Core Minutes 3/22/2005
Present: Joanne Bogart, Anders Borgland, Toby Burnett, David Chamont,
James Chiang, Johann Cohen-Tanugi, Seth Digel, Richard Dubois, Warren Focke,
Riccardo Giannitrapani, Berrie Giebels, Tom Glanzman,
Navid Golpayegani, Tony Johnson, Micahel Kuss, Matt Langston,
Chuck Patterson, Igor Pavlin, James Peachey,
Leon Rochester, Tom Stephens, Tracy Usher
- Readiness for data-taking (Richard)
- EM v2r060302p6 Thanks to heroic efforts by Anders last
weekend, a new bug in processing of diagnostic blocks has been fixed.
This tag also contains the CAL energy bump fix. There is one known
small bug in the Glt part of the merit ntuple, but it is not a showstopper.
Anders is happy with System Tests output. We have a CCB-certified release!
During Heather's absence, Joanne is subbing as EM gatekeeper. Please contact
her with any requests to promote new package tags to EM Head.
- CCB procedure See the
CCB confluence page for background. See also the
page for the
baseline release.
Proposed changes to any of the controlled software should be described in a
JIRA issue and tested with the test or dev pipeline. Then the proposed changes should be aired at an Instrument
Analysis meeting (they happen every Friday).
Finally, all the supporting information should be sent
to Richard, who will in turn forward it to the full CCB. For Pipeline
front-end, which is on the boundary of CCB control, not adequately exercised
by System Tests, and which is likely to require quick turnaround, the plan
is to appoint a two-person committee to verify the change is ok. CCB
can then use their report as input.
(Tony) What about System Tests front-end? Is this covered by CCB?
(Richard) No clear need for it; he would prefer it not be.
- Data is coming when? Scheduled for today, but, as we've
learned, that doesn't guarantee it will actually appear today. Last
known hang-up had to with GASU verification.
- Data handling: (Tony)
- Developers' workshop is taking place this week.
Dan is holding a seminar on how to fix likely Pipeline problems,
so that he won't have to be on call 24/7. Determining exactly what
are the likely Pipeline problems is still a work in progress.
Other Pipeline activity includes joint Dan/Navid projects
on archiving and on batch submission. Later this week the group will
hold brainstorming sessions on future developments.
- ROOT classes for access to dataset catalog (Richard)
Tom Glanzman has started work on this. Tony & Co. will put
together a web interface, similar to the one Navid provided
for DC1.
- GlastRelease: (Toby) Keeping Tuple
documentation current has been a perennial problem. Finally an
approach which seems to cover all bases is in progress.
The Confluence pages which
the Workbook uses will contain information for old released versions, but
will also point to Doxygen documentation for latest changes. Tuple
maintainers will be responsible for keeping Doxygen documentation current.
(Richard) Can we keep certain old versions of the Confluence page and index
them? (Matt) Yes, Confluence has support for these things.
(Toby) Thanks to Riccardo's good works, in the next GlastRelease, Gleam
will start up Fred automatically.
- RM: (Navid) Checkout, compile and unit
test phases are now all being done.
He had to exclude building the test program for one package since on
Windows it wants to run interactively.
(Richard) We will need more Windows resources for these builds.
We don't have anything like the resources
of the Linux farm at our disposal, in fact at the moment we have one
dual-cpu box. This is fine for one build at a time, but 3 or 4 concurrent
builds are common; 10 or more have been seen. We expect to buy a
few more Windows boxes. Exactly how many is still under discussion.
(Matt) See some statistics
on RM performance from the
glast-ground home page. (Toby) Policy for triggering LATEST builds
affects the performance. He doesn't like the current arrangement.
(Richard) Will Installer now be able to distribute Windows libraries,
executables as well as external libraries? (Navid) Yes, but haven't
done it yet.
(Richard) What about RM submitting System Tests?
(Navid) Will come. Right now Julie has
a script she uses for manual submission. The script is still under
development.
- Calibration access: (Richard)
During Integration production calibration files will be written to
/nfs/slac/g/glast/calibrations/ and its subdirectories.
Anyone
wishing
to access calibrations while examining the data-to-come from a machine
which can't mount this disk will need
- a network, to communicate with the MySQL metadata database.
- copies of the calibration files themselves
The disk is ftp-accessible and is served by Samba (for Windows users at SLAC) Julie will rsync at Goddard. We need to
look into other mirroring tools for other situations.
- Science Tools checkout: (Seth)
So far, so good.
J. Bogart, Last Modified:
01-Jun-2010 15:48:29 -0700