Repository navigation
Fix four bugs ahead of the next release - #1175
Merged
Merged
Conversation
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
![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: 629bea4447
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".
NishanthNalluri
force-pushed
the
fix/pre-release-bug-fixes
branch
3 times, most recently
from
October 29, 2026 13:56
a78aba4 to
92f0683
Compare
ShashwatDash23
approved these changes
Oct 30, 2026
SetupRBAC creates only resources the controller does not watch and records nothing in status, so the reconcile that ran it produced neither a requeue nor a watch event and the cluster stalled before the StatefulSet until the next cache resync. Fixes #1174.
The StatefulSet builder rebuilds the pod template from the spec, which dropped the marker the restart actor writes onto the live template. The next deploy then saw a template change and rolled every pod a second time, including for changes that should not touch pods. Fixes #1171.
Next returns false both when the rows are exhausted and when the stream fails, so a truncated read was returned as a complete one. Closes #1162.
The annotation selects the restart action, so a swallowed update error left it in place and the next reconcile rolled the cluster again. Retry on conflict before giving up, as the other CR annotation writes do. Closes #1164.
NishanthNalluri
force-pushed
the
fix/pre-release-bug-fixes
branch
from
October 30, 2026 08:51
92f0683 to
88d68c9
Compare
The version checker Job uses fixed resource requests, so the test's large CR resource requests no longer keep its pod pending. This commit uses required node affinity to make the pod unschedulable and checks that Kubernetes reports it that way.
This commit upgrades k3d from 5.2.2 to 5.8.3 to support newer Kubernetes versions in local development and CI.
This was referenced Oct 30, 2026
Closed
Merged
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.
Four independent fixes, one commit each.
controller: requeue after a successful action (#1174) — SetupRBAC creates only resources the controller does not watch and records nothing in status, so the reconcile that ran it produced neither a requeue nor a watch event. A cluster that reached SetupRBAC as the only pending queue entry stalled before the StatefulSet until the next cache resync, with no error and
clusterStatusstuck atStarting. Reconcile now requeues after an action succeeds; once no action is left the director returns nil and the loop stops as before.resource: keep the restart marker on the pod template (#1171) — the StatefulSet builder rebuilds the pod template from the spec, dropping the
crdb.io/restartmarker the restart actor writes onto the live template. The next deploy saw that as a template change and rolled every pod a second time, including for changes that should not touch pods such as a node count change.clustersql: check rows.Err() (#1162) —
Nextreturns false both when rows are exhausted and when the stream fails, so a truncated read ofcrdb_internal.zoneswas returned as a complete one. Same change as #1162.actor: return the restart annotation cleanup error (#1164) — the annotation is what selects the restart action, so swallowing the update error recorded the restart as complete while leaving the annotation in place, and the next reconcile restarted the cluster again. Same change as #1164, with the cleanup pulled into its own method so the failure path can be tested without a live health checker.