Cannot record events outside reconciliation loop or for arbitrary resources (`RegisteredController#eventRecorder` missing)
I maintainer di solito rispondono entro 1 giorno
Valutazione
- Difficoltà
- 4/5
- Tempo stimato
- 3-5 giorni
- Idoneità per principianti
- 54/100
- Tipo di issue
- Bug
- Chiarezza
- Specificata chiaramente
- Stato di attività
- Attiva
- Stack tecnologico
- java
- Ambito
- backend-api-design
Direzione di ricerca
Read the v5.6 event recorder release notes and trace the API changes described for PRs #3570 and #3581, then inspect RegisteredController, Context, and DefaultEventRecorder. Confirm how recording is currently scoped to the primary resource and where an unbound recorder or explicit target resource could fit; done means events can be recorded outside reconciliation and targeted at a parent resource from a sub-resource controller.
Scritto dal modello di indicizzazione a partire dal testo della issue.
Descrizione
Bug Report
What did you do?
We attempted to record a Kubernetes event for our primary/main resource from within the reconciliation loop of a sub-resource / secondary controller, following the v5.6 release notes:
https://javaoperatorsdk.io/blog/2026/09/02/version-5.6-released/#kubernetes-event-recorder
According to the release documentation:
"Outside of the reconciliation loop — from a status listener or a background task, or about an object other than the primary — use the unbound form from
RegisteredController#eventRecorder(), which is scoped to the controller rather than to a reconciliation:EventRecorder recorder = registeredController.eventRecorder(); recorder.record(someOtherResource, EventRecord.normal("Noticed", "something happened"));
What did you expect to see?
RegisteredControllerexposes aneventRecorder()method (or equivalent API) to obtain an unboundEventRecorder.EventRecorderprovides a way to record events for an arbitrary target resource (e.g.recorder.record(someOtherResource, EventRecord.normal(...))), allowing a controller reconciling a sub-resource to emit an event on its main/parent resource.
What did you see instead? Under which circumstances?
RegisteredControllerdoes not containeventRecorder(). CallingregisteredController.eventRecorder()fails to compile.- In
Context<P>,eventRecorder()returns aResourceEventRecorderstrictly bound tocontext.getPrimaryResource(). - In
DefaultEventRecorder,toEvent(...)hardcodes the target resource tocontext.getPrimaryResource():var regarding = context.getPrimaryResource(); and EventRecorder#record only accepts (EventRecord event, Context<?> context).
As a result, when reconciling a sub-resource (where context.getPrimaryResource() is the sub-resource), there is no supported way to publish an event targeting the main/parent resource.
Environment
Kubernetes cluster type: any
java-operator-sdk version: 5.6.3
java -version: 17 / 21
kubectl version: 1.30+
Possible Solution
Tracing through the PR history:
- PR #3570 (feat: Add Event Recorder) originally introduced RegisteredController#eventRecorder() and supported recording events for arbitrary objects.
- PR #3581 (refactor: configurable event recorder) refactored the API to be stateless and removed RegisteredController#eventRecorder(), inadvertently dropping the ability to record events outside a reconciliation context or targeting a resource other than context.getPrimaryResource().
Suggested fixes:
- Re-introduce an unbound EventRecorder accessible via RegisteredController, ConfigurationService, or Operator.
- Allow passing an explicit target HasMetadata targetResource (or via EventRecord / EventRecorder#record(targetResource, ...)), rather than exclusively binding to context.getPrimaryResource().
Additional context
Our use case: We have separate controllers reconciling sub-resources. When state changes or failures occur during the reconciliation of a sub-resource, we need to publish a Kubernetes event targeting the main/parent resource so users inspecting the parent resource (kubectl describe ) have visibility into events triggered by its sub-resources.
- Lingua principale
- Java
- Stelle
- 945
- Fork
- 247
- Merge medio
- 2g 10h
- PR unite (30g)
- 27
Preparare l'ambiente
- Nessun Dockerfile né file Docker Compose
- Nessun modello di pull request
- Leggi la guida per i contributori
Come iniziare
- Leggi tutta la issue e poi la guida ai contributi del progetto.
- Commenta sulla issue per dire che te ne occupi tu — evita che due persone facciano lo stesso lavoro.
- Fai un fork del repository e lavora su un branch.
- Apri una pull request che faccia riferimento al numero della issue.
Altre issue di operator-framework/java-operator-sdk
-
Make failure handling in EventRecorder configurable (fail-on-error option)Forse già presa Una pull request collegata a questa issue è aperta o già unita. Aperta
Difficoltà 3/5 1-2 giorni Idoneità per principianti 68/100
operator-framework/java-operator-sdk#3687 · 3 commenti ·
I maintainer di solito rispondono entro 1 giorno
-
on-demand
Difficoltà 4/5 3-5 giorni Idoneità per principianti 42/100
operator-framework/java-operator-sdk#3671 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 4/5 3-5 giorni Idoneità per principianti 62/100
operator-framework/java-operator-sdk#3643 · 7 commenti ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 4/5 3-5 giorni Idoneità per principianti 45/100
operator-framework/java-operator-sdk#3568 · 1 commento · 1 reazione ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 5/5 Più di una settimana Idoneità per principianti 25/100
operator-framework/java-operator-sdk#3563 ·
I maintainer di solito rispondono entro 1 giorno
Tutte le issue di operator-framework/java-operator-sdk
Issue simili
-
ImporterNoConsensus and ImporterFileVerificationErrors alerts can't fire on consensus failureApertabug
Difficoltà 2/5 1-3 ore Idoneità per principianti 78/100
hiero-ledger/hiero-mirror-node#14665 ·
I maintainer di solito rispondono entro 1 giorno
-
kind/bug
Difficoltà 2/5 1-3 ore Idoneità per principianti 65/100
cadence-workflow/cadence#8745 ·
I maintainer di solito rispondono entro 1 giorno
-
Difficoltà 2/5 1-3 ore Idoneità per principianti 68/100
oracle/graalvm-reachability-metadata#12168 ·
I maintainer di solito rispondono entro 1 giorno
-
area/frontend area/test kind/tech-debt
Difficoltà 2/5 1-3 ore Idoneità per principianti 82/100
camunda/camunda#69073 · 1 commento ·
I maintainer di solito rispondono entro 1 giorno
-
Patient profile returns a server error for a person with observation periods but no clinical recordsForse già presa @DG47 l’ha presa oggi. Aperta
Difficoltà 2/5 1-3 ore Idoneità per principianti 75/100
I maintainer di solito rispondono entro 1 giorno