Minutes SLAC-ALTAS TDAQ Meeting 20-Jul-2006
-------------------------------------------

Present: Steffen, Sarah, Rainer, Stephen (phone)


Stephen: 

Working on DetCommon, have set up tag collector (which tags of which
packages), currently four packages, will add another two: TriggerTools
and TriggerMenuCompiler.  Ran across a dependency problem.  The four
were all "glue" packages (inside ATLAS core), so now DetCommon will
depend on core.  Talked to Dave Quarrie and that's ok.  People at
Point 1 eager to use this, too, since, first four were offline.  Next,
with finalized tag collector, build into a release.  Need to sort out
dependency of xml parser, external and not in ATLAS core.

Situation at SLAC: release 12.0.1 installed, added latest patch,
announced in HN.  Next release 12.0.2 will be built next Tuesday.
Takes ~1 day.  Giving Zach instruction for how to import.


Sarah:

Some folks around Hans vdS, now discussing the use of Shared memory
between dbproxy and HLT.  Amedeo has a meeting at CERN trying to
figure out some open questions.  Currentlly launching dbproxy with
ssh, see if we could use the DAQ partition mechanism.  Also checking
policy on using udp, semaphores, etc.  Also checking namespaces,
conventions.  Dependency on MySQL.  No problem using C++ exceptions,
they weren't initially, but are now supported.  Code now in our own
CVS repository (set up by Stephen).  Need to figure out where this
should eventually go.  Create a CERN package select a project,
librarian.


Steffen:

Recap decision on buing 3750 switch: no other parts on site, so would
have to buy two.  Rather, found matching (old) network equipment,
still slightly better than what's now in IR2.  But will get this for
free.  It's a Cisco 6500 supervisor 2 plus 16-port optical GB line
cards.  16 GB backplane good enough.  Supervisor has a router so has
it's own subnet, ok.  CERN switches: for SLAC they would be $5k rather
than 2700 CHF.  For SCCS no problem if Steffen hooks them up to the
network.  Explore possibility to borrow switches from CERN purchase.

Changes: No more opportunity to install some of the new BaBar boxes
during a ROD at IR2.  a) they essentially want to install all at once,
and b) there will be no ROD.  Installation of BaBar may be much later
by Aug 21, depending on peoples schedules (Ray etc.).  Still could
pull old farm after that date, but power issues in building 50.  Some
uncertainty there.  Need to talk to SCCS, probably need a more formal
request.

Discussion on for how long the farm would need to be supported.  Want
to come back to this at the next Wednesday meeting and make a proposal
for a concrete timeframe so can make compromises.