Core Minutes 6/7/2005
Present: Joanne Bogart, David Chamont,
James Chiang, Johann Cohen-Tanugi,
Seth Digel, Richard Dubois, Dan Flath, Warren Focke, Riccardo Giannitrapani,
Berrie Giebels, Tom Glanzman,
Navid Golpayegani,
Tony Johnson, Heather Kelly, Michael Kuss, Matt Langston,
Chuck Patterson, Igor Pavlin, James Peachey, Tracy Usher
- Data handling: (Tony)
- Pipeline Plan for weekend was for Dan to put archiving facility
into production, then start on automated archiving this weekend. Dan later
confirmed that this is where we are. Next he plans to drop in code, written
a while back, to use Navid's batch interface (has advantages like using
fewer db connections). Dan plans to do more testing on dev pipeline, then
will coordinate with SVAC to to find a time when he can put it into
production. Meanwhile Igor is learning about Java messaging and how it
can be applied job submission, ultimately to be used in the Data Server and
next-generation Pipeline.
- JIRA administration Matt organized a meeting on this subject.
He'll be writing up the results. Among them are the intention to
implement a scheme to disallow submission of spurious
issues by unauthorized people.
- Data Server It is generally working well, except that putting
new data in is very slow. The system uses a MySQL database. Performance
will also be evaluated with Oracle, but the expectation is that results
will be similar.
- Peeler (Tom G.) Progress is being made. Most of the time
is being spent on testing and bug fixing. If there is a need, Tom
promises 72 hour delivery.
- Schedule Richard asked Tony to make a schedule for all the
Data Handling components, perhaps to be discussed next week.
- ROOT Review: (Heather) Next Monday at
9 some BaBar folks will be looking at our use of ROOT, primarily the
(Event) ROOT classes and related matters. Richard doesn't expect
any suggestions for radical changes, particularly since in the past
we've been reviewed by Mr. ROOT himself, Rene Brun.
- Cividale meeting: (Richard) As usual,
our Italian collaborators were
superb
hosts.
- Alignment (Tracy) Sara's presentation showed that there are internal
(to a tower) misalignments. Michael will provide a text file describing the misalignments for Recon.
When he returns next week Leon will make code upgrades required to use it.
- CalRecon (Tracy) See the
CAL
confluence page for the to-do list resulting from discussions among Tracy,
Bill and David Chamont. (Richard) Will this be ready by the time DC2 needs
it? (Tracy) Most of the items are cut and dried and will be straightforward
to implement. Only a couple are more open-ended.
- Maintenance (Riccardo) is fixing some problems in MRvcmt and
FRED that Toby pointed out.
- G4 upgrade: (Tracy) Should we be upgrading
to 7.0 or 7.1? Geant people think so. Finally 7.0 has the multiple
scattering problem fixed so we won't have to pull in an old version of the
module.
Francesco has some more validating to do; otherwise everything is in
place for a move. (Richard) Users from the Linear Collider may have detected
a problem with thin volumes in sampling calorimeters which could affect us
with either of the new versions.
- Matt report: (Matt)
- Windows central installation: At the moment you can only get
to our Citrix server via vpn, which is not what we want [this is now fixed]; however, once
you do get in, it works as it should. Glast09, 10 and 11 are not yet
ready for us to use, in particular, LSF in not installed. This may
take some nagging of SCS. The machine serving Confluence and JIRA
periodically has problems requiring restart of servers or reboot. There may
be configuration errors or we might conisder upgrading versions.
- Windows web mirroring Mirroring of the SLAC Windows web to UW is
working, but there isn't (yet) enough disk space over there to hold
everything, so only part of the SLAC web is mirrored.
- Calibration file syncher Matt has written one for Windows. [On
Linux the rsync command can be used for this purpose.] Once installed it
works autonomously, like Windows Update, downloading new files as it
detects them as long as it's not pre-empted by other, higher priority
network traffic.
- Navid report:
(Navid)
- Windows RM builds are suspended pending installation of
new disks, which should be soon. He doesn't want to have to deal with
killing and then restarting a bunch of builds.
The Citrix box is at Windows 2000 rather than XP, so has trouble
with builds because of the limitation on command line length input to
cmd.exe. (Matt) It's 2047 characters for Windows 2000,
but a much larger number for XP.
The alternate scripting environment,
WSH, has no such restriction.
- Tag collector The new version, integrated with RM, is about
ready to go.
- Automated Systests almost. RM does most of the work for each
tagged release; there still is one little bit Julie has to deal with.
For the time being, only one Systests job can run at a time (a
limitation of Systests itself, which writes some information to a fixed
place).
- gcc upgrade: (Richard) Is it time? We've
been at our current 3.2.3 for a while. (Michael) Fedora Core 4 will be released
soon, perhaps only with gcc4. (FC3 comes with gcc3.4 and has gcc4 as a
test package). Some of our extlibs (e.g. Gaudi, Geant4) don't compile with
gcc3.4 It might be best to foxus on platform (OS version), which will
determine compiler version.
(Richard) would like to make it easier for people to install and use
a new (gcc + glibc) version, regardless of platform default.
(Berrie) has done this with no particular difficulty. (Navid) RM
can easily enough make its builds with a different compiler version;
just need a new CMT tag.
- Heather report: (Heather)
- EM tags New base will be GlastRelease v6r8. Almost everything
has been promoted that needs to be; waiting for calibGenCAL to settle
down.
- Recent activities include wrestling with CINT and work with
the ACD team. Bob Hartman has joined and is soaking up ROOT lore like
a sponge, or maybe a root.
- Data-taking: (Anders) [This just in. ed.]
Data taking this weekend with 4 towers was problematic. calibGenCAL had several
bugs which meant that we had to set the CAL LO (the 100 MeV CAL trigger)
to a fixed threshold value by hand for the two new CALs. Then we realized
that the same 12 LAC (zero-suppression) thresholds were applied for all 8
layers in the new CALs. We decided to go ahead and take data
anyway. However, a GASU problem I had just started to notice last week in
some test runs suddenly appeared in the SVAC data runs. This means that in
nearly all the data we took the trigger word (GEM conditions summary word)
is zero. The ELX people are able to reproduce the failure and have a way
around it to avoid it in the future. See
SVAC-57.
By the way, the 4 tower runs are taken with the 1-shot TKR trigger
which is the new default configuration from now on.
We are currently installing the fifth tower and should have six towers
in the grid by Friday. We are supposed to run the CAL calibration runs
June 15 and 16, and then the SVAC data runs June 20.
- Miscellany: (Richard)
- Next
ADASS meeting
is in October in Madrid. Need to decide who should
go and what they should talk about.
- DC2 planning meeting is at the end of June.
- Some folks will be visiting CAL France in mid-July.
J. Bogart, Last Modified:
01-Jun-2010 15:47:31 -0700