Attendance:  

  Raha,  Greg,  John, Warren,  Joe,  Martin,  Siong, Masha,  Luca,
   Ed,  Peter,  Kent,  Erik,  Jolien,  Alan ...........


Action Items:
  Stuart will discuss with Benoit,  John and Albert about the local
  time information stored in the detector geometry.  

  Jolien to e-mail Albert about the information from the summary of
  changes to the frame spec that did not make it into the text.

  Duncan to ciculate a proposal for the channel and interferometer
  information.  Peter will iterate.


Discussion of detector geometry information:   

John Whelan started discussion.   The summary of changes to frame spec
mentions some items which are not in the text.  Jolien thinks we might
want a linked list of the detector structures in the  structures.
Kent agrees.    Will add pointer to next det geometry structure.  

General discussion of local time information in the detector geometry
structure.  Some people felt there might be a need for it.  Siong and
John claimed that since Allegro will restart each time they re-orient
it.  Stuart will ask that local time be officially dropped by Benoit,
John and Albert.    


Another change request:

Duncan:  interferometer name and channel name.   Currently needs to be
parsed out of the name in multiDimData.  Put things into domain
structures ...... Would be adding another set of meta-data that get
stored in another place.   Duncan will put together a proposal with it
-- multiple interferometers,  multiple channels.    Action peter will
send an alternative proposal.  


Scalability of LDAS:

Have run jobs which used 296 nodes at UWM.  We're limited by crappy
frameAPI ultra 30 not by other things.  We will be doing many
simulations using LDAS in the next while.  Stuart thinks this is a
great opportunity to run on a large number of nodes.   How did it
scale since the fix?  Duncan will start to do scalability tests.  Are
there any intrinsic bottlenecks.  Lamboot ....... etc.  Can we run
lam daemons continuously?


Inquiry about requirements of Frame/Disk Cache API:

Can handle automount, but better not to since it make for significant
overhead in the frameAPI.   Can tell you what frames are available on
the system.  Globus could request this if ........... and put it into
it's catalog.   The rest of that catalog is for storing virtual data
products.  


Database issues:

The question of where to put entries from search codes run on channels
other than the GW channel came up again.  No resolution.  Should be
put on the agenda again.


DSO issues:

Philip Charlton reported that the FCT code should be available for
scalability and testing.  The flat search part is going well.  

Rauha is working on a DSO to do triggered searches.  This led to the
general question of how long the E7 data would be available at the
sites.   Stuart mentioned 1-2 months;  2 months being the outside
limit.

Where will reduced data sets go?  Where does it get archived?   Stuart
would put 0.5 -- 1 Tbytes RDS on all the lab LDAS sites.