCache
and SRM it seems they want a backup. Not sure what the timescale
might be for this. We should make sure we are not surprised by
their desires. Could link up Andy with someone at BNL.
- Jamboree
Borrowing disk space from BaBar is progressing, officially put
request into UNIX group to do the installation. Need to know what
type of service wished, current guess is a big glob of disk
served via NFS. Don't expect it will be wished back very soon,
probably have it for two years till the equipment reaches it end
of life. Probably want to integrate into
/afs/slac/g/atlas/work/.. area for the moment. For data transfer
should encourage users to use the dq2 server to avoid
transferring the same data multiple times.
Users will probably log in to yakuts. Norics might be less busy
but yakuts are closed to CERN setup.
Action Items:
-------------
061004 Randy Find out about xrootd for ATLAS plans
061004 Richard Announce election of Chair/Chair-Elect of WT2 AB.
061004 Randy Project number of nodes for purchase for both 1GB/core
and 2GB/core
060927 Stephen Write web page with description of SLAC batch system
061004 Copied BaBar page and started editing, hopefully in
better shape next week;
http://www.slac.stanford.edu/~gowdy/batchDesc.html
060830 Richard Talk to Gregory about "getting" disk
060906 Did try but not managed yet.
060920 Trying to get together tomorrow to talk about it.
060927 Don't know the status.
061004 Done.
060816 Chuck Check with Bob about web server approval need
060823 To be done.
060830 To be done.
060906 Gary believes that it wouldn't need approval but should
check with Bob.
060913 Still need to talk to Bob.
060920 No news.
060927 Nothing yet.
061004 Not done yet.
