Repository navigation
Description
Delivery bundle: B: Lifecycle correctness and status. See #397 and #368 for the four-PR stack.
Parent: #393
Roadmap: #368
Plan item: PR-08
Priority: P2
Dependencies: None
Kubernetes support: Rolling upstream-maintained minors; use the oldest/newest boundaries pinned in config/kubernetes-versions.env. See #368 for the current policy. Unrun jobs are not passed validation.
Problem Statement
Evidence: Ordinary cluster reconciliation writes Reconciling then Running; ensureFinalizers always updates; KafkaUser status writes are unconditional. Multiple controllers share KafkaCluster status.
Proposed Solution
Files: pkg/k8sutil/status.go, its tests, controllers/kafkacluster_controller.go, controllers/kafkauser_controller.go, Cruise Control status writers, and existing controller tests.
Interface contract: Each writer changes only its owned fields and avoids requests when desired state is unchanged. For contested fields, use a fresh-read retry with optimistic locking; never replay a stale whole status. Lists/finalizers must preserve other writers' entries.
- Add request-count tests showing settled resources perform no unnecessary status/finalizer writes.
- Add two-writer tests updating unrelated status fields and finalizers; assert neither loses the other's changes and real conflicts remain retryable.
- Replace broad writes selectively with no-op-aware updates/patches. Preserve rolling-upgrade and broker-state signals used by other controllers.
- Retain the PR-01 execution-persistence invariant if modifying that path; JSON merge patch is not a blanket substitute for field ownership.
- Validate with
go test ./controllers ./controllers/tests ./pkg/k8sutil; compare write/conflict metrics on the PR-07 workload.
Done when: Settled reconciles have a documented lower write count and concurrent status writers preserve each other's data.
Delivery: Four consolidated PRs, approved November 16, 2026: A -> B -> C; D branches from A. Foundation is #395. The 22 PR-XX identifiers are work-item IDs, not separate PRs. Preserve every work-item acceptance criterion and native dependency. No gated experiment is automatically enabled, and no issue is complete merely because one PR references it. D must satisfy its native #375 gate, including lifecycle-domain signals assigned to B; branching from A does not waive that merge prerequisite. Write focused regression tests first; record exact commands/results and run make check plus this issue's real-cluster/measurement gates.
Alternatives Considered
Twenty-two individual PRs add unnecessary integration overhead, while one omnibus PR obscures correctness and rollout boundaries. Deliver this scope in its approved dependency-aligned bundle. Do not expand into deferred consistency, storage, or concurrency changes without their own decision gates.
Additional Context
Baseline: 95138e0d; detailed roadmap scope: #368, section PR-08. The roadmap #368 and its native sub-issues are the planning source of truth. No planning document is committed in the final PR diffs. Re-read affected source and overlapping work before implementation. Historical baseline test passes are not validation of this change.
Activity
amuraru commentedon Nov 17, 2026
Concrete status-conflict fixes move to #391; measured redundant-write reductions move to #393 under #368. No blanket status/patch migration. Closing this split tracking scope does not claim all original work implemented.