Minutes from OGSA BES WG session #1, GGF14

Minutes (by Mark Morgan)

BES

* Agenda
	- IPR Note
	- Agenda Check
	- F2F debrief
	- Review of Service Container interface
	- Discussion of un-resolved topics:
		-- Legacy Application interface (Chris Smith)
		-- Resource model discussion
		-- Container Management interfaces
			--- Start/Stop accepting activities
			--- Shutdown Container
		-- Register for events
			--- e.g. activity/container state change.

* EMS Scope Review
* Face to face debrief
	- went over minutes from London F2F
	- staging as defined in JSDL
	- Presentation of results of resource modeling to be given -- probably on telecon
* Have VERY rough draft of specification
	- Very important to keep in mind that this is "BASIC" execution, not execution in general.
	- Is this a spec for a single web service?
		-- Yes, unless we incorporate the interface for the activities.
	- Is this state (getActivityState) held in the container?
		-- Implementation dependent
		-- Out of scope
	- Are you going to specify types of exceptions?
		-- Yes.
		-- Start making your list of favorite faults
	- Everyone is happy with the high level view of the container.
	- Going to cover states in more detail tomorrow
* Legacy Application Interface
	- getJobState/getJobInfo
		-- Retrieve run time information about the job
		-- While notifications on job state change is important, is it within the scope of a "basic" interface?
		-- Notification through some event mechanism is good, but out of scope to choose which one we'll use right now
	- getJobDefinition
		-- Retrieve teh original JSDL document that was used to define this job instance
			--- Depending on the implementation, this document might not be available
		-- Need to be able to retrieve audit information (well, you want to at least)
		-- Discussion on whether or not this information is available in the container, the job, or both....
	- suspend/resume
		-- These operations will initiate a suspension of the job's activity and resume job's activity
		-- Keep in mind with these things, we are "asking" the endpoint to do something -- that doesn't mean that it has to do something.
	- checkpoint/restart
		-- Can be out of band ways of providing information about checkpoint and restart (like, what image to use).
		-- Can you have the application checkpoint and keep on running?
			--- Yes, that happens all the time.
		-- Could implement migrate using this
	- terminate/signal
		-- Soft terminate of the job.
* Resource Model Discussion
	- might be best to take this to the list
	- history
	- two minute overview of CIM
	- Deciding which RM model to use isn't on the critical path
	- If you want to take CIM, it will take too long to come up with a complete schema
	- Can we agree that resource model is out of scope for this iteration of the specification?
		-- We need to say that there needs to be one, but we aren't going to decide on this now. 
	- To wrap up
		-- It clearly doesn't need to be in BES
		-- It clearly needs to be out there
* CreateActivityFromJSDL
	- as per document changes

* Telecons start two thursdays hence
	- morning US, afternoon UK, etc.
	- deadline by end of July
	- F2F co-located with OGSA F2F (plans are for early Sept. in LaHoya)