ATLAS SCCS Planning 22Aug2007
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Wei, Stephen, Chuck, Len, Randy, Richard, Booker
Agenda:
1. DQ2 Status/Web Proxy
Problems today with the PANDA server today so probably is probably
poor. A lot of problems while BNL was updating their dCache. They
have unexpected hardware problems. So production has been down
quite a lot recently. Some worries about the quality of the DQ2
software, doesn't seem to be production quality even after two
months of deployment. The recent functional tests have worse
results than the last try last year.
2. Tier-2 Hardware
Stephen posted earlier about some measurements with the latest
Intel and AMD CPUs. The basic result was that the Intel GHz were
worth 10% more than the AMD ones. AMD was less effected by loading
up a box completely (the 10% is for a fully loaded box, it was
nearer 15% with only one job per box). BaBar saw more-or-less
parity between them. For acquisitions therefore might assume more
like 5%. Will not know how the prices compare until the 4-core AMD
chips come out.
Have for the last year being doing 1gig ether net on the batch
machines, this is really needed with the large number of cores per
box. Might even need to bind together the two on board ones for the
Dell boxes being tested, having four would have left more headroom
but have only two.
3. AOB
- Next weeks meeting
Stephen will be travelling, should others meet? Wei also away so will
cancel next week.
- OSG Counsel Meeting
Noticeable that SLAC doesn't get any statistics in to the Gratia
accounting system. There is a large downside for the OSG that we
don't do this. Need to get this working. There are some technical
issues with assumptions made (might be an assumption that there
is only one queue used, takes all info from that queue and a
single user).
OSG needs some amount of political help. They've been funded for
one year for $6M (half DoE, half NSF), NSF has pulled out $1M
from a second year. Some folk having been hearing this grid stuff
will disappear in a year or so and we'll be back at TeraGrid
(which means large NSF facilities). We should make sure that OSG
can report truthfully the large amount of work being done.
Some milestones were written down that don't actually reflect
work being done. OSG needs to make sure they have the right set
of milestones.
This that more than half of OSG is reporting via Gratia. So it is
useful to use it for the next year till it gets
replaced. Fermilab is responsible for Gratia just now, initially
it was developed with PPDG money prior to OSG.
- Power work
It will happen on Wednesday next week, all day. There will be
some service outage.
There is a new outages planning mailing list. Charlie is on it.
There will be a whole building outage not before next March, this
will be for an entire day.
Action Items:
-------------
070815 Wei Thing about how we maintain lists of local people etc
070822 Should create another LSF group. The current one is for
all Tier-2 users. These are only for local use. Should
just LSF groups to maintain these lists. Done.
070725 Stephen Try to test eval01
070801 Didn't have access when attempted, Booked fixed that.
Problem with ATLAS software (hopefully trivial).
070815 Installed new software to get around the problem. Not
tested yet. Have order 128 cousins for it, have bought
the machine.
070822 Done.
070711 Stephen Find out about benchmarks for CPUs for next purchase
070718 Not done yet.
070801 Extracted data from our Production Database, need to
analyse it still.
070815 Not done yet.
070822 eval01 test produced results needed just now, so forget
about this for now.
|