[FLINK-25433] Integrate retry strategy for cleanup stage - ASF JIRA

XML

Word

Printable

JSON

Release Note:

Hide
Adds retry logic to the cleanup steps of a finished job. This feature changes the way how Flink jobs are cleaned up. Instead of trying once to clean up the job, this step will be repeated until it succeeds. Users are meant to fix the issue that prevents Flink from finalizing the job cleanup. The retry functionality can be configured and disabled. More details can be found in the documentation under Deployment / Configuration / Retryable Cleanup.

Show
Adds retry logic to the cleanup steps of a finished job. This feature changes the way how Flink jobs are cleaned up. Instead of trying once to clean up the job, this step will be repeated until it succeeds. Users are meant to fix the issue that prevents Flink from finalizing the job cleanup. The retry functionality can be configured and disabled. More details can be found in the documentation under Deployment / Configuration / Retryable Cleanup.

The ResourceCleaner should be able to cleanup not only once but retry infinitely.

causes

FLINK-26331 Make max retries configurable

FLINK-26388 Release Testing: Repeatable Cleanup (FLINK-25433)

relates to

FLINK-26114 DefaultScheduler fails fatally in case of an error when shutting down the checkpoint-related resources

links to

GitHub Pull Request #18637