Jenkins 2.413 as next LTS baseline?

[Alexander Brandes]

Hey everyone,

I would recommend choosing 2.413 as baseline for the next LTS, as it finalizes fixing a major regression from 2.409.

An alternative would be 2.412, and backport the fixes for JENKINS-71553.

Despite the obvious hijacked voting results for 2.413 on the changelog, I would favor it nonetheless, given the lack of reports outside voting.

Thoughts? Comments?

Br,

Alex

[Mark Waite]

On Wednesday, July 12, 2023 at 7:45:07 AM UTC-6 Alexander Brandes wrote:

Hey everyone,

I would recommend choosing 2.413 as baseline for the next LTS, as it finalizes fixing a major regression from 2.409.

I think that 2.413 is a very safe choice as the next LTS baseline.  It has good ratings and the preceding releases also have good ratings.

I think that we should consider 2.414 even a little bit above 2.413 because it includes the improvements for the safe restart process and has all the benefits of 2.413.  My personal preference is for 2.414 with the assurance that if there are surprises in the safe restart improvements, we'll backport the fixes.

 

An alternative would be 2.412, and backport the fixes for JENKINS-71553.

I don't see anything in Jira reports, mailing lists, or on community.jenkins.io that makes me think that 2.412 would be a better choice than 2.413 or 2.414.  I prefer 2.414 as my first choice and 2.413 as my second choice, with 2.414 being only a very little bit ahead of 2.413 in my prioritization.  I'd be happy with either of them.

 

Mark Waite

[Alexander Brandes]

No objections against 2.414 from my side either.

[Basil Crow]

2.414 does have some benefits, like a Prototype fix and an
optimization to the End of Life subsystem. The only risky portion is
the Safe Restart change, but as Mark mentioned, this could always be
patched or backed out if necessary. I audited the Safe Restart change,
and it seems safe (no pun intended) to me. So my vote is for 2.414.
But I don't feel strongly either way.

[Basil Crow]

There do not appear to be any known user-visible regressions in production code in 2.414, but jenkinsci/jenkins#7872 implies a change in Login Theme (3,757 installations) which has not yet been completed (tracked in JENKINS-71238) and jenkinsci/jenkins#7942 implies a change in Calendar View (1,866 installations) which has not yet been completed (tracked in JENKINS-71663). If these tasks are not completed by the time 2.414 LTS ships, they should be documented as known defects in the release notes.

[Mark Waite]

On Monday, July 17, 2023 at 9:41:51 AM UTC-6 Basil Crow wrote:

There do not appear to be any known user-visible regressions in production code in 2.414, but jenkinsci/jenkins#7872 implies a change in Login Theme (3,757 installations) which has not yet been completed (tracked in JENKINS-71238) and jenkinsci/jenkins#7942 implies a change in Calendar View (1,866 installations) which has not yet been completed (tracked in JENKINS-71663). If these tasks are not completed by the time 2.414 LTS ships, they should be documented as known defects in the release notes.

Thanks for noting those needed additions to the upgrade guide.  Since those changes are in Jenkins 2.413 and in Jenkins 2.414, it makes sense to me that we choose Jenkins 2.414.

Tim Jacomb, do you agree that Jenkins 2.414 should be the baseline for the LTS release scheduled for August 23, 2023?

Mark Waite

[Tim Jacomb]

Yes

--
You received this message because you are subscribed to the Google Groups "Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-de...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-dev/2fad8aa3-2774-413d-b41b-1f2e1ee1a94dn%40googlegroups.com.

[jn...@cloudbees.com]

2.414 has a regression in the plugin manager.
I would be -1 on this unless that is fixed (currently missing a Jenkins bug but has been the cause of a lot of ATH failures)

https://github.com/jenkinsci/acceptance-test-harness/issues/1284 for the current status/investigation, I will file a Jenkins bug on Monday to track.

/James

[Basil Crow]

On Fri, Jul 21, 2023 at 10:42 AM 'jn...@cloudbees.com' via Jenkins
Developers <jenkin...@googlegroups.com> wrote:
> 2.414 has a regression in the plugin manager.
> I would be -1 on this unless that is fixed (currently missing a Jenkins bug but has been the cause of a lot of ATH failures)
> https://github.com/jenkinsci/acceptance-test-harness/issues/1284 for the current status/investigation, I will file a Jenkins bug on Monday to track.

Thanks for starting to investigate this. As I wrote in the discussion
at jenkinsci/acceptance-test-harness#1284, I was able to definitively
implicate jenkinsci/jenkins#8025 as the cause of the ATH test failure
by widening the race window with some sleeps, and I may have been able
to chase away the ATH test failure and/or fix the bug by removing the
asynchrony from the original change.

Since jenkinsci/jenkins#8025 was the last commit in 2.414 and since
2.414 contains some other changes that we want, I would like to
propose that we keep 2.414 as the next LTS baseline but simply revert
jenkinsci/jenkins#8025 from the stable branch before shipping. (An
equally plausible alternative would be to choose 2.413 and backport
the other changes that we want.)

As far as the main branch is concerned, now that I have come up with a
way to reproduce the original problem, I would like to propose that we
give folks a limited amount of time (say, a week) to come up with a
solution and otherwise revert jenkinsci/jenkins#8025 on the main
branch as well to restore stability to the ATH test suite.

[Tim Jacomb]

Shall we go with 2.415 instead as it will have security fixes (given they would be backported anyway, it's less confusing if we start with the version they were introduced in I think)?

The plugin manager bug in 2.414 has been fixed.

--
You received this message because you are subscribed to the Google Groups "Jenkins Developers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to jenkinsci-de...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-dev/CAFwNDjpCWAkp6nfB07z3wc9bMqaFs091tbGWyW3CVMQZm8dFmw%40mail.gmail.com.

[jn...@cloudbees.com]

Thanks Tim & Basil for continuing this.

I filed https://issues.jenkins.io/browse/JENKINS-71698 for LTS tracking and tagged the jenkins PR 

>  Shall we go with 2.415 instead as it will have security fixes 

not sure I follow 2.415 is not a security release?  are you meaning 2.416 which will be the security release which will naturally contain the 2.415 bits?

2.415 appears to have a regression in the dialog handling that I just found with a minimal amount of testing, so possibly too new to have had exposure  :-(

[jn...@cloudbees.com]

>  2.415 appears to have a regression in the dialog handling that I just found with a minimal amount of testing, so possibly too new to have had exposure  :-(

https://issues.jenkins.io/browse/JENKINS-71699

[Tim Jacomb]

Ah I see I missed that 2.415 was out.

Let’s continue with 2.414 and backport security fixes

To view this discussion on the web visit https://groups.google.com/d/msgid/jenkinsci-dev/7d39f095-36c4-4164-a4b9-34f920b5d3dcn%40googlegroups.com.

[Alexander Brandes]

Alternatively, 2.417 could be a viable candidate, as it includes the security fix, and could include both regression fixes for JENKINS-71698 and JENKINS-71699, if reviewed within the next week.

Thanks to Tim and Markus for filing the couple of PRs so quick!

[Basil Crow]

On Mon, Jul 17, 2023 at 8:41 AM Basil Crow <m...@basilcrow.com> wrote:

There do not appear to be any known user-visible regressions in production code in 2.414, but jenkinsci/jenkins#7872 implies a change in Login Theme (3,757 installations) which has not yet been completed (tracked in JENKINS-71238) and jenkinsci/jenkins#7942 implies a change in Calendar View (1,866 installations) which has not yet been completed (tracked in JENKINS-71663). If these tasks are not completed by the time 2.414 LTS ships, they should be documented as known defects in the release notes.

How close are we to completing these tasks? It is not desirable to ship an LTS release with known defects. If our users upgrade and encounter regressions, they will lose trust in our software.

In addition to the above, jenkinsci/jenkins#7951 implies a change in Global Build Stats (6,641 installations) which has not yet been completed (tracked in JENKINS-71741). If this task is not completed by the time 2.414.1 LTS ships, this should be documented as a known defect in the release notes.