ATLAS SCCS Planning 17Oct2007
-----------------------------
9am, SCCS Conf Rm A, to call in +1 510 665 5437, press 1, 3935#
Present: Wei, Stephen, Len, Charlie, Booker, JohnB, Chuck, Randy,
Richard
Agenda:
1. DQ2 Status/Web Proxy
PANDA has been switch between DQ2 and PandaMover but things look
okay at SLAC. There have been a lot of updates at BNL, the latest
to DQ2 0.4.0 which seems to have make everything run well again.
Need to work on getting a GSI authenticated PandaMover
working. Were looking for a new machine but could perhaps use the
existing machine as has was quite low CPU usage.
In the future it is better to know the plans ahead of time to get
security involved before software is deployed. Not sure there are
existing communications channels that would allow this, may need to
discuss with Michael or Torre.
2. Tier-2 Hardware
Evaluating three candidates for BaBar, all Intel based. Will almost
certainly want to get the same for ATLAS. For BaBar want to order
mid-November and installed in December. To order together the
money would need to arrive this week so unlikely. However, should
be able to order more similar machines later.
GLAST will probably be ordering on the same timescale as ATLAS so
can put those together. GLAST want their machines by sometime
mid-January, although their launch date has slipped a little.
Almost certainly Intel quad core. AMD have had difficulties getting
theirs out and it looks like GA would be first quarter of 08.
Will be running RHEL4 on the new machines.
3. SSH w/ token passing
With the switch v2 and Kerberos 5 you cannot pass CERN tokens to
SLAC anymore. You could use things like getToken and setToken to
pass it. You don't need to be root to use setToken, not sure about
getToken. These are external commands (not standard AFS).
Normal work method is to acquire tokens for SLAC and CERN on laptop
then you can log on to both CERN and SLAC and use both cells.
No one seems to be complaining though. Beyond the general issue CVS
might be an issue. Booker/Wei will talk to DavidM to see if it all
works okay.
Will also try to think of other issues. Wonder if other labs also
have an issue. Might be worth discussing at HEPiX.
4. Tier2/3 Workshop
Wei forwarded an email about the agenda to everyone from
MichaelE. RobG has setup a page at CERN with the agenda. The
InfoMedia people have sent Wei a page for the meeting. It looks
very nice and Wei has asked them to go ahead. There is an account
to accept registration and payment. This needs to be added to the
web page. They would also like to setup early registration before
14th November. Need to make sure there is enough time for folk to
register. Wei will forward the URL of the InfoMedia page.
5. AOB
- Network Tuning
Single transfer can get 700-800Mbits/s in each directory. The
transfer buffer was changed and a hardware function was turned
on. There are statistics from BNL to various Tier-2 sites, you can
see SLAC jump from 150Mb/s to 800Mb/s there. 600Mb/s works out at
5TB/day, we don't really have that much data to transfer. Right now
there is a transfer test every two to three hours.
Action Items:
-------------
071017 Wei Arrange meeting with DavidM.
071017 Stephen Think about other issues than might arrive w/out tokens
|