o Improve web pages so that information about the software and services
  can be very easily found.

o GAT tutorial: will repeat this on the Access Grid in August. Ed will
  publicize with Ohio etc and fix a date. 

o Important that everyone developing is able to use the testbed!!!
  Mirek will send around a list of suspects who are not in the gridmapfile.

o Would be nice to track the stability of the testbed, save number of 
  services working on each machine with the date, have prizes for machines
  which are always working, and motivate machines not working into doing
  better, etc.

o Add to the machine status and service status web pages ...
  - which machines does GRMS run on? 
	o test running echo
	o test running simple Test.c program
	o test running simple Test.f program
	o test last two but with MPI
	o test the GAT Migration application
  - send Mirek tests for all services
	o data movement
	o test application contacting monitoring (Kelly)
	o firewall information needed (check each site against each other site)

o Kelly: command line scripts

o Gabrielle suggested that GRMS only uses machines on the testbed which 
  are fully working (as indicated on the testbed web pages). Is any of 
  this information kept in MDS ? Maybe this could be part of the adaptive 
  component? 

o Chris: will fix the web page with pictures describing the scenario, removing
  a couple of additions and adding in the missing items. Also it would be 
  useful to add in the indentations which Tom used in the original. (Actually 
  what about using HTML so that there is a short version of the scenario where
  you can then click on links to expand each component? If we do this then
  there should be a button to expand everything out). Fix STATUS items and 
  then Kelly/Tom will check.

o Zoltan: will suggest a good metric to use for the trigger for the migration 
  scenario. (e.g. cpu speed of processors, average load, etc ... some of these 
  don't make much sense in a HPC environment).

o Tom/Kelly/Zoltan: Sort out the format in which information is passed back
  from monitoring.