Minutes SLAC-ALTAS TDAQ Meeting 09-Nov-2006
-------------------------------------------

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


Stephen:

Working with Jiry on release tests.  Amazing how large ouptut is of
jobs: 10 events create 10 MB of log file.  Running athena with one of
the slices.  Want to replicate the same test with AthenaMT.  At the
moment using the e-gamma slice.  Want to include muon and jet slices.

Want to try to install these tests on our SLAC test farm.

A question came up in the Trigger Configuration meeting.  They need to
have some TrigerDB data stored in COOL.  Question is if HLT could
write this.  If dbproxy sees that the nodes want to write, what does
it do?  Doesn't seem like that this will work.  Should let Amedeo
comment.  This might be information that should be written by
RunControl.


Sarah:

Tau slices not installed as one of the current tests, because haven't
gotten athena to run with it.  Has been adding the Lvl2 slices.  There
is going to be another HLT release on the time scale of a week or two.
A chance that the tau slice will be there.

LST: As of this morning hadn't decided which node to use for the SQL
server.  Dbproxy waiting for that.  As soon as that is set up, will
move forward with the proxy.

Trying to reproduce a recent crash.  One difference in the log files
is the version of a table (which looks like a detector description).
Wonder if the non-atomic calls seemed to have happend with one version
of the table but not the other.  In terms of the job options it seemed
like the only difference.  Problem is not obvious.

Short discussion on getting CORBA support for guaranteeing requests to
be atomic.  One of the CORBA developers agrees there would be a
solution to that issue.  They don't consider this to be a bug, because
CORBA itself keeps track of where the request was coming from.  There
was also the question about thread-safety.  Sarah will follow-up with
the developer.



All:

Brainstorming what's needed to get the farm running.

Ingredients: MySQL server, POOL files, POOL file catalogue, an HLT
image.

Discussion of whether to use the nightly or some fixed release.

How long are the nightlies kept?  Just one week.  12.0.x are bugfix
nightly.  Shoot for importing the nightly once a week.

There is an rpm created at CERN that is available for import by sites.
Tends to go with major releases of reconstruction software.  Probably
includes binary database files.  Steffen: Need to find out which
version of mysql they are using.

Also need the trigger db.  Rainer will talk to one of the developers
to see where and how we get this.  Also ask Richard about the POOL
files.

Andre might be the right person to ask for an image.  Sarah will do
that.