KEP-4443keps/sig-apps/4443-configurable-pod-failure-policy-reasons
More granular Job failure reasons for PodFailurePolicyRule
This KEP proposes adding an optional name field to PodFailurePolicyRule in the Job API. When a rule with action: FailJob matches a failed Pod, the Job controller will set the Job Failed condition reason to PodFailurePolicy_<ruleName>. If the field is unset, the controller derives the suffix from the rule index, preserving a deterministic machine-readable rea… Full summary
This KEP proposes adding an optional name field to PodFailurePolicyRule in the Job API. When a rule with action: FailJob matches a failed Pod, the Job controller will set the Job Failed condition reason to PodFailurePolicy_<ruleName>. If the field is unset, the controller derives the suffix from the rule index, preserving a deterministic machine-readable reason without requiring users to name every rule. The purpose of naming rules is to expose more detailed failure information in the Job status condition. This lets users configure multiple pod failure policy rules and lets higher-level controllers distinguish which rule, if any, caused a Job failure.
Facts
- Authors
- @danielvegamyhre
- Approvers
- @soltysh
- Reviewers
- @kannon92@alculquicondor@ahg-g
- PRR approvers
- @jpbetzalpha
- Tracking issue
- #4443(opens GitHub in a new tab) openmilestone v1.38sig/appsstage/alphatracked/yeswg/batchlead-opted-in
- Milestones
- alpha v1.38
- Participating SIGs
- none
- Feature gates
- PodFailurePolicyName
- Created / last updated
- 2024-02-05 / none