[Proposal] - Release Jenkins 2.7.4 on this week

202 views
Skip to first unread message

Oleg Nenashev

unread,
Sep 6, 2016, 2:21:01 AM9/6/16
to JenkinsCI Developers, Kohsuke Kawaguchi, Oliver Gondža, Daniel Beck
Hi,

This is just a follow-up to the discussion in https://groups.google.com/forum/#!topic/jenkinsci-dev/KY6AVXyTAAU.

In Jenkins 2.7.3 we have introduced a serious regression, which impacts detached plugin update flows on Windows:
This issue is pretty serious since we expect pretty much updates to 2.7.3 over next two months before 2.19.x gets stabilized.

I propose to release 2.7.4 with the fix. IMHO it's more preferable compared to reverting the JENKINS-37041 fix, because it's also a serious issue. The proposed fix is available in 3 releases, hence it is eligible for backporting.

WDYT?

Thanks in advance,
Oleg

P.S: I have +1 from Daniel Beck in the original thread

Oliver Gondža

unread,
Sep 6, 2016, 2:41:06 AM9/6/16
to Oleg Nenashev, JenkinsCI Developers, Kohsuke Kawaguchi, Daniel Beck
On 2016-09-06 08:20, Oleg Nenashev wrote:
> Hi,
>
> This is just a follow-up to the discussion in
> https://groups.google.com/forum/#!topic/jenkinsci-dev/KY6AVXyTAAU.
>
> In Jenkins 2.7.3 we have introduced a serious regression, which impacts
> detached plugin update flows on Windows:
>
> * JENKINS-37041 <https://issues.jenkins-ci.org/browse/JENKINS-37041>
> <https://github.com/jenkinsci/jenkins/commit/6ea7883fed497b68eec9f89ec89082edf1b39ce5>)
>
> * This fix in weeklys caused critical regression (JENKINS-37332
> <https://issues.jenkins-ci.org/browse/JENKINS-37332>)
> * Fix for this regression
> <https://github.com/jenkinsci/jenkins/pull/2516> has not been
> backported, the issue has been raised by a user in JENKINS-37332
> <https://issues.jenkins-ci.org/browse/JENKINS-37332>
>
> This issue is pretty serious since we expect pretty much updates to
> 2.7.3 over next two months before 2.19.x gets stabilized.
>
> I propose to release 2.7.4 with the fix
> <https://github.com/jenkinsci/jenkins/pull/2516>. IMHO it's more
> preferable compared to reverting the JENKINS-37041
> <https://issues.jenkins-ci.org/browse/JENKINS-37041> fix, because it's
> also a serious issue. The proposed fix is available in 3 releases, hence
> it is eligible for backporting.
>
> WDYT?

I agree that backporting the other fix seems less risky than reverting
the fix once released. I have backported that on top of stable-2.7 and
create temporary CI jobs testing it right now:

https://jenkins.ci.cloudbees.com/job/core/job/jenkins_lts_branch-2.7/
https://jenkins.ci.cloudbees.com/job/core/job/acceptance-test-harness-stable-2.7/

Do we want the release to be expedited skipping the rc? I am undecided
on the issue urgency as there seems to only single person complaining in
JENKINS-37332 for now.

--
oliver

Daniel Beck

unread,
Sep 6, 2016, 5:05:30 AM9/6/16
to Oliver Gondža, Oleg Nenashev, JenkinsCI Developers, Kohsuke Kawaguchi

> On 06.09.2016, at 08:41, Oliver Gondža <ogo...@gmail.com> wrote:
>
> Do we want the release to be expedited skipping the rc? I am undecided on the issue urgency as there seems to only single person complaining in JENKINS-37332 for now.

Unsurprising given the kind of bug. If this takes until the end of the week or so, that should be fine, but any longer and we could as well tell them to wait for 2.19.1.

ogondza

unread,
Sep 7, 2016, 3:59:42 AM9/7/16
to Jenkins Developers, ogo...@gmail.com, o.v.ne...@gmail.com, k...@kohsuke.org, m...@beckweb.net
The test results[1] looks good. I am pushing the RC at the moment for manual retesting. Once confirmed, I am ok to get that released.

FYI, I suggest[2] to have an explicit link type for issue that introduced another one. So hopefully, such situation will be lot less likely to reoccur.

[1] https://jenkins.ci.cloudbees.com/job/core/job/acceptance-test-harness-stable-2.7/1/
[2] https://issues.jenkins-ci.org/browse/INFRA-915

--
oliver

Oliver Gondža

unread,
Sep 8, 2016, 10:44:10 AM9/8/16
to Jenkins Developers, o.v.ne...@gmail.com, k...@kohsuke.org, m...@beckweb.net
Kohsuke, Please consider releasing this off-cycle version to plug the
hole in .3. I still do not feel all that strongly if this is necessary
but it would be nice to have LTS _without_ known regression to show on
Jenkins World ;)

--
oliver

Kohsuke Kawaguchi

unread,
Sep 8, 2016, 2:13:41 PM9/8/16
to Oliver Gondža, Jenkins Developers, o.v.ne...@gmail.com, m...@beckweb.net
Ack. I'm releasing this now.
Reply all
Reply to author
Forward
0 new messages