[DRMAA-WG] DRMAA Version 2 - Final Document

Discussion:

[DRMAA-WG] DRMAA Version 2 - Final Document - Draft 1

Peter Tröger

2011-03-29 12:27:31 UTC

Dear all,

it took 3 months instead of 3 weeks, but we finally made it. Attached you can find the first official draft for the final DRMAA v2 specification.

I attached a preview for the official text, plus an annotated version with ToDo markers + explanations + line numbers. The latter should be used as foundation for all upcoming discussions.

We will start again with weekly (!) conf calls beginning from next week, until the remaining issues are solved. Please feel free to use the list for initial discussions, this saves time on the phone.

Best regards,
Peter.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: drmaav2_draft1.pdf
Type: application/pdf
Size: 473522 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/drmaa-wg/attachments/20110329/ae191cd3/attachment-0002.pdf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: drmaav2_draft1_annotated.pdf
Type: application/pdf
Size: 620455 bytes
Desc: not available
Url : http://www.ogf.org/pipermail/drmaa-wg/attachments/20110329/ae191cd3/attachment-0003.pdf
-------------- next part --------------

Daniel S. Katz

2011-03-29 14:52:27 UTC

Permalink

Hi,

I just want to say hello to the list, as a new OGF Area Co-Director for Applications.

I look forward to hearing about this discussion and seeing the progress towards the spec. I will be following the email list, but probably not following the phone calls.

If have a problem where you feel I can help, please let me know.

Thanks
Dan

Post by Peter TrÃ¶ger
Dear all,
it took 3 months instead of 3 weeks, but we finally made it. Attached you can find the first official draft for the final DRMAA v2 specification.
I attached a preview for the official text, plus an annotated version with ToDo markers + explanations + line numbers. The latter should be used as foundation for all upcoming discussions.
We will start again with weekly (!) conf calls beginning from next week, until the remaining issues are solved. Please feel free to use the list for initial discussions, this saves time on the phone.
Best regards,
Peter.
<drmaav2_draft1.pdf><drmaav2_draft1_annotated.pdf>
--
drmaa-wg mailing list
drmaa-wg at ogf.org
http://www.ogf.org/mailman/listinfo/drmaa-wg

--
Daniel S. Katz
University of Chicago
(773) 834-7186 (voice)
(773) 834-3700 (fax)
d.katz at ieee.org or dsk at ci.uchicago.edu
http://www.ci.uchicago.edu/~dsk/

Peter Tröger

2011-03-29 19:56:31 UTC

Permalink

Dear Daniel,

congratulations, seems like I missed the announcement for the position change. I left out some OGF conferences in the past, sorry for this ...

As you might have heard from Steven, DRMAA has the typical high / low activity phases of a OGF working group. We were really busy with pushing the DRMAA version 2 development in 2010, including meetings with SAGA, OCCI and different user communities. DRMAA has (or should I say had) the advantage of a large end user community with Sun Grid Engine, which lead to many other adoptions of the spec (check drmaa.org). The second version basically considers everything we got as feedback for DRMAA in the last years.

It would be great if you could help us a little bit to promote the new version inside of OGF when the public comment period is over. SAGA and OCCI are already in close contact with us, OGSA-BES and JSDL are at least mentioned. Spreading the word inside and outside of OGF will become crucial to hopefully reach the same level of uptake as with the first version.

Thanks and best regards,
Peter.

Post by Daniel S. Katz
Hi,
I just want to say hello to the list, as a new OGF Area Co-Director for Applications.
I look forward to hearing about this discussion and seeing the progress towards the spec. I will be following the email list, but probably not following the phone calls.
If have a problem where you feel I can help, please let me know.
Thanks
Dan

--
Daniel S. Katz
University of Chicago
(773) 834-7186 (voice)
(773) 834-3700 (fax)
d.katz at ieee.org or dsk at ci.uchicago.edu
http://www.ci.uchicago.edu/~dsk/

Daniel S. Katz

2011-03-29 20:01:40 UTC

Permalink

See: http://ogf.org/News/news.php?id=144

Regarding DRMAA, I see it as a useful and well-used standard. I will be happy to help promote the new version.

Dan

Post by Peter TrÃ¶ger
Dear Daniel,
congratulations, seems like I missed the announcement for the position change. I left out some OGF conferences in the past, sorry for this ...
As you might have heard from Steven, DRMAA has the typical high / low activity phases of a OGF working group. We were really busy with pushing the DRMAA version 2 development in 2010, including meetings with SAGA, OCCI and different user communities. DRMAA has (or should I say had) the advantage of a large end user community with Sun Grid Engine, which lead to many other adoptions of the spec (check drmaa.org). The second version basically considers everything we got as feedback for DRMAA in the last years.
It would be great if you could help us a little bit to promote the new version inside of OGF when the public comment period is over. SAGA and OCCI are already in close contact with us, OGSA-BES and JSDL are at least mentioned. Spreading the word inside and outside of OGF will become crucial to hopefully reach the same level of uptake as with the first version.
Thanks and best regards,
Peter.

--
Daniel S. Katz
University of Chicago
(773) 834-7186 (voice)
(773) 834-3700 (fax)
d.katz at ieee.org or dsk at ci.uchicago.edu
http://www.ci.uchicago.edu/~dsk/

Daniel S. Katz

2011-03-31 00:03:31 UTC

Permalink

Some comments from a scan of the document:

I would like a rational section for this document - something that says what the issue with DRMAA v1 were, and what changes have been made in v2 to address those issues - I know there are footnotes in the doc that talk about differences, but it would be good to have all the differences in one place, along with the reasons.

Sun Grid Engine is probably renamed just Grid Engine or Univa Grid Engine now - not sure, but this should be checked.

line 80 - stake holders should be stakeholders
line 81 - which should be that
line 86 - not clear what resp. means
line 100 - 2nd paragraph - remove "itself"
line 122 - API's should be APIs
line 149 and 150 - change to "with a resolution at least to seconds"
line 151 and 152 - change which to that
line 152+ - A line for UNSET should follow.
line 156 - another "resp." that I don't understand.
line 159+ call-by-value has the wrong initial quote - probably a LaTeX error in not using `` - this seems to be a general problem, and a global fix should be applied
line 160 footnote - change "from DRMAA" to "out of DRMAA"
line 161 - section 4.1 - are there other OSs that should be added? newer windows? Cray? (maybe this is Linux?)
line 233 - define Kibibyte - maybe define all units somewhere?
line 247 - are there systems that have a wall-clock time limit in their schedulers? I'm familiar with limits of run-time, but these are usually in CPU time, not wall-clock time. I don't know of a wall-time limit that includes time in the queue, but perhaps I've missed something.
line 481 - why would they be defined otherwise? This seems odd.
line 539-540 - I'm not sure what "according race conditions." means
line 541, 546 - more "resp." this continues in other places...
line 648 - should be "is not allowed."
line 650 - In case should be "In this case".
line 652 - "information" should be deleted or changed to another word. "Every ... information" isn't grammatical English.
line 694 - "allows to" should be "allows one to"
line 863 - how should the checkpointability of a job be denoted?
line 917 - of should be or
line 953 - It's not clear what dictionary is referred to here.
line 1051 - beside should be "in addition to"
line 1054 - "allows to" should be "allows one to"
line 1095 - "allow tho" should be "allows one to"
line 1156 - DRM should be "a DRM"
line 1195 - ending should be end
line 1240 table - "were the" should be "where the"
line 1353 - "return also" should be "also return"
line 1371 box - I'm not sure what "iterator-alike" means. Is this simply "iterator-like"?
line 1624 - that should be who
line 1629 - ", however" should be "; however"

--
Daniel S. Katz
University of Chicago
(773) 834-7186 (voice)
(773) 834-3700 (fax)
d.katz at ieee.org or dsk at ci.uchicago.edu
http://www.ci.uchicago.edu/~dsk/
--
Daniel S. Katz
University of Chicago
(773) 834-7186 (voice)
(773) 834-3700 (fax)
d.katz at ieee.org or dsk at ci.uchicago.edu
http://www.ci.uchicago.edu/~dsk/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.ogf.org/pipermail/drmaa-wg/attachments/20110330/155b29b7/attachment.html

Andre Merzky

2011-03-31 20:31:11 UTC

Permalink

Hi Peter,

I only skimmed through the document, but promise to read it in more
detail once it enters public comment! So, for now only two small
comments:

- In table 3, all DRMAA job states which have no entry in the SAGA
column would map to the "Running" state, with a DRMAA state-detail.
For example, the DRMAA state "REQUEUED" would map to SAGA's "Running"
state, with the SAGA state-detail set to "drmaa:REQUEUED".

I am less sure about the missing values in BES, but I think you
should try to complete the table for BES as well. I would assume that
all missing states get mapped to BES's "Pending" state, with no
condition on the state-detail string.

- The contact information for some contributors seem incomplete.

Cheers, Andre.

--
So much time, so little to do...
[Garfield]