[Iwg-arbitration-committee] Emulex Arbitration Issue

Mon May 27 14:24:23 PDT 2013

Bob, thanks, this is all extremely helpful and informative. I hope you weren’t bothered, much less insulted by my earlier comment. Nothing along the lines of that was intended

Regards, Jim

From: Bob Noseworthy [mailto:ren at iol.unh.edu]
Sent: Monday, May 27, 2013 8:49 AM
To: Ryan, Jim; iwg-arbitration-committee at lists.openfabrics.org
Subject: RE: [Iwg-arbitration-committee] Emulex Arbitration Issue

Hi Jim.   If the result was erroneously logged as a fail, or just a "refer to comments" I would agree.

Instead, the result was a Pass, logged as a "Pass with comments", which the report result key shows is still a "Pass",  yet Emulex was apparently concerned with anything that is not a pure "Pass".
  Arbitrating a passing result is unusual (a first I believe)
  It is our custom to highlight any deviation from the OFA test plan as a "Pass with comments" or "Refer to comments" as appropriate.  I'm sure you would agree that deviations from the test plan should not be undocumented.
  The fact that Emulex was concerned about the perception of a "Pass with comments" vs a "Pass" is actually a positive sign of their desire to market/promote the result - though traditionally if there were such vendor concerns re: wording/etc, they would be caught during informal vendor report reviewing before official report issuance (where we typically ask for things like product identification, etc to be reviewed prior to formal report issuance - which would also be a waste of everyone's time if taken through arbitration)
  For the next reports we'll try to be clearer with Emulex,  IBM, etc that their early review and comment is welcome so non-technical issues like this can be resolved prior to formal report issuance thus requiring arbitration.

BR
  Bob

"Ryan, Jim" <jim.ryan at intel.com<mailto:jim.ryan at intel.com>> wrote:
Pardon my uninformed comment, but this suggests there has to be a better way. This process required Emulex to take issue with something that wasn’t their fault – IMO an unnecessary and unwarranted imposition on their time

Jim

From: iwg-arbitration-committee-bounces at lists.openfabrics.org<mailto:iwg-arbitration-committee-bounces at lists.openfabrics.org> [mailto:iwg-arbitration-committee-bounces at lists.openfabrics.org] On Behalf Of Bob Noseworthy
Sent: Friday, May 24, 2013 12:16 PM
To: iwg-arbitration-committee at lists.openfabrics.org<mailto:iwg-arbitration-committee at lists.openfabrics.org>
Subject: Re: [Iwg-arbitration-committee] Emulex Arbitration Issue

Hi all,
  Just closing the loop from my perspective.   I'm fine with the request.  The purpose of the "Pass with Comments" was to draw necessary attention to the deviation from the TEST PLAN,  not to highlight an issue with the PRODUCT (there is none).    The reason why we typically flag a result such as this as "Pass with Comments" is to make sure the result gets a little more attention than a mere glance - and also to ensure that the deviation from the test plan is noticed (and, if necessary, challenged if anyone believes the deviation was NOT warranted).
  Jeff certainly makes the case that the change is warranted and clearly wishes to have a 'cleaner' report,  which I have no objection to.

Unless there's objection,  we shall reissue the report as a "Pass" (leaving the comment as is)

Have a great weekend all,

- Bob Noseworthy
  Chief Engineer / Technical Sherpa
  +1-909-891-0090 {unified phone number for office, cell, etc}
  +1-603-862-0090 {IOL Main number-associate this with any shipments}
  University of New Hampshire's InterOperability Laboratory (UNH-IOL)
On 5/24/2013 11:08 AM, Rupert Dance wrote:
Hello all,

The arbitration submission period for the February 2013 Logo GA Event officially ends May 24th 2013 at 5:00 PM Eastern but since we are having an Interoperability program meeting today, I wanted to take the opportunity to review the only arbitration claim we have received so far for the February Logo GA Event.  You have requested to be included in the Feb 2013 Arbitration committee and so  I need to remind you that you are under NDA and none of the information revealed during this process should be shared with anyone outside of the committee members. Here is the list of committee members for this event:

·         Bob Noseworthy – UNH-IOL

·         Jim Ryan – Intel

·         Paul Grun - Cray

·         Rupert Dance S- Software Forge

Start of Arbitration Claim:

Jeff Kopko has submitted an arbitration request on the results of the February 2013 Logo Event Report. Here are the details of the arbitration request:

________________________________

Emulex would like have our results for MPI reviewed.

Current Reported Results for test 13.7 MPI  = Pass with Comments

Desired Results = Pass

Reasoning:

The issue is that the Global Routing Header will not be set if rdmacm is not specified for openmpi.  GRH is required for RoCE.

Spec Details.

In current OFED stack implementation, to comply to RDMA RoCE Annex specification, To satisfy requirement CA16-4. And CA16-22, rdmacm is necessary.

________________________________

I have attached the following documents for us to review in support of this arbitration claim:

·         Emulex Logo Report: OFILG_2013-Feb_Logo-Report_Emulex_Beta_v1.1

·         RoCE Annex – see pages 3 and 9

·         Interop Test Plan for May Logo GA Event: OFA-IWG Interoperability Test Plan-v1.48-v4

Here is the call in information for the meeting today:

Friday, May 24, 2013, 09:00 AM US Pacific Time
916-356-2663, Bridge: 4, Passcode: 1639619

_______________________________________________

iwg-arbitration-committee mailing list

iwg-arbitration-committee at lists.openfabrics.org<mailto:iwg-arbitration-committee at lists.openfabrics.org>

http://lists.openfabrics.org/cgi-bin/mailman/listinfo/iwg-arbitration-committee

--
Sent from Kaiten Mail. Please excuse my brevity.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/iwg-arbitration-committee/attachments/20130527/f1fa075b/attachment.html>