feat: EnvoyPatchPolicy support targetRefs#8713
feat: EnvoyPatchPolicy support targetRefs#8713zirain wants to merge 11 commits intoenvoyproxy:mainfrom
Conversation
✅ Deploy Preview for cerulean-figolla-1f9435 canceled.
|
Codecov Report❌ Patch coverage is Additional details and impacted files@@ Coverage Diff @@
## main #8713 +/- ##
==========================================
- Coverage 74.35% 74.32% -0.03%
==========================================
Files 245 245
Lines 38847 38987 +140
==========================================
+ Hits 28883 28979 +96
- Misses 7963 7995 +32
- Partials 2001 2013 +12 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
@codex review |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 75527614e2
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: f26683595c
ℹ️ About Codex in GitHub
Codex has been enabled to automatically review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
When you sign up for Codex through ChatGPT, Codex can also answer questions or update the PR, like "@codex address that feedback".
Signed-off-by: zirain <zirain2009@gmail.com>
Signed-off-by: zirain <zirain2009@gmail.com>
Signed-off-by: zirain <zirain2009@gmail.com>
Signed-off-by: zirain <zirain2009@gmail.com>
Signed-off-by: zirain <zirain2009@gmail.com>
zirain
force-pushed
the
epp-multi-targetrefs
branch
from
f266835 to
f7983f5
Compare
Signed-off-by: zirain <zirain2009@gmail.com>
|
@codex review |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 6a5bcf3f2c
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| } else { | ||
| // Keep existing ancestor that wasn't updated | ||
| mergedAncestors = append(mergedAncestors, existingStatus.Ancestors[i]) |
There was a problem hiding this comment.
The merge branch preserves any existingStatus.Ancestors entry that is missing from the newly translated status (else path here), which makes ancestor status sticky across updates. If a policy previously targeted two Gateways and later removes one targetRef, the removed target is no longer emitted in the new translation, but this code keeps it forever, so users continue to see outdated attachment/condition state. The merged result should prune ancestors absent from the latest status (or gate retention by generation) instead of always carrying them forward.
Useful? React with 👍 / 👎.
| Name: string(refName), | ||
| } | ||
| irKey = t.IRKey(gatewayNN) | ||
| ancestorRef = getAncestorRefForPolicy(gatewayNN, nil) |
There was a problem hiding this comment.
For non-merged mode, ancestorRef is derived with getAncestorRefForPolicy before validating targetRef.Group/Kind, so every entry is normalized to a Gateway ancestor even when the input ref is invalid. With targetRefs, an invalid and a valid ref that share the same name can collide on the same ancestor: the invalid one sets Accepted=False, and the valid one then cannot flip it to accepted because the accepted condition already exists. This produces incorrect status (rejected) even when patches were successfully applied to the valid target.
Useful? React with 👍 / 👎.
Signed-off-by: zirain <zirain2009@gmail.com>
2 participants
xref: #8623
targetRefsto EnvoyPatchPolicyspec.targetRef