Open Stack

Here is the proposed schedule and adjusted ordering based on risk due to
code changes. Please let me know if you would like any of these dates
changed.  You will see that I have bumped modules that were upgrades to
later so that we can stop prior to this changes if we feel they are too
risky.  We should be safe due to zero puppet code changes up until the ssh
module.

Today 7/31
librarian - https://review.openstack.org/#/c/202763/
stdlib - https://review.openstack.org/#/c/203386/

Monday 8/3
concat - https://review.openstack.org/#/c/203387/
inifile - https://review.openstack.org/#/c/203395/

Tuesday 8/4
xinetd - https://review.openstack.org/#/c/203393/
ssh - https://review.openstack.org/#/c/203392/

Wednesday 8/5
ntp - https://review.openstack.org/#/c/203390/

Thursday 8/6
apache - https://review.openstack.org/#/c/203388/

Monday 8/10
firewall - https://review.openstack.org/#/c/203396/

This commit should be signed off by the teams who will ultimately be
responsible for it to ensure they are familiar with the workflows
involved[0]. It has no code changes, but may impact bug fixing so we may
want to wait until 8.0 for this. If not, it should be done last.
cinder - https://review.openstack.org/#/c/203394/

This cannot be merged until we get a fuel-infra mirror, but is a change
with no code changes so can be done whenever the mirror gets resolved.
apt - https://review.openstack.org/#/c/203389/

If at any point there are issues and we decide to revert back, they commits
just need to be reverted in reverse order and we can use the webui to do
so.  I will finish re-ordering the patches today, but we should be OK
through the ntp module at this time.

Thanks,
-Alex

[0] https://wiki.openstack.org/wiki/Fuel/Library_and_Upstream_Modules

On Fri, Jul 31, 2015 at 11:39 AM, Vladimir Kuklin <vkuklin at mirantis.com>
wrote:

> Okay, folks, we had a short meeting to synchronize our vision of how it
> should happen.
>
> We will start by merging least-invasive modules like stdlib today and then
> continue doing merges one by one in discrete manner and revert things
> immediately if something goes wrong.
>
> So there is a list of action items:
>
> Alex Schultz will send a schedule of which modules will be merged on which
> week and ensure that core reviewers know which commits they should merge
> when either by keeping W-1 on particular commits or by sharing the schedule
> in commit message so that noone can forget about it or maybe some other
> conveinent method that he can invent.
>
> I will remove my -2 for the inital librarian commit.
>
>
> Thanks everyone for the collaboration and not calling me a selfish lunatic
> :-)
>
>
> On Fri, Jul 31, 2015 at 6:29 PM, Mike Scherbakov <mscherbakov at mirantis.com
> > wrote:
>
>> Vladimir,
>> can you please elaborate on "such invasive changes"?
>>
>> There was a plan developed, including risk mitigation, etc. - like do
>> 'grep -r' to check, and revert the change all together right away if we see
>> regression. So far, everyone was aligned with the plan. It was discussed
>> yesterday during IRC meeting [1]. Again, no one had objections.
>>
>> Please provide your concerns, explain your opinion in more details. I'd
>> like other core reviewers to jump in here and reply. If you need details of
>> the approach, please jump in a call with Alex Schultz.
>>
>> Thank you,
>>
>> [1]
>> http://eavesdrop.openstack.org/meetings/fuel/2015/fuel.2015-07-30-16.00.html
>>
>> On Fri, Jul 31, 2015 at 5:52 AM Vladimir Kuklin <vkuklin at mirantis.com>
>> wrote:
>>
>>> Folks
>>>
>>> I do actively support our initiative to use librarian to be as close as
>>> possible to upstream, but let's not merge such invasive changes until we
>>> announce Hard Code Freeze and create stable/7.0 branch. So far I put my -2
>>> onto the first commit in the chain. Let's get through 7.0 and then land
>>> this code in the master as early as possible after HCF.
>>>
>>>
>>> On Fri, Jul 31, 2015 at 3:21 PM, Aleksandra Fedorova <
>>> afedorova at mirantis.com> wrote:
>>>
>>>> > So far CI has been successful on all of these
>>>> changes, and bvt is currently running.
>>>>
>>>> Small update - BVT test passed.
>>>>
>>>> On Fri, Jul 31, 2015 at 6:27 AM, Alex Schultz <aschultz at mirantis.com>
>>>> wrote:
>>>> > Hey everyone,
>>>> >
>>>> > During on the fuel meeting today we discussed the librarian changes
>>>> and
>>>> > their status.
>>>> > As part of this work, the wiki page was updated and a first attempt at
>>>> > migrating the
>>>> > following modules has been completed pending merge:
>>>> >
>>>> > stdlib
>>>> > concat
>>>> > inifile
>>>> > ssh
>>>> > ntp
>>>> > apache
>>>> > firewall
>>>> > xinetd
>>>> > cinder
>>>> > apt*
>>>> >
>>>> > It should be noted that apt is currently blocked by the lack of a
>>>> mirror so
>>>> > while it has
>>>> > been prepared, it should not be merged at this time.
>>>> >
>>>> > As part of this migration we are doing two things. The first is an
>>>> update to
>>>> > the build
>>>> > process that is included as part of the initial librarian[0] patch.
>>>> The
>>>> > other patches
>>>> > consist of the actual module code changes.
>>>> >
>>>> > Here is the list of the diffs for each change so that it can be
>>>> reviewed and
>>>> > people can
>>>> > raise concerns if there are any with this change. As part of the
>>>> migration,
>>>> > I inspected
>>>> > the code and file differences for each module to determine how much
>>>> impact
>>>> > they might
>>>> > have.  I chose the list of modules based on their minimal differences
>>>> from
>>>> > the upstream
>>>> > or if they already had our forked differences rolled into a newer
>>>> version of
>>>> > the module.
>>>> > For this list, I took the current stable iso (#110) and rebased the
>>>> changes
>>>> > on top of this
>>>> > to create a custom iso with just the librarian changes. We have
>>>> kicked off a
>>>> > bvt_2 test for
>>>> > the custom iso as well. From this iso I have extracted the
>>>> fuel-library
>>>> > package from both
>>>> > of these isos and exploded the fuel-library folder structure to do the
>>>> > diffs.
>>>> >
>>>> > Code Changes:
>>>> >
>>>> > For stdlib, the only differences are related to git, travis or
>>>> fixtures[1].
>>>> > There are no
>>>> > puppet code changes as part of the librarian migration.
>>>> >
>>>> > For concat, the only differences were a git folder and in a custom
>>>> change to
>>>> > the spec tests[2].
>>>> > The test difference[3], was a change we made because it was failing
>>>> our
>>>> > syntax checker.
>>>> > This change has been included in a newer version of concat (1.2.4)
>>>> but are
>>>> > not necessary
>>>> > when the module gets moved to be included via librarian.
>>>> >
>>>> > For inifile, the only difference is the addition of git and metadata
>>>> > files[4].
>>>> >
>>>> > For ssh, the only difference is a single line to have the config
>>>> notify
>>>> > service[5]. This
>>>> > difference is already covered by another file and is not needed[6].
>>>> >
>>>> > For ntp, this change introduces more code changes[7] because we are
>>>> updating
>>>> > the module
>>>> > to the 4.0.0 version because of previous extending of functionality
>>>> that is
>>>> > now covered by
>>>> > 4.0.0 vs 3.3.0[8]. The changes in our fork were upstreamed and are
>>>> include
>>>> > in 4.0.0.
>>>> >
>>>> > For apache, this change includes an upgrade from 1.2.0 to
>>>> 1.3.0[9][10]. Our
>>>> > fork had a
>>>> > customization made which was contributed upstream.
>>>> > (apache::mod::proxy_connect)
>>>> >
>>>> > For firewall, this change also includes an upgrade from 1.0.2 to
>>>> 1.2.0[11]
>>>> > as our fork had
>>>> > mac supported added[12] in which is now covered upstream.
>>>> >
>>>> > For xinetd, the only change was the addition of a .git folder and a
>>>> > .gitignore with librarian.
>>>> >
>>>> > For cinder, the only change was the addition of .git, .gitignore, and
>>>> > .gitreview.
>>>> >
>>>> > Once we can get the apt mirror created, the only change for that is
>>>> also the
>>>> > addition of
>>>> > .git.
>>>> >
>>>> >
>>>> > If there are any of these upgrades/changes that we do not want to
>>>> tackle
>>>> > right now, I can
>>>> > adjust the review order such that it can be skipped for now.  Please
>>>> take
>>>> > some time to
>>>> > review these changes and raise concerns.  So far CI has been
>>>> successful on
>>>> > all of these
>>>> > changes, and bvt is currently running.
>>>> >
>>>> > Also please take some time to review the changes themselves:
>>>> >
>>>> https://review.openstack.org/#/q/status:open+project:stackforge/fuel-library+branch:master+topic:bp/fuel-puppet-librarian,n,z
>>>> >
>>>> > Please raise any concerns as quickly as possible as this is the last
>>>> call
>>>> > for objections
>>>> > for these reviews.  This has been talked about extensively and these
>>>> reviews
>>>> > have
>>>> > been available for several weeks now.
>>>> >
>>>> > Thanks,
>>>> > -Alex
>>>> >
>>>> >
>>>> > [0] https://review.openstack.org/#/c/202763/
>>>> > [1] http://paste.openstack.org/show/406523/
>>>> > [2] http://paste.openstack.org/show/406524/
>>>> > [3] http://paste.openstack.org/show/406525/
>>>> > [4] http://paste.openstack.org/show/406526/
>>>> > [5] http://paste.openstack.org/show/406527/
>>>> > [6]
>>>> >
>>>> https://github.com/saz/puppet-ssh/blob/v2.4.0/manifests/server/config.pp#L9
>>>> > [7] http://paste.openstack.org/show/406536/
>>>> > [8]
>>>> https://github.com/puppetlabs/puppetlabs-ntp/compare/3.3.0...4.0.0
>>>> > [9] http://paste.openstack.org/show/406538/
>>>> > [10]
>>>> https://github.com/puppetlabs/puppetlabs-apache/compare/1.2.0...1.3.0
>>>> > [11]
>>>> https://github.com/puppetlabs/puppetlabs-firewall/compare/1.0.2...1.2.0
>>>> > [12] https://review.openstack.org/#/c/92167/
>>>> >
>>>> >
>>>> __________________________________________________________________________
>>>> > OpenStack Development Mailing List (not for usage questions)
>>>> > Unsubscribe:
>>>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>>>> > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>> >
>>>>
>>>>
>>>>
>>>> --
>>>> Aleksandra Fedorova
>>>> Fuel CI Engineer
>>>> bookwar
>>>>
>>>>
>>>> __________________________________________________________________________
>>>> OpenStack Development Mailing List (not for usage questions)
>>>> Unsubscribe:
>>>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>>
>>>
>>>
>>>
>>> --
>>> Yours Faithfully,
>>> Vladimir Kuklin,
>>> Fuel Library Tech Lead,
>>> Mirantis, Inc.
>>> +7 (495) 640-49-04
>>> +7 (926) 702-39-68
>>> Skype kuklinvv
>>> 35bk3, Vorontsovskaya Str.
>>> Moscow, Russia,
>>> www.mirantis.com <http://www.mirantis.ru/>
>>> www.mirantis.ru
>>> vkuklin at mirantis.com
>>>
>>> __________________________________________________________________________
>>> OpenStack Development Mailing List (not for usage questions)
>>> Unsubscribe:
>>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>>
>> --
>> Mike Scherbakov
>> #mihgen
>>
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe:
>> OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>
>
> --
> Yours Faithfully,
> Vladimir Kuklin,
> Fuel Library Tech Lead,
> Mirantis, Inc.
> +7 (495) 640-49-04
> +7 (926) 702-39-68
> Skype kuklinvv
> 35bk3, Vorontsovskaya Str.
> Moscow, Russia,
> www.mirantis.com <http://www.mirantis.ru/>
> www.mirantis.ru
> vkuklin at mirantis.com
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150731/2fd7e5f2/attachment.html>