Core Minutes 4/5/2005
Present: Joanne Bogart, Anders Borgland,
James Chiang, Johann Cohen-Tanugi, Richard Dubois,
Nicola Giglietto,
Navid Golpayegani, Tony Johnson, Heather Kelly, Michael Kuss, Matt Langston,
Chuck Patterson, Igor Pavlin, James Peachey,
Leon Rochester, Tom Stephens, Tracy Usher
- Data-taking
- Hardware (Richard, Anders) Tower A
(that is, a
real CAL + TKR tower) is in the grid; tower B may be in by tomorrow. Data
taking in this configuration is expected on Tuesday. It will include the
new 32-bit event counter (supported by LATTE 4.8).
- Overly long About 10% of the runs failed in the xlong queue
because they ran out of
time. The failures
are Van der Graaf runs which tend to be very noisy, averaging about 65 hits
per event; some events had as many as 1000. The jobs have since been
resubmitted to the xxlong queue.
Meanwhile Anders tried one on his 3.2 Gigahertz desktop. It took 14 hours
to complete and broke his fan: a killer run! Of this about 40
minutes was spent writing the ROOT file, a reasonable percentage.
We need to think
about cuts, e.g. on hit multiplicity, so that we quit processing these
time-consuming and uninteresting events. This morning Leon had a vision,
or at least a bright idea, that an ntuple variable containing cpu time
expended on that event would help him to design an appropriate cut
strategy.
- Multi-tower calibrations Calibration infrastructure and tracker
code essentially support them. Missing pieces should be done shortly:
finish up charge scale support
in tracker code (Leon); enhance XML to ROOT translator by implementing
an append function (Joanne).
- New EM tag (Heather) p9 is coming soon, perhaps today. It
will include the new LDF external
library, required for the 32-bit event counter. It also will include modified
RootIo job options parameter for max file size, Johann's 3-package update to
trigger bits (note the current 3-in-a-row will get a new name), and calibration
infrastructure support
for charge scale calibrations, affecting another 3. Zach's update to make
flight gains the default, recently promoted to GR HEAD, is already in
the current EM tag.
- Data handling: (Tony)
First release of data server for DC2 is almost ready. Front end
and back end are in hand; there are still some issues to be resolved in
communication between them. Meanwhile, Jean-Paul is working on the
"real" data server.
Pipeline is undergoing gradual improvements. Since it was difficult to
pick up all the little odds and ends of file output for archiving (not just
the more well-known outputs), SVAC/Online will tar everything up into just one
file to be archived. (Richard) Dan is going to make use of Navid's
interface to the batch farm; to date interaction with the batch farm has
been one of the more problematical parts of the Pipeline.
- RM on Windows: (Navid) It's up and running;
JIRA issue RM-17
has been closed. There may still be the occasional odd problem with
incorrectly installed external libraries, but practically all functionality
is now there. Windows binaries are not yet being generated for the Installer;
that could be ready by the end of the week. He has held off
instituting email notification of failures
initially since there were many spurious failures, but that has now settled
down. (Leon) Not completely. RM claims TkrDigi is failing the Test step
but the log file looks fine.
(Richard) It would be nice to be able to
use a Windows RM build from the GLAST Citrix farm by including it in
your CMTPATH as can be done on Linux, but there is still a small glitch to
be ironed out there. We have ordered 3 more boxes like glast06 to do
RM builds.
- CAL news: (Tracy) He got the mip-finding
tool to run on Linux. The TDS classes will be discussed at the next
CAL meeting.
- Science Tools Checkout 2: (James)
It's generally going well, but some new issues have come up (spotted
by David Band) concerning burst tools; fixes are in progress. (Jim) will
tag a new ST release once James checks in his latest changes.
- SCS dealings: (Richard)
Would like to set up a series of meetings with SCS to keep them abreast of
our needs and to solicit advice in designing aspects of the data handling.
J. Bogart, Last Modified:
01-Jun-2010 15:48:21 -0700