LSC Data Analysis Software Working Groups

Agenda: Wed, 18 April 2007

Minutes

Attendance:

John
Xavi
Stuart (minutes)
Steve
Dave
Britta
Kipp
Brian
Kevin
Adam
Filippo
Greg
Ben



Announcements:

h(t) generation has started for the released H1 and L1 V3 calibration.
H2 will start once the calibration data are available.  There are two levels
of files: Strain-L1 that includes all the necessary information to regenerate
h(t) or run other calibration studies, and the minimal Strain-L2 that just has
the double precision 16kHz h(t) channel.

Progress in transferring test Virgo frames to UWM is currently held up with
problems in configuring an FTS channel from Lyon, and the necessary debugging
work for this needs to take place at Cascina, but the computer systems are
currently down there in preparation for Virgo joining S5.


LSCSoft:

Ed is combining the 32-bit and 64-bit FrameCPP RPM spec files, and Adam
is making progress on respinning the trio: root/dol/framecpp. It is expected
that this process will converge in the next few days.

Glue:

Kipp has posted documentation on the new coinc tables and this will be
discussed next week. Kipp will commit the circulated changes to segments.py today.

LDAS:

Ed is working on a few corner case frameAPI and diskcacheAPI crashes.
Greg's week long createRDS tests are running well on the new Opteron servers
with a pre-release of LDAS.

LDR:

Kevin has cleaned up most (if not all) of the corrupt frames on Nemo and
will be circulating a report to users on what files where found to be bad.

LDG:

Greg has officially released version 4.4 which is now running on all the
Lab clusters and Nemo. There is still a reported problem by Murali with
pyglobus on RHEL4 that has been directed to the VDT team. Paul is also
reporting problems with FC5 that have not been seen in other FC5 tests so
this continues to be tracked down.

Onasys:

Kipp has fixed a new problem with onasys and h(t) generation at LHO.

OSG:

David is working on a web service container thread problem and has successfully
run nanohype using GT4 (aka web services) globus jobs. However, at the current
scale of running there does not appear to be any obvious performance improvement
in switching from GT2 to GT4.

$Id: 070418.html,v 1.2 2007/04/18 23:34:53 anderson Exp $