Core Minutes 4/4/2006
Present: Joanne Bogart, Anders Borgland,
Philippe Bruel, Toby Burnett, David Chamont,
Jim Chiang, Johann Cohen-Tanugi, Richard Dubois, Dan Flath,
Berrie Giebels, Riccardo Giannitrapani, Tom Glanzman,
Heather Kelly, Michael Kuss, David Landriu, Bryson Lee, Francesco Longo,
Chuck Patterson, Igor Pavlin, James Peachey, Leon Rochester,
Tracy Usher
- Processing needs: (Richard) It's a regular
3-ring circus:
- Regenerate the DC sky. Problems were detected with pulsars and with
GRBs, in particular correlation of GBM and LAT.
Resources:
Last time it took a
few days and about 3 terabytes of disk.
Target start time:
about April 13.
- Bill, just back from Patagonia and fully energized, thinks we
need
at
least one more background run before hand-off of the LAT (August or
September) of 2-5 billion events with various improvements (e.g.,
upgraded ACD geometry).
Resources: About 7-10 days and 2 terabytes.
Target start time: Yesterday.
- Beamtest simulation.
Resources About 2 terabytes.
Target start time: Next week.
We may be disk-shy. We have 60 terabytes on the way, but installation is backed
up behind some electrical work at SCS, due to be completed by mid-April. Tom believes we can retrieve 5 terabytes which could be backed
up to tape, and we still have 2 terabytes left from our last disk acquisition.
- GR code merge for external libraries: (Toby)
This is a painful subject because of various Windows problems:
- For a substantial period all RM Windows compiles were failing.
- Once the compile problem was fixed,
a different problem manifested itself: now all test
programs fail.
- There are too many simultaneous builds and they take too long. Turn-around
has been 6-7 hours. (Richard) Need to investigate to know whether the long build
times are due to many simultaneous builds or something else* (for example, check
that all three Windows boxes were up and running).
(Toby) Particularly in light of the global compile and test program failures,
the Windows RM needs to be actively monitored. (Richard) This is the job of
Windows developers at least as much as it is Navid's, as owner of RM. (Heather)
Since these kinds of global failures can be caused either by changes in critical
packages like GlastPolicy or by changes to RM itself, it would be helpful if
Navid would send an announcement when changes to RM go in. (Richard) In the
meantime, when RM builds are failing or just slow, Windows developers can grab
the source tarball and build on their own. For building locally, the Linux source tarball is adequate. (Toby) If there
is any chance of making commits to CVS, the Windows source tarball (which normally
shows up later, after the RM build is complete — could this be
changed so that the tarball is generated first?) should be used to avoid
end-of-line problems.
- Support for charge injection data:
(Bryson, Heather) The goal here is to make this data available offline
where additional tools are available to look at it. This involves work
in CHS/EventRet, and
new data structures, both in the TDS and ROOT classes, to keep track of
the information in the calibration request (DAC
settings, etc.) for each calibrated subsystem. This all should be
ready to go by the end of the week.
- Beamtest: (Michael, Francesco)
The new BeamtestRelease container package
is nearly ready for a serious tag. Beamtest needs Geant visualization,
which is not part of the Glast build of G4
[but should be! ed.]. So
far Michael has not been able to successfully build the new Geant with
all the visualization libraries; Heather may be able to help out.
(Richard) What's needed before Beamtest will be ready for the Pipeline?
(Francesco) There will be some changes to the tuple. Also, we need
a plan for local file space. Should be ready to go by Thursday.
- Databases: (Bryson, Richard)
There are or will be quite
a few resident at SLAC, with varying degrees
of dependency on each other and overlap in clients. One goal of
discussions which will start this week is to come up with a common
interface for analogous queries as much as possible.
- CMT and MRvcmt: (David Chamont)
Riccardo has released a new version of MRvcmt, v0r22, which
supports CMT v18 as well as v16. David has been using it (as well
as the new CMT) and has seen no problems; that is, no new
problems. It would be well if other people, who may use these
tools differently, would also try them out.
-
Announcements: (Richard) Heather
will be visiting SLAC Thursday and Friday of this week. She'll be
spending time with Chuck on Workbook issues.
J. Bogart, Last Modified:
01-Jun-2010 15:45:53 -0700