v1.4.0 Release-Cycle: GEP Iteration and Review Phase

[shaneutt]

IMPORTANT: The GEP Iteration and Review cycle for v1.4.0 has ended, and this thread is closed

The scoping phase for standard and experimental channels for v1.4.0 has concluded.

The following standard features are accepted for the release:

• GEP: TLS from Gateway to Backend for ingress (backend TLS termination) #1897
• GEP: SupportedFeatures #3164
• GEP: Configurable Retries in HTTPRoute #1731
• GEP: Add Name to HTTPRouteRule and HTTPRouteMatch #995

As for experimental, we currently have capacity for 4 regular issues, and 2 issues that support the transition from ingress to Gateway API based on the current experimental pipeline. These experimental features are accepted:

Normal Pipeline (max 4):

• GEP: Support E/W authorization #3779 - Missed Implementable deadline and extension. Out of scope for this release.
• GEP: Support ClientIp based session persistence #3798 - Missed Implementable deadline. Out of scope for this release.
• GEP: Off-Cluster Gateways #3792 - Extension to pull the Mesh resource out of this GEP for inclusion in v1.4.
• Mesh Resource Definition #3949 - ✔️
• GEP: Default Gateways #3793 - ✔️

Ingress Pipeline (max 2):

• GEP: Gateway/HTTPRoute level authentication #1494 (special iteration, already provisional) - Merged. ✔️
• GEP: Client Certificate Verification for Gateway Listeners #91 (special iteration, not graduating), Extension until 25th July. ✔️

The following timeline shows the expectations for the experimental feature transitions:

Date	Event Description
June 10	GEP PR with "What?"/"Why?"/"Who?" is ready for review ✔️
June 17	GEP PR with "What?"/"Why?"/"Who?" is merged ✔️
July 1	GEP PR with proposed API details (the "How?") is ready for review ✔️
July 22	GEP PR with final API details is merged in implementable state ✔️

After this, then final implementations will be be due in time for code freeze on August 26th. See the milestone for more details after the GEP iterations and review cycle.

Note: The strong emphasis on getting the "What?", "Why?" and "Who?" submitted first, to build consensus before worrying about the implementation details. This will be key to smooth iterations throughout this phase.

Warning: Features that don't meet these timelines will be assessed, and may unfortunately have to be dropped from the release for timing and capacity reasons. When in doubt, please communicate any expected issues with the maintainers so we can coordinate with you!

The timeline for standard features will be done a little different and communicated uniquely for each issue. Expect updates and check-ins on each individual feature during this phase, and if you have any questions, concerns or need support please reach out!

[mikemorris]

Could you provide more context on this note for #91?

(special iteration, not graduating)

[robscott]

This is an existing experimental GEP that is not attempting to graduate to another level in this release, just make some meaningful adjustments to the existing GEP. It's rather ambiguous how we handle this in terms of release scoping and if it should fill up a slot in the release or if these kinds of changes should always be considered in scope as they're intended to progress an existing feature towards a GA state.

cc @arkodg and @kl52752 to make sure I got the details right here.

[candita]

Should it be assigned @robscott ?

[shaneutt]

Note: we have moved forward the timeline to have Experimental GEPs ready for review with the "How?" (the implementation details) by one week, putting that at July 1st. Future timelines remain the same, but we made this extension in response to several contributors feedback that the previous transition was not feasible.

[shaneutt]

📣 An update for those who have GEPs moving forward in this release: We haven't had a super-official way to request extensions for timelines as the release process is still shaping up. However, we recognize that we need something.

If you need an extension, these are our guidelines (so far) based on the Kubernetes process:

• Extensions can be granted on a per-GEP basis. The owners of the GEP have to ask and provide a timeline (measured in days) as to when they believe the GEP will be ready for merge.
• The request and approval for a GEP extension needs to be in public.
• Extensions can only be granted with a majority agreement by maintainers / release-manager

(note that in the future we're going to try and iterate and improve on this, as well as the whole release process based on what we learn during each one).

So if you need an extension, please create a comment here on this discussion with the request and let us know.

[youngnick]

I think it's important to mention that only the GEP document update needs to be merged for the GEP to be Implementable. The GEP moves to Experimental once the API changes are made, which has up until August 26th. I think we probably need some further clarifications to the GEP process page on the exact steps in our round of changes after this release.

[candita]

What GEPs remain that aren't listed below?

[kl52752]

Request for timeline extension for: #3760 (comment)
Reason: We reach out the agreement about proposal. Update for #3942 is in review. After that PR with API changes needs to be merged.
ETA: EOW

[shaneutt]

maintainer +1

[robscott]

maintainer +1

[youngnick]

+1 from me, although I think the GEP changes (which are the ones under deadline) will make it in today, most likely. Thanks for asking for the extension @kl52752!

[youngnick]

With that, the extension is approved (majority of maintainers).

[kflynn]

I'd like an extension for Default Gateways until the end of the week. I expect to have resolved all the comments today, but I want to give folks adequate time to review before it merges.

[shaneutt]

maintainer +1

[robscott]

maintainer +1

[youngnick]

maintainer +1.

With that, the extension is approved (majority of maintainers).

[kflynn]

I'm going to preemptively request a special dispensation here too: although I don't think it's realistic to really get Off-Cluster Gateways into 1.4, I would really like to get the Mesh resource in. I'll work with @karthikbox here to get a minimal definiton GEP for Mesh in tomorrow.

[youngnick]

I am +1 to an extension for a cut-down Mesh resource GEP. No longer than a week though!

This is because it will be useful to be able to start prototyping SupportedFeatures on the Mesh resource also.

[robscott]

maintainer +1 for up to one week

[shaneutt]

+1

[youngnick]

Extension was approved in time.

[LiorLieberman]

I'd like a week-long extension for E/W authz GEP until 29th July. Based on recent useful feedback and comments (some of which were not on GH), I expect to work with other folks this week to figure out if and how we want to change the API details to be able to merge a good-enough experimental API.

[youngnick]

maintainer +1 for one week.

[robscott]

maintainer +1 for up to one week

[shaneutt]

+1

[youngnick]

Extension was approved.

[shaneutt]

Thank you everyone for your time and efforts on this part of the release cycle.

The GEP Iteration and Review part of the cycle has completed, and we're now on to implementations on our way to code freeze on August 22nd.