[ewg] OFED Updates

[Shuichi Ihara]

Hi Rupert,

This is what I asked you a couple of month ago, but I'm still really confused even on 3.5-2.

_version and _release are defined in 3.5-1-RC1's compat-rdma.spec below.
%{!?_version: %define _version 3.5}
%{!?_release: %define _release OFED.3.5.1}

But, with 3.5-2 today...

%{!?_version: %define _version 3.5}
%{!?_release: %define _release 1}

Why _release number is 1 in OFED-3.5-2? it is wrong and it should be 2 shouldn't it?
What is exactly correct "_release" number?

Please advise. I just filed this on bugzilla, anyway.

Thanks
Ihara

On Jun 27, 2013, at 5:02 PM, Shuichi Ihara <sihara at ddn.com> wrote:

> 
> Rupert, 
> 
> Thanks clarification on this.
> My question was versioning for compiled compat-rdma RPMs below. The version will be fixed and
> changed with correct version in the next RC build?
> 
>> compat-rdma-3.5-2.6.32_358.6.2.el6.x86_64.x86_64.rpm
>> compat-rdma-devel-3.5-2.6.32_358.6.2.el6.x86_64.x86_64.rpm
> 
> I can expect compat-rdma-3.5-1-2.6.xx_xxx, instead of compat-rdma-3.5-2.6.xx_xxx, correct?
> 
> Thanks
> Ihara
> 
> On Jun 27, 2013, at 4:16 AM, Rupert Dance <rsdance at soft-forge.com> wrote:
> 
>> Hello Ihara,
>> 
>> I discussed this with Vlad before we did these builds and here was his
>> reasoning for suggesting these changes:
>> 
>> "We should change versioning from "3.5.x" to "3.5-x". Current versioning may
>> be confusing as OFED-3.5.1/2 is not based on linux-3.5.1/2 but on
>> linux-3.5."
>> 
>> Thanks
>> 
>> Rupert
>> 
>> -----Original Message-----
>> From: Shuichi Ihara [mailto:sihara at ddn.com] 
>> Sent: Wednesday, June 26, 2013 7:26 AM
>> To: Rupert Dance
>> Cc: <ewg at lists.openfabrics.org>; <ofalab at iol.unh.edu>; <ofs at soft-forge.com>
>> Subject: Re: [ewg] OFED Updates
>> 
>> 
>> Hello,
>> 
>> I have an confusion a little bit.
>> 
>>> ·         OFED 3.5-1 not OFED 3.5.1
>> 
>> I built OFED-3.5-1-RC1.tgz against RHEL6.4 kernel and the compile was fine.
>> However, the compiled compat-rdma and compat-rdma-devel rpms's version is
>> still 3.5, not 3.5-1.
>> 
>> compat-rdma-3.5-2.6.32_358.6.2.el6.x86_64.x86_64.rpm
>> compat-rdma-devel-3.5-2.6.32_358.6.2.el6.x86_64.x86_64.rpm
>> 
>> _version in compat-rdma.spec should be 3.5-1, instead of 3.5, shouldn't it?
>> 
>> %{!?_version: %define _version 3.5}
>> %{!?_release: %define _release OFED.3.5.1}
>> 
>> Please advise.
>> 
>> Thanks
>> Ihara
>> 
>> On Jun 25, 2013, at 5:08 AM, Rupert Dance <rsdance at soft-forge.com> wrote:
>> 
>>> Meeting minutes and notes from the OFA EWG Meeting held on 6/24/2013
>>> 
>>> 
>>> 
>>> Attendees
>>> 
>>> ·         Steve Wise - Chelsio
>>> 
>>> ·         Robert Woodruff - Intel
>>> 
>>> ·         Tatyana Nikolova - Intel
>>> 
>>> ·         Tom Elken - Intel
>>> 
>>> ·         Rupert Dance - Software Forge
>>> 
>>> 
>>> 
>>> Updates
>>> 
>>> Vlad suggested via email that we name OFED 3.5 point releases using a 
>>> hyphen rather than a dot release. His reasoning is that the EWG is 
>>> working to insure that we stay synchronized with kernel.org and there 
>>> is no such thing a 3.5.1 or 3.5.2 kernel. So the EWG "point" releases 
>>> will follow thins
>>> format:
>>> 
>>> 
>>> 
>>> ·         OFED 3.5-1 not OFED 3.5.1
>>> 
>>> ·         OFED 3.5-2 not OFED 3.5.2
>>> 
>>> ·         etc
>>> 
>>> 
>>> 
>>> OFED Status
>>> 
>>> OFED 3.5-1
>>> 
>>> ·         OFED 3.5.1 RC1 released on 6/24/2013
>>> 
>>> o   This release is identical to OFED 3.5 except that it adds support for
>>> RHEL 6.4. There are no new packages.
>>> 
>>> o
>> http://www.openfabrics.org/downloads/OFED/ofed-3.5-1/OFED-3.5-1-RC1.tgz 
>>> 
>>> ·         OFED 3.5.1 RC2
>>> 
>>> o   Tom Elken from Intel mentioned that they have found an issue with the
>>> Intel True Scale PSM package with RHEL 6.4. This affects both RHEL 6.3 
>>> and RHEL 6.4. Tom wants to review the status and provide a fix which 
>>> will be available for RC2. There is a current work around using 
>>> ZStream kernels
>>> 
>>> o   Pending the Intel fix - goal is July 15th
>>> 
>>> ·         OFED 3.5.1 GA 
>>> 
>>> o   After fixes have been applied in RC2 we will do a GA. Goal is end of
>>> July.
>>> 
>>> 
>>> 
>>> OFED 3.5.2
>>> 
>>> ·         OFED 3.5.2 Daily Build released on 6/24/2013
>>> 
>>> o   This release support RHEL 6.4
>>> 
>>> o
>>> http://www.openfabrics.org/downloads/OFED/ofed-3.5-2-daily/OFED-3.5-2-
>>> 201306
>>> 24-0804.tgz
>>> 
>>> o   New packages added
>>> 
>>> §  dapl-2.0.37
>>> 
>>> §  libcxgb4-1.3.0
>>> 
>>> §  libibverbs-1.1.7
>>> 
>>> §  libmlx4-1.0.5
>>> 
>>> §  librdmacm-1.0.17
>>> 
>>> §  mstflint-3.0-0.5
>>> 
>>> §  opensm-3.3.16
>>> 
>>> §  perftest-2.0-0.50
>>> 
>>> ·         OFED 3.5.2 RC1
>>> 
>>> o   Woody mentioned that SLES-11 SP3 is about to be released and he would
>>> like to include support for this OS. His team is already working on an 
>>> update for their MIC OFED version so he will provide the patches to do 
>>> this and we will include it in the OFED 3.5-2 Daily build before we do an
>> RC.
>>> 
>>> o   2 weeks after Tom Elken, Steve Wise, Tatyana and others have a chance
>> to
>>> review and communicate their plans with their teams.
>>> 
>>> ·         OFED 3.5.2 GA - target is early August 
>>> 
>>> 
>>> 
>>> OFED 3.11 or 3.12
>>> 
>>> ·         Goal is to start work in August 2013 and RC in October GA in
>> time
>>> for SC 2013
>>> 
>>> 
>>> 
>>> Open Issues
>>> 
>>> Mellanox has requested that Software Forge (SFI) take over the 
>>> maintenance of OFED and so we have been making a transition from Vlad 
>>> being the maintainer to Rupert maintaining this project. Please notify 
>>> me of any updated packages and issues that you discover during 
>>> testing. There are two addresses that may be used:
>>> 
>>> ·         Rupert: rsdance at soft-forge.com
>>> 
>>> ·         SFI OFA Team: ofs at soft-forge.com 
>>> 
>>> 
>>> 
>>> Thanks
>>> 
>>> 
>>> 
>>> Rupert
>>> 
>>> 
>>> 
>>> <OFED-3.5-1-RC1_release_notes.txt><OFED-3.5-2_release_notes.txt>______
>>> _________________________________________
>>> ewg mailing list
>>> ewg at lists.openfabrics.org
>>> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
>> 
>> 
>