[ogsa-hpcp-wg] Minutes for OGSA HPC Profile telecon (Aug 18 2006)

[humphrey at cs.virginia.edu]

[[ these will be uploaded to sourceforge shortly ]]

OGSA HPC Profile Minutes – Fri Aug 18 2006

Participants:
	Marty Humphrey	 (University of Virginia)
	Marvin Theimer 	(MSFT)
	Glenn Wasson 	(University of Virginia)
	Donal Fellows 	(University of Manchester)

Minutes: Marty Humphrey

* Summary of Actions:

	AI-HPCP-0818a: Donal Fellow to review HPC Profile Application.
	AI-HPCP-0818b: Marvin Theimer will add paragraph to HPC Basic Profile
			stating: "all other elements of the JSDL 1.0 MAY appear
			but MAY	also result in a 'not implemented' fault, but
			only the list here must be supported"
	AI-HPCP-0818c: Marvin Theimer will add first draft of section 4 to
			HPC Basic Profile that restricts BES operations to
			singletons

* Previous minutes: NONE (this is the first teleconference)

* Review of Actions: NO Previous Actions (this is the first teleconference)

* Discuss the State of BES
--------------------------
	Marvin Theimer has grabbed the pen on BES doc (plan: to send it out  
by Wednesday – including
		refactoring: [1] arbitrary clients vs [2] sys admins; management  
would only have “start”
		and “stop”) (ie. Put “query” into the “factory”); Main thing: add  
fleshed-out extended
		state model (current draft: “there WILL be an extended state model,  
and here’s the basic one").
		extend data-staging example so that Ian F and Dave S can see the  
details and/or are satisfied
	Glenn: factoring into two port types (or three: activity, factory,  
management: but Activity
		is not current in it), not going to be in an Appendix
	Interesting issue: what ultimately shows up in the activity  
management interface
	Glenn: seemed like from the base case that the only thing you would  
do from the interface is “Cancel”
	Marvin recapped his email to BES re: three renderings
	Claim: basics of BES are shaping up to be what we expected (shaping  
up in a good way)

* HPC Profile Application Extension - JSDL extension for describing  
executables, etc.
-------------------------------------------------------------------------------------
	fairly far along, are we happy with what’s in there? Consensus is  
yes, we’re happy.
	Next step is to have someone other than Chris to go through it to see  
if there are any
		details that they don’t like.
	Donal: the document needs: which bits are the normative MUSTs

		AI-HPCP-0818a: Donal Fellow to review HPC Profile Application. Specifically
			to ensure that normative text use is correct. Also, adding
			normative text. Hopefully done before the next JSDL call on Wed.
			Goal: to get on the agenda for JSDL. Donal will contact Andreas to
			get on the JSDL WG agenda.

* HPC Basic Profile
-------------------
	Issue: what’s the style of writing this doc? How do the restrictions work?
	Previous discussion: do we really want to create a doc that only has MUSTs?
	Currently, the style is: "These are the elements that MUST be  
supported and all other
		aspects COULD be supported but you also might get a fault back."  
Consensus is
		that this is the best style.
	Missing: another paragraph : "all other elements of the JSDL 1.0 MAY  
appear but MAY
		also result in a 'not implemented' fault, but only the list here  
must be supported."
		Example: file system element.
	Glenn: I don’t like the Strike-out approach because you need both docs.

		AI-HPCP-0818b: Marvin Theimer will add paragraph stating: "all other  
elements of
			the JSDL 1.0 MAY appear but MAY	also result in a 'not implemented'
			fault, but only the list here must be supported." Donal says: probably
			want to add more text with regard to range values. E.g., it must
			support “JSDL exact” but the “exact” allows to specify what you mean
			by “Exact”. In the case of range values, you also want to specify
			to mandate the syntactic requirements of integer values. I.e.,
			set-value resources or single value resources. If NOT "set", then
			go with "exact".

	Marvin: Do we want to, in the base case, restrict total cpu count to  
be exact? COnsensus in
		call is that this is tricky, because it’s arguably not as “simple”  
as we would
		like. Consider putting this issue in a tracker.
	This leaves section 5 of BES
our only restriction is to eliminate  
required support for arrays.

		AI-HPCP-081c: Marvin Theimer will add first draft of section 4
			that restricts BES operations to singletons. This will
			result in our first complete draft. He will add in a
			first draft of the security section.

* Other news
------------
	Donal: At the end of the OGSA call yesterday, it was mentioned that  
EMS and data people are
		looking to do a combined use case, based on the HPC Profile. We all  
agreed on the
		call that this was great news and we're anxious to work with them.











----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.