[DRMAA-WG] Conference call - May 11th - 19:00 UTC

[Peter Tröger]

Dear all,

the next DRMAA conf call is scheduled for May 11th, 19:00 UTC. We meet
on Skype, please find me under my user name "potsdam_pit".

Preliminary meeting agenda:

1. Meeting secretary for this meeting?
2. Review of advance reservation research work by Mariusz
(http://bit.ly/kLwNe7)
3. Discussion of usersACL proposal by Mariusz
4. Discussion of NOW constant proposal
5. Other issues in DRMAAv2 Draft 4 (see attachment)

Best regards,
Peter.





-------------- next part --------------
A non-text attachment was scrubbed...
Name: drmaav2_draft4_annotated.pdf
Type: application/pdf
Size: 755322 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/drmaa-wg/attachments/20110511/7d520c6d/attachment-0001.pdf

[Peter Tröger]

Participants: Daniel, Andre (SAGA), Peter, Mariusz (partially), Roger
(partially)

Post by Peter Tröger
1. Meeting secretary for this meeting?

Peter.

Post by Peter Tröger
2. Review of advance reservation research work by Mariusz
(http://bit.ly/kLwNe7)
3. Discussion of usersACL proposal by Mariusz
4. Discussion of NOW constant proposal

Dropped due to problems with the connection to Mariusz.

Post by Peter Tröger
5. Other issues in DRMAAv2 Draft 4 (see attachment)

Line 101 - Use term "process" for instances of the executable in a
parallel job

Page 14 - no use case for "comment" attribute from SAGA perspective, no
support in Grid Engine

Line 381 - Caching behavior for job information completely unspecified,
could be non-existent for distributed DRM systems; reformulate to make
no promises on JobInfo availability in terminal states

Line 729 - Discussion to either remove maxSlots or make it optional;
expected use cases are billing limitations and parallel job scalability
considerations - serious enough to make it an optional attribute; if not
supported, maxSlots should be assumed to be equal to minSlots.

Line 769 - VIRTUAL_MEMORY violation should lead to job FAILURE state.

Page 26 - Discussion about accumulated resource limits in case of
parallel jobs (e.g. CPU_TIME); semantics are completely
implementation-specific, state like this

- Proposal for DRMAA F2F meeting in Potsdam, Peter sets up Doodle poll