Core Minutes 2/15/2005
Present: Joanne Bogart, Anders Borgland, Toby Burnett,
David Chamont, James Chiang, Johann Cohen-Tanugi, Seth Digel, Richard Dubois,
Warren Focke, Navid Golpayegani, Tony Johnson,
Heather Kelly, Michael Kuss, Matt Langston, Julie McEnery, Chuck Patterson,
James Peachey, Dirk Petry, Leon Rochester, Tom Stephens, Tracy Usher
- Data Handling: (Tony)
- System Tests A new version was released yesterday. It
features improved navigation. See the
Release Notes.
- Pipeline Matt and Dan are working on setting up a Test database
(in addition to already-existing Prod and Dev). They'll use this
to work on the next project, allowing one task to get its input from another.
- JIRA, Confluence upgrade has occurred. There were some
minor problems, now all resolved or worked-around.
- Data Server Looking into plug-ins to dbs to do
geographical search.
Planning on a two-phase strategy:
- Client makes cuts on time, energy and position which are used to
search metadata and return count of events satisfying the cuts.
Anticipate this phase will be relatively fast,
on the order of seconds.
- Client may refine search, get an event display, make cuts on a
much larger class of variables, and select that part of the
per-event data (ntuple, particular ROOT branches, all,...) to be
returned. This can take substantially longer.
Would like to experiment with Tera-memory to see if it significantly
increases speed.
(Richard) To keep abreast of developments, see
Confluence page
and
data-handling mailing list.
We also want ability to prune backgrounds
and to request a particular set of events, as could be done with the
DC1 Data Server.
- GlastRelease: (Toby) Latest tag (v6r1p1)
is up-to-date except for new Calorimeter code and a couple odds and ends: a new
patch tag for HepRepSvc and a new
ldfReader tag. (Heather) Could go ahead
and promote ldfReader. The still-newer LDF
library could go in when new CAL code does.
(Richard) Zach is finishing up the rest of the CAL work; close
to done.
Richard would prefer that Gaudi complain and stop processing when
it encounters an undefined parameter in a job options file. This would
help to catch misspellings, for example. However another, less clear-cut
situation is when you wish to remove some algorithm or service for
a particular run. Tracking down all the parameters belonging to it
can be messy.
(Richard) Tracy has been attempting to track down memory leaks (which
Julie has seen in SysTests, particularly with low-energy runs) with
ValGrind. He has run into resource limits when submitting these jobs to
batch, but now knows about special ValGrind queues for this very purpose.
- RM: (Navid) Nothing new on the Linux front;
everything seems to be in order. On Windows Navid succeeded in doing a
checkout by logging in and running a script. Next step is to get LSF
to do this.
About the sometimes sluggish performance of the Web interface: Navid
plans to get together with someone from SCS to track it down. It
probably isn't the Web server, nor is it the database access, which
runs quickly enough when done from the machine hosting the database.
-
Personnel: (Richard) We have interviewed
two quite acceptable candidates for Alex's position. One seems like a
particularly good match, if he is willing to take the salary we're
offering. However, there has been no progress in finding a way to
keep Marco on GLAST. If we lose him, it will have serious impact on
MRvcmt, MRcmt and rdbGUI.
∫ Flight(w, ..) dw
-
EM: (Heather)
She is working on a patch tag to include the new LDF library, already
built on both platforms. Next she would like to make a release including
the new TkrRecon. The new CAL code will follow in a separate release.
-
ToT: (Leon)
pieces are there, though one step (translation from XML to ROOT) has
to be done by invoking the translator program by hand. (Leon) expects
to get back to this approximately now.
- Callable interface to metadata dbs: (Joanne)
Work is not going as fast as she had hoped or would like (lots of distractions
both internal and external to GLAST;
similarly for Leon),
but it is going. First phase – transparent reorganization
of calibUtil services needed by CalibSvc to
fetch calibrations – is close to done; should finish today or tomorrow.
- Calibration meeting: (Richard) Eddy would like
to have one next week. Calibrations taken with the assembled tower should
start appearing soon and he would like to be sure that
...
J. Bogart Last Modified:
01-Jun-2010 15:45:52 -0700
previous
