Agenda and Minutes, 21 Mar 2006
Agenda: Tue 21 Mar 2006 (LSC Meeting)
Minutes
Attendance:
John Whelan
Patrick Brady
Jolien Creighton
Kipp Cannon
Adam Mercer
Duncan Brown
Stuart Anderson
Warren Anderson
Siong Heng
We started the meeting by putting together a list of items we wanted
to check the status of and make plans for:
1. Porting LSCSOFT to 64 bit architecture
2. Publication of GEO data
3. Calibration codes for LAL and frame generation
4. Documentation
5. Sys admin grid camp
6. Web servers
Other long term issues which were brought up for discussion included:
7. Preparing for sharing data with VIRGO
8. How to gain better access to GEO clusters
9. General problems
Porting LSCSOFT to 64 bit:
- UWM will get an undergrad to build a clean LSCSOFT for 32 bit FC4
including an lscsoft release of fftw and gsl. We will then
work toward removing gsl and fftw from LSCSOFT.
- UWM will get an undergrad to build a clean LSCSOFT for 64 bit FC4;
initially there will be no support for condor compilation.
- VDT pyGlobus is not working on 64 bit fedora core: maybe
suggest to them a subset of VDT that we need. At the present
time, the list is pretty small, but we may need to keep
VOMS/GUMS to participate in OSG.
- This is to be done by 9 June 2006. (Stuart and Patrick are to
figure out who will do this.)
Publication of GEO data:
- Siong and Kipp would like to have GEO data quality information
available in the segment database so that they can run excess
power on it. Ben has nearly got this working modulo a small bug
in the FrameL; this was communicated to Benoit and will be fixed.
Calibration codes:
- Jolien explained that he now has codes to generate calibration
frames from the ascii files that will be produced by the
calibration team. He showed the group how they work.
- Jolien has also updated the calibration codes for LAL which allow
a simple interface to the calibration data. These codes work and
are ready to be moved from LALApps to LAL CVS.
- Jolien has sent the frame files to Stuart to check if they are
consistent and can be read by various tools.
- Jolien and Brian will coordinate to generate the frame files,
check them, and carry out the test against the point estimate of
the calibration provided by the calibration team.
- We will transition to these new codes after the test is complete;
probably about the end of March or early April.
Documentation:
- we did not spend much time on this, but agreed that we need to
get started developing some how-to documents for the
collaboration.
Sys admin grid camp:
- Warren asked if there was still interest in sys admin grid camp.
The answer was yes. He asked how was the best way to find out
what information would be most useful to the sys admins. It was
agreed that Warren would attend one of the Monday telecons to
discuss this and get some ideas.
- In addition, Warren reminded us that the GRIDs folks are willing
to give us a one hour presentation about cool grid tools during a
DASWG telecon to see if there is anything else we might be
interested in trying out.
- Warren will put together a plan for an August camp in coordination
with the GRIDs folks, Stuart and the sys admins.
Web servers:
- There was a lot of moving of web servers over the last few months
as we scrambled to provide these services the LSC members doing
data analysis.
GEO:
clusters and how to get at them
Web servers
will use names www.ldg-uwm.ligo.org
ldg1-uwm.ligo.org
June LSC target date
Need to get documentation on the web site for sys
admins
Problems:
* RLS is becoming more or more flaky
* Is data replicating automatically if nodes fail
* Documentation
Igor is taking over segment database
LDR services are about a day a week at the sites.
- there are a few corner cases that have not quite been figured out.
- patrick to make
Onasys database and web service:
-
Wish List:
- condor has fallen over a few times
- Alex 500 dags
- Patrick's student crashed it.
- the biggest bug in condor is Duncan
- there is no fair sharing
- need somebody to go through the bologna batch system
- how to put matlab in a box?
Matlab:
- cannot do 64 bit matlab
- vanilla condor in 64 bit
How to get jobs moving around the grid:
- try condor-c: yes. But who will do it?
- Do this in early June.
Data Publication:
will come back to it in about 6 months to see
DMT offline rpms:
need to get this and root working together. Disable globus and
something else.
Are we missing any big pieces?
* Ben mentioned the publication of the segment information for GEO.
* Data discovery for products
Calibration Frame Stuff:
Frame reading and writing code is in place. The ascii reading code
is coming along. Once this is done, he will update the program.
$Id: 060321.html,v 1.1 2006/03/29 18:31:00 patrick Exp $