Core Minutes 7/5/2005
Present: Joanne Bogart, Anders Borgland, Toby Burnett,
David, Chamont, Johann Cohen-Tanugi,
Richard Dubois, Warren Focke, Riccardo Giannitrapani,
Tony Johnson, Michael Kuss, Matt Langston, Julie McEnery,
Igor Pavlin, Leon Rochester,
Tom Stephens, Tracy Usher
- I & T: (Anders) The trigger test run
for the TEM diagnostic scan was taken on Friday. Results should be
available today. Expect data-taking Wednesday (tomorrow) or Thursday.
He will check into disk space for these runs [and he has: we have
420 GBytes left on the relevant disk, u16]. Some reprocessing needs
to be done of SVAC ntuples for 2 and 4 tower data, because of a bug
in filling one of the variables. The plan is to only reprocess the SVAC
ntuples, not rerun recon. This means we have to change some of our pipeline
scripts.
- Pipeline: (Richard for Dan)
Planned move this weekend of production pipeline to
new version
making use of archiver and Navid's batch interface didn't happen; Dan will
coordinate with I&T to find a good time to do it. The code has been
running happily in the dev version.
- Windows hardware & RM: (Richard) The
2 terabytes of disk have arrived; Navid is configuring. He expects it
to be ready no later than the end of the week. The 3 dual-cpu Windows
boxes for doing RM builds have also arrived. SCS has completed their
set-up of these boxes, but we may have to do additional configuration
specific to our use. (Matt) is in the process of applying security patches
and so forth. Navid should check with him before making serious use
of these boxes.
- Common lib directories
(Richard for Navid) Navid is working on creation of a common lib directory
for Linux builds, actually one per CMT path segment.
- DC2 planning meeting: (Richard) happened
last week. Fallout includes
- Schedule. May not keep all of end-of-July background run
output; will depend on availability of disk space at that time.
Significant demands will be put on Pipeline: up to a thousand concurrent
runs for DC2 plus LAT Integration data-taking runs.
- CalDigi is taking as much time as G4Generator. Zach now has one of
the two copies of AQtime so he can do some profiling.
- Hot/Dead channels, ToT (Leon) is coming up with sets of hot and
dead channels appropriate for the DC2 samples. There had been some confusion
between Leon and Hiro about ToT, now straightened out. In
this plot (which
comes from Monica Brigida, a follow-up to one she showed at the last I&T meeting) notice
that the light blue MC curve (zero occupancy) does a very good job of approximating real
ToT data except for the first couple bins. The difference is explained by
the fact that the official MC occupancy number, 5*10-5 (used in the green
plot), is much too high; Hiro says it has been measured and ranges from (0.9 to 1.7) * 10-6, depending on
tower.
- CalRecon status (Tracy) Generally speaking, data produced
at Bill's urging a couple weeks ago with new CalRecon looks good. Remaining
tasks include
- Improve background rejection by using MIP finder
- Iteration of moments analysis
- Odds and ends, e.g. probable memory leak to be diagnosed and fixed.
May not get much past the first two for the July 13 test run.
- ROOT problem: (Toby) is now mysteriously
unable to read ROOT trees from a simple little program like
this.
Help from ROOT experts is requested.
- FRED problem on Linux: (Riccardo) The
problem is not yet fully understood, but it can be avoided by using
v1r4 of HepRepCorba rather than any newer tag.
This works regardless of OS or compiler version. A
possibly-significant difference between v1r4 and and later tags is an
increase in the maximum message size.
J. Bogart, Last Modified:
01-Jun-2010 15:47:55 -0700