ATLAS SCCS Planning 21Mar2007
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Wei, Chuck, JohnB, Stephen, Randy, Gordon, Charlie, Gary,
Richard, Len
Agenda:
1. DQ2 Status/Web Proxy
Running out of space. Recent jobs seem to be more data intensive
than old jobs, as they are reconstruction jobs as they need the
output from the simulation. Hope to get the new space available,
currently are turning off the Fetcher for 12 hours, then on
again. Currently only keeping three days worth of data. Only have
232GBs worth of space free currently. The AOD distribution has used
about 600GB and only a 1 or 2 GB per day so not something that
could be turned off to make a big difference.
Some email discussion about DQ2 0.3. Recently installed 0.2.12, but
not really in production. Paul Nilsson has got things working but
don't believe will put it into production. Want to deploy 0.3 in a
quick timescale. We want to install DQ2 via RPMs instead of
pacman. Will be slightly better to allow patching of the web
server, if not the MySQL (as require version 5 and RHEL4 comes with
MySQL 4). Discussions going on about having central US installation
or site administrators to do it. If it is by folk outside SLAC will
need to get those folks accounts at SLAC. Timescale is still at
least a month before 0.3 will be used.
2. Tier-2 Hardware
In purchasing, not gotten out. Seems to be very slow over there
just now.
Is the memfs server moving out of building 50, so does that free
power for other uses? Power usage needs to be prioritised. The ATLAS
(and GLAST) disk should be able to be powered up by the space freed
up. Can't see where to put the batch nodes till the racks are built
in May. Looks like it will move in a couple of weeks.
Should also discuss if tories and nomas should be removed to
provide power to the newer machines. For the lab it makes sense.
3. xrootd/srm
No news. From the Tier-2 meeting quite sure that people will want
it. SW Tier-2 is looking at xrootd but they want SRM
interface. Hope can get something, or at least decide if it is
doable by the next Tier-2 meeting, that should be sometime in
June.
4. AOB
- Web Site
At the Tier-2 meeting they mentioned they wanted a common
look-and-feel for the Tier-2 web sites. Wanted to raise this at
today's Facilities meeting but it was cancelled, will need to try
it next week. A quite look round would point this towards being a
Wiki page, but which isn't actually Wiki but is static.
The goal of a common technology is probably unrealistic, a common
look-and-feel might be. The page that InfoMedia solutions have done
for us is probably better than what we have. Believe these are
static UNIX web pages (could be static windows ones but isn't a big
difference). New pages look much more professional and more like a
Wiki page to look more like other Tier-2s.
Should be one more round of feedback to InfoMedia before moving all
the content. Will go back to them and discuss adding content.
- Tier-3 Discussion at UCSD Tier-2 meeting
Discussion of network and Tier-3 user support. There was not so
many ATLAS folk but quite a few CMS. There was faculty person from
Santa Cruz. It looks like he doesn't know much about what we do at
the Tier-2. He wanted to know about the best way to use resources,
like having their own mini-farm at Santa Cruz. Also had a similar
request from Andy Langford. While making the Tier-2 proposal also
discussed hosting institution machines at the Tier-2. Probably
won't work given the power issues at SLAC, but would really like to
do it.
At UCSC they don't have any substantial bandwidth to their
Campus. Normally sites would be interested in having one to two
racks of systems. Many sites still have a learning curve on running
more than a few systems.
They did come out with a number of 20TB per user, where they are
looking at $300/TB. May be the 20TB is based on money, we could see
what could be provided for the same money, certainly less.
This meeting was organised by Internet2. This was to try to get the
network people and physicists at Universities together. It did wake
up some of them to start them talking. No concrete followup to the
meeting at the moment beyond circulating some information.
- Miron Livny (CONDOR)
Will be visiting tomorrow.at 10am. He will be at SLAC for the
morning. Hasn't been around SLAC much but has been involved in the
Grid for HEP for many many years. Will be valuable for him to see
how life is in the trenches of the OSG.
Will make sure there are a good number of appropriate people
present.
- CMS Support
For priority for hardware resources ATLAS should get appropriate
response. There is the idea of the Grid where you can easily use
the same resources for different purposes. To allow SLAC to be used
efficiently could support CMS use. CMS currently need batch
visibility of the Internet, which isn't possible at SLAC (although
have heard the opposite). Need to make sure we don't get distracted
by CMS in making sure SLAC is fully efficient for ATLAS. CMS did
run batch jobs at SLAC before ATLAS. At the time they didn't find
it interesting as they did need a large amount of resources. Would
actually like to put CMS lower down the list than other lab
requests (like from KIPAC and GLAST). It is important that SLAC is
seen as an important Grid site for HEP. This could occasionally
attending to a CMS issue prior to KIPAC (as an example). It is the
human cycles which are the important (and scarce) resource.
There is a perception of SLAC not being fully ready. Need to try to
understand this. There are problems with disk space currently but
many sites have this issue. There is the issue of xrootd not being
fully available for ATLAS use via an SRM interface but this is a
bonus and not needed for production use.
- Next Meeting
Stephen will be at the ATLAS Software Week in Munich;
http://indico.cern.ch/conferenceDisplay.py?confId=5060
Will clarify on Tuesday next week if there is a meeting and who
will be chairing it.
Action Items:
-------------
070321 Gordon Discuss perception of SLAC Tier-2 with external folk.
070321 Wei Send example new web page link to list
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.
070124 Don't know the status.
070131 Don't believe this happened.
070207 Have not done this. Randy has talked to Heather, didn't
have any time to comment today but she is aware about
it. Will treat each VO as an enclave, if you are using
anonymous accounts need to be able to show how ran a job
and when. The main issue is that VOs are not legal
entities and anyone can declare themselves a VO. Would
need to actually test that the required information can
actually be found.
070221 No info.
070228 No info.
070314 No info.
070321 No info.
