chore(main): release 0.41.0 - #285
Merged
sunib merged 1 commit intoJul 30, 2026
Merged
Conversation
github-actions
Bot
force-pushed
the
release-please--branches--main--components--gitops-reverser
branch
2 times, most recently
from
July 29, 2026 21:42
3976ed9 to
3d57bff
Compare
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
github-actions
Bot
force-pushed
the
release-please--branches--main--components--gitops-reverser
branch
2 times, most recently
from
July 30, 2026 15:34
6feb4dd to
a2238f1
Compare
github-actions
Bot
force-pushed
the
release-please--branches--main--components--gitops-reverser
branch
from
July 30, 2026 17:14
a2238f1 to
407e67e
Compare
sunib
approved these changes
Jul 30, 2026
sunib
deleted the
release-please--branches--main--components--gitops-reverser
branch
July 30, 2026 18:42
Contributor
Author
|
🤖 Created releases: 🌻 |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Release 0.41.0
Nine commits since v0.40.1. Two of them are worth the release on their own: attribution no longer needs Redis, and the machinery underneath it is now the shape a second replica needs.
Attribution without Redis
Naming the human behind a commit used to mean running Valkey. It no longer does.
Attribution facts moved from a Redis keyspace the watch side polled to a log it follows, with an in-process index in front of it. Two consequences:
--author-attribution-transport=memorycarries the facts in an in-process ring buffer. A single-replica install now gets named commit authors with no Valkey/Redis at all — running a StatefulSet just to writealiceinstead of the bot intogit logwas out of proportion, and now you don't have to.Redis stays the default and is still the right choice for anything you'd call production — it is the only transport whose facts survive a restart, and the only one that can reach a second process.
One step toward HA
The Redis transport is now one stream per (audit route, type), fan-out, with every follower reading every entry from its own cursor. That is deliberately the shape multiple processes need: no consumer groups, because each replica needs the whole stream, not a share of it.
This is a step, not the destination — HA is not shipped, and
memoryis refused at startup with more than one replica precisely because it cannot cross a process boundary. But the attribution side is no longer the thing standing in the way.The index is keyed by route first, which matters more than it looks: one index serves the whole process while streams are per (route, type), so a type-keyed index would pool two clusters' facts and could hand a watch event on cluster B an author from cluster A.
⚠ Breaking: placement no longer infers from your repository
A new document's path now comes only from what you declared, the folder's single kustomization root, or the built-in canonical path. It is never inferred from where the repository happens to keep the other documents of that type.
Sibling-cohort inference is gone. The reason is not primarily the bug it produced: inference let a human's edit to the repository change where the operator writes next, with no Kubernetes object changing and nothing in status recording the move. The bug is the evidence — a vacuous namespace guard let one namespace's object append into another's file, which legitimized the bundle for every later object, and collapsed a whole type into a single file. A rule inferred from mutable state has failure modes that feed themselves.
The kustomize-root fallback stays, because it is not inference: a file no kustomization can reach is never rendered, so placing a new document beside the folder's one root follows from there being one root. Two roots is still ambiguous and still declines.
If you relied on inference, declare the layout with
spec.placement.byType/default. SeeUPGRADING.md.⚠ Breaking: two smaller ones
exact_deletecollection_itemis replaced bycollection_uidandcollection_scope, which carry different confidence: a UID the API server said it deleted, versus a match on namespace, selector and window alone. Dashboards selectingresult=~"exact_.*|weak"should selectresult!="absent"instead, or they will read the collection tiers as misses.Also in this release
multi_ack, which go-git only implements from v6 — so ADO could not be fetched at all before. This moves to go-git v6 rather than bundling a systemgitbinary, which measured at +723 MB of image and left the CRITICAL scan gate blind to git, OpenSSH and OpenSSL. Covered by a credential-free simulator in CI plus opt-in tests against a real tenant. (#297, guide)Upgrading
Nothing to do unless you relied on placement inference, or your dashboards select attribution
resultlabels by prefix. Both are covered above.Full changelog
See
CHANGELOG.mdin this PR's diff for the generated entry.