Plan to release etcd v3.5.17

[ivanvc]

What would you like to be added?

The etcd patch release criteria has been met for our release-3.5 stable release branch so we should release v3.5.17.

The list of commits included since the previous release is: v3.5.16...release-3.5:

• [3.5] docs: remove warning that etcd 3.5.0 was not yet released #18579
• [3.5] Ensure commits are signed in scripts/release.sh #18599
• [3.5] Bump go version to 1.22.8 #18669
• [3.5] github/workflows: use ECR mirror for Trivy's DB #18687
• [3.5] Remove defunct release-3.5 arm64 github actions workflows #18802
• [3.5] [serverWatchStream] terminate recvLoop on sws.close() #18784
• [3.5] manual: Fix risk of a partial write txn being applied #18799
• [3.5] chore(e2e): backport TestNoErrorLogsDuringNormalOperations test #18807
• [3.5] fix(defrag): handle errors during defrag #18842
• [3.5] update go to 1.22.9 #18849
• [3.5] fix(defrag): close temp file in case of error #18854

Work in progress CHANGELOG is: https://github.com/etcd-io/etcd/blob/main/CHANGELOG/CHANGELOG-3.5.md#v3516-tbc

List of pull requests we still need to backport from main to release-3.5 before the patch release is issued:

• Fix tx buffer inconsistency if there are unordered key writes in one tx. #17263
  • [3.5] Backport fix tx buffer inconsistency if there are unordered key writes in one tx #18861
• add tls min/max version to grpc proxy #18816
  • [3.5] add tls min/max version to grpc proxy #18829
• Use GitHub's CLI gh to publish releases #18604
  • [3.5] release: use GitHub's gh to create GitHub release #18876
• release: properly change working dir if tmp location already exists #18812
  • [3.5] release: properly change working dir if tmp location already exists #18863

Release team

GitHub handle	Role
@jmhbnz	Release Advisor
@ivanvc	Release Lead
@ghouscht	Release Shadow
@stackbaek	Release Shadow

Why is this needed?

Regular patch releases are vital to ensure our users have bug-free and secure software.

[ivanvc]

As discussed on Slack, it would be an excellent opportunity to do the release in person next week at KubeCon.

[ivanvc]

We should update to Go 1.22.9 (#18847)

[ivanvc]

I volunteer to be either the release lead or a shadow.

[ghouscht]

I volunteer to help as a shadow. However I'm not at KubeCon and the timezone difference to central europe might complicate things. If you find someone else to help, I'd be happy to renounce.

[jmhbnz]

Discussed during sig-etcd triage meeting. Thanks @ghouscht for volunteering as shadow, I will volunteer as release advisor and with @ivanvc as release lead we are all set to go. We are looking at Tuesday next week MST timezone.

We are thinking early morning on that Tuesday Utah time, how would that work for you @ghouscht?

[stackbaek]

@ivanvc - I'd like to be a release shadow for 3.5.

[ghouscht]

Discussed during sig-etcd triage meeting. Thanks @ghouscht for volunteering as shadow, I will volunteer as release advisor and with @ivanvc as release lead we are all set to go. We are looking at Tuesday next week MST timezone.

We are thinking early morning on that Tuesday Utah time, how would that work for you @ghouscht?

Yes, that does work for me 👍🏻

[ivanvc]

Tentatively, the release will be done at 9:00 am MST.

[ivanvc]

Can you help me do a pass at issues and PRs that may need backporting since 3.5.16 (September 10th)?

• https://github.com/etcd-io/etcd/issues?q=is%3Aissue+created%3A%3E2024-09-10
• https://github.com/etcd-io/etcd/pulls?q=is%3Apr+base%3Amain+created%3A%3E2024-09-10

I'll also do a pass later this weekend.

[chaochn47]

Hi @ivanvc, is it a good timing to catch the release train for #17263?

This bug fix seems like an important one we need to get into release branch.

[ivanvc]

Hi @chaochn47, it's still a good time if we can confirm that we want to backport that pull request. I may suggest opening the backport PR and getting feedback from @ahrtr / @serathius / @fuweid in the backport PR to expedite merging it before Tuesday :)

[ivanvc]

I swept through issues/PR and added the ones that caught my eye. Please let me know if you think there are more we should consider.

@ahrtr, should we consider #18829 for this release? Or do we want to hold it because of the tests?

[ahrtr]

@ahrtr, should we consider #18829 for this release? Or do we want to hold it because of the tests?

I already approved #18816, we need second approval to merge the PR. Once the PR for main is merged, then we can approve & merge PR for 3.5.

[ghouscht]

Can you help me do a pass at issues and PRs that may need backporting since 3.5.16 (September 10th)?

* https://github.com/etcd-io/etcd/issues?q=is%3Aissue+created%3A%3E2024-09-10

* https://github.com/etcd-io/etcd/pulls?q=is%3Apr+base%3Amain+created%3A%3E2024-09-10

I'll also do a pass later this weekend.

Had a look at the PRs/issues and I think you already caught whats needed. I also opened a PR to backport #18812 to 3.5 and assigned it to you: #18863

[jmhbnz]

Hey Team - Update, as discussed in person at KubeCon release 3.5.17 will proceed on schedule tomorrow morning MST and the following items will be deferred to 3.5.18 as they are not considered critical release blocking items:

• [3.5] Backport fix tx buffer inconsistency if there are unordered key writes in one tx #18861
• [3.5] add tls min/max version to grpc proxy #18829

[ghouscht]

Looks like we need to update the release steps/script. During the release we noticed that

etcd/scripts/release.sh

Line 133 in 7ab7612

if ! gh auth status &>/dev/null; then

does not exit with an error so the script doesn't work properly. Maybe also document that gh (or specifically the token gh uses) must have permission on the etcd-io org here:

etcd/Documentation/contributor-guide/release.md

Lines 65 to 66 in 7ab7612

	5. Install gh, refer to [GitHub's documentation](https://github.com/cli/cli#installation). Ensure that running
	`gh auth login` succeeds for the GitHub account you use to contribute to etcd.

[ivanvc]

Thanks, @ghouscht, for the note.

Just clarifying:

does not exit with an error so the script doesn't work properly.

The script does exit with an error (that's why it halted), but in the release instructions we should add to verify that gh auth status has a clean exit code.

We also need to update the documentation with detailed instructions on publishing the GitHub release.

[ivanvc]

Thanks, team (@ghouscht, @stackbaek, @jmhbnz), with all follow-up tasks closed and the release in place. I'll go ahead and close this issue 🎉