Minutes SLAC-ALTAS TDAQ Meeting 24-Aug-2006
-------------------------------------------

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


Stephen:

DetCommon has been in the nightly now.  There are two parallel builds:
A development nightly (numbered 12.x.0 etc, currently: 12.2.0) and a
bugfix nightly (for production, 12.0.x etc, currently: 12.0.2).
Regression test started running in the nightly.  Didn't pass the first
time, will update scripts and see again tomorrow.  Releases for LST:
offline plans to use a specific build based on 12.0.3 called
12.0.3-LST.  For the TDAQ part we want to use 1.6.  For MST: 1.4.1 for
the TDAQ and 11.0.6 offline.  Talked to the developers of the
TrigComp* packages, including Joerg and sorted out issue from last
week.  Muon system might be interested in using DetCommon.


Sarah:

Tried to build dbproxy against 1.4.1, not successful because ERS
changed significantly from 1.4.1 to 1.6.  Talked to Haimo who is
getting the rest of TDAQ to work on the Manchester cluster, running on
Grid machines.  Haimo not convinced that they will get it to work in
that environment.  Very intersted in testing on a SLAC farm.  Could
aim for some form of test at SLAC in mid-October.  Don't want to spent
too much time now on making dbproxy work with ERS as in 1.4.1, because
use of 1.4.1 for tests now questionable.  Works in 1.6 anyway.  Not
possible to run DAQ with dbproxy on lxplus just now.  Currently only
posssible on the real farm P1, but limited access.  More reason to get
ready for setting up tests at SLAC.  One of the next steps is to sort
out the different versions of MySQL in use that test proxies would
connect to.


Amedeo:

Completed multicast.  Plan to make a commit at CERN today.  Using port
that was already there for control, now can also send data through it.
Not the reliable multicast (yet) of course: If one datagram is dropped
you lose the entire entry.  Can already measure the performance.
There was a suggestion made to Sarah to handle connection errors
through IPC.  Somewhat mystified by this. [Discussion] Sarah will talk
to the RC person about it again and find out.


Steffen:

Accounts for non-SLAC people need to be sorted out ahead of time.
Discussing requirement for the farm to be ready for tests, e.g., OS:
SL3 (or SL4), Release: 12.0.2, TDAQ 1.6.  IR2 farm turned off for now
for the lack of cooling.  Will configure the new farm with SL4.  The
new machines will not move in before October.  Could be end of October
because Ray is tied up with LST cabling [no pun intended].  Move out
probable in mid-October.  Disruption may be about two days.  Aim to
configure farm to ATLAS machines in situ at IR2, including SL4, taylor
etc.  Server can be a Sun box.  Database server to be sorted out.
Talk to John Weisskopf about the schedule for the move.  This might
take an additional week.  New switch purchase in the process.  Found a
small vendor, cost just under $5000.  44 ports available.  Eventually
might pick one of the nodes to provide nfs when in building 50.
Propose 500GB of server space.  Could use a spare T3 RAID from BaBar.


Action items:

20060824 Sarah/Stephen: Find out the requirements on the IR2 farm and
servers (including disk space) for TDAQ test.