Attendance:

  Kent Blackburn,  Duncan Brown,  Patrick Brady,  Alan Wiseman,  Jolien
  Creighton,  Phil Ehrens,  Masha Barnes,  


Action items:
  
  Expect an e-mail from Kent about scheduling and running at the sites
  

Status of LDAS:

  Testing very hard a frozen version of repository.  Have had several
  restarts.  There is an ability to detrend,  but there is an issue to deal
  with gappy data.   Fills gaps with zeros and then can fill gap with
  something else;  then detrending could cause some problems.   If there are
  no gaps,  there is no issue.  

  Timeout when system is heavily loaded with requests for wrapper jobs.  It
  queues up job requests with an infinite wait.   A failure was occuring when
  datacondAPI would time out.  Should have jobs queue only long enough to have
  things handle -- now set to 10 seconds.  They have implemented a fixed delay
  and when there are still no nodes,  it cans the job and reports an error.  

  If you request too many nodes,  it will report back that not enough nodes
  are available.   

  This problem is only running on test and dont have enough information to
  understand what is happening.   The timeout can be set in a resource file.
  The datacondAPI currently has a timeout set to 300seconds.  

  They have turned all LDAS programmers into read only and only Ed can check
  in code.   Each day,  once they have identified and fix that works,  Ed puts
  it in,  then he tags the code.   

  They have been putting the pre-release on dev,  test.   Working well all
  week.   Some of these bugs are exotic functionality.  However,  when they
  fix code

  Hope not to make any changes at all.  They will only change one file that
  causes only file that causes the problems.   Want a release a week or so
  after the engineering run to capture some of the missing functionality.
  There are a couple of other things which have been requested;  kent needs to
  get Disk Cache API.   4--6 weeks after that release.   
  
LAL and LALWrapper releases:

  FCT.   They have a problem with the hierarchical is they have not got the
  stuff for computing the correct output stats from flat search code.  Defines
  n-dimensional solid regions in parameter space,  but does not refine into
  smaller blocks.   Hierarchical search code will be rather slow.  Not run
  under LDAS at this time.   

  Greg would fix this quickly.  One column needs to be put into that new
  column.   This would be good because they use his code to verify that the
  frame writing capabilities 

  A small amount of E6 data will be kept at the site including some
  overlapping data and the data with the injected signals.   

  E7 tapes will be sent to Caltech how often?  Continuous data is 2-3 days.
  Data will be available at the sites for the entire time.  
  
  current version of code from greg 


Calibration:

  Code to produce from ZPG representation will go into LAL before E7
  release version  


Miscellaneous:

  Set up a web page for each of the systems which has a table saying
  search group versus