Reconcile stops for FAILING SparkApplication

[kapilgarg1996]

What happened?

• ✋ I have searched the open/closed issues and my issue is not listed.

https://github.com/kubeflow/spark-operator/blob/master/internal/controller/sparkapplication/controller.go#L627-L649

2026-03-10T05:15:34.811Z INFO sparkapplication/event_handler.go:84 Spark pod updated {"name": "app-1773119700000918904-driver", "namespace": "helm", "oldPhase": "Running", "newPhase": "Failed"}
2026-03-10T05:15:34.817Z INFO sparkapplication/controller.go:171 Reconciling SparkApplication {"name": "app-1773119700000918904", "namespace": "helm", "state": "RUNNING"}
2026-03-10T05:15:34.861Z INFO sparkapplication/event_handler.go:188 SparkApplication updated {"name": "app-1773119700000918904", "namespace": "helm", "oldState": "RUNNING", "newState": "FAILING"}
2026-03-10T05:15:34.866Z INFO sparkapplication/controller.go:171 Reconciling SparkApplication {"name": "app-1773119700000918904", "namespace": "helm", "state": "FAILING"}

There are no reconcile logs for this application after this

Reproduction Code

No response

Expected behavior

No response

Actual behavior

No response

Environment & Versions

• Kubernetes Version:
• Spark Operator Version:
• Apache Spark Version:

Additional context

No response

Impacted by this bug?

Give it a 👍 We prioritize the issues with most 👍

[github-actions]

🎉 Welcome to the Kubeflow Spark Operator! 🎉

Thanks for opening your first issue! We're happy to have you as part of our community 🚀

Here's what happens next:

• Our team will review your issue soon!
• If you'd like to contribute to this issue, check out our Contributing Guide for repo-specific guidelines and the Kubeflow Contributor Guide for general community standards.

Join the community:

• Slack: Join our #kubeflow-spark-on-kubernetes Slack channel.

Feel free to ask questions in the comments if you need any help or clarification!
Thanks again for contributing to Kubeflow! 🙏

[tariq-hasan]

Hi @kapilgarg1996 thanks for raising the issue.

I looked at the code block you referenced. But I wasn't able to reproduce the issue locally - both the retry and no-retry cases seem to work if I use a non-existent JAR, for example. If you can share your SparkApplication spec and the full operator logs around that timestamp I can take another look at the issue.

[kapilgarg1996]

Hi @tariq-hasan , thanks for looking into it.
I actually found this issue in 2.0.2

Bug Details: The operator will not reconcile/retry spark application if it fails before the retryInterval duration. For eg. if the retryInterval is 30s and the driver fails within 20s, then the SparkApplication would always remain in FAILING state.

Code reference: https://github.com/kubeflow/spark-operator/blob/v2.0.2/internal/controller/sparkapplication/controller.go#L509-L520

Reason: When retry is not due, the operator sends an update with no change to k8s for which it doesn't receive a watch event and reconciliation stops.

I think this has been fixed in latest versions with requeueAfter, but we haven't tested it yet.

[tariq-hasan]

Understood. The #2241 PR might have addressed this issue I believe.