Core Minutes 5/3/2005
Present: Joanne Bogart, Anders Borgland, Toby Burnett, David Chamont,
James Chiang,
Seth Digel, Pol D'Avezac, Richard Dubois, Zach Fewtrell, Berrie Giebels,
Tom Glanzman,
Tony Johnson, Heather Kelly, Michael Kuss, Matt Langston, Julie McEnery,
Chuck Patterson, Igor Pavlin, Dirk Petry, Leon Rochester, Tracy Usher
- Data handling: (Tony)
- Installer GUI See
description and instructions. Windows users are encouraged to try it out and give feedback to
Tony. Currently there is only a Windows version, but it should not be
difficult to port it to Linux or even Mac. Don't attempt to install
GR v6r7; there is some problem with it. (Toby) That's because it doesn't
build properly on Windows. It can be fixed with a small patch.
- Data server Jean-Paul and Tony almost had the new version up,
but ran into problems with lack of space for the database. They have more
space now and expect to make another attempt.
- System Tests There was a new release recently, featuring some
bug fixes. Remaining to-do items don't appear to be very high priority,
so the group expects to focus on other projects, such as the data catalog.
- Pipeline New release with archiving should be ready shortly.
- GlastRelease:
(Toby) Problem with v6r7 points up need to label good releases, delete or
hide bad ones. (Richard) On a related topic, Navid recommends that we not
try to rename good
HEAD releases since it would not be so easy to get this right; better to
make another build.
- Using PyROOT: (Pol) See
his
write-up on the topic, including a couple example scripts.
(general
discussion) PyROOT is not the only Python interface to ROOT (for example, Paul
Kunz has written one), though it is the officially supported one. It comes
pre-built in the Linux ROOT distribution (4.02) we're using; Windows
distribution has only the source, but Toby had no trouble building it for
glast_ts and also made an installer which should make the process easier for
other Windows users. See his
notes, including a link to the installer.
- Calibration infrastructure: (Joanne) will
disappear tomorrow for 10 days. See this attempt to capture everything most worth knowing
about calibration plus a couple remarks about geometry.
She's writing a diagnostic to be used to verify that the metadata
database is in good shape.
Here are some early results. Some form of the program will be released and tagged today in
the calibUtil package.
- I & T data-taking: (Richard, Anders)
Calibration runs are happening. Module 104 is done; 105 is in progress.
(Richard) How does it happen that the modules are being calibrated
separately? (Zach) Only muon data-taking is being done multi-tower for now.
Otherwise the procedure is tower by tower, with a (scripted) merge step
afterwards.
- EM releases: (Heather)
Current
EngineeringModel tag is v4r060302p22, which looks pretty good. It has
a new TkrRecon, addressing some JIRA issues.
- Computing hardware: (Richard)
Final price will be $120K for 32 terabytes of Linux disk. This leaves
us with a little more cash than we might have had, to be used to buy
2 Oracle servers ($40K for them and their disk space). We'll also
be getting 2 Terabytes of disk for Windows and we'll be contributing
25 dual cpu's to the batch farm. This guarantees us at least this
much compute power; in practice we get significantly more. Now that the
big ticket items have been taken care of, we can think about adding
a server here and there so that there won't have to be as much sharing
of servers by different applications.
J. Bogart, Last Modified:
01-Jun-2010 15:45:31 -0700
previous
