Skip to content

[FLINK-39073][runtime] Defer alignment check for idle splits#27638

Open
Efrat19 wants to merge 2 commits intoapache:masterfrom
Efrat19:FLINK-39073
Open

[FLINK-39073][runtime] Defer alignment check for idle splits#27638
Efrat19 wants to merge 2 commits intoapache:masterfrom
Efrat19:FLINK-39073

Conversation

@Efrat19
Copy link
Contributor

@Efrat19 Efrat19 commented Feb 19, 2026

What is the purpose of the change

If a split emits watermarks far into the future and then goes idle, alignment check will incorrectly mark it paused.
As max allowed watermark advances, Source operator will transition the split back to active state, (while its still idle)
This change aims to fix the issue by deferring the alignment check for idle splits until they break out of idleness.

Brief change log

  1. Tracking currentlyIdleSplits in sourceOperator
  2. Avoiding pause of idle splits
  3. Adding alignment check when a split switches to not idle to compensate on 2)
  4. Improved logging around invalid split transitions

Verifying this change

This change added tests and can be verified. (unitTest)

Does this pull request potentially affect one of the following parts:

  • Dependencies (does it add or upgrade a dependency): no
  • The public API, i.e., is any changed class annotated with @Public(Evolving): no
  • The serializers: no
  • The runtime per-record code paths (performance sensitive): no
  • Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Kubernetes/Yarn, ZooKeeper: no
  • The S3 file system connector: no

Documentation

  • Does this pull request introduce a new feature? no
  • If yes, how is the feature documented? not applicable

If a split emits watermarks far into the future and then goes idle, alignment check will incorrectly mark it paused.
As max allowed watermark advances, Source operator will transition the split back to active state, (while its still idle)
This change aims to fix the issue by deferring the alignment check for idle splits until they break out of idleness.
@flinkbot
Copy link
Collaborator

flinkbot commented Feb 19, 2026

CI report:

Bot commands The @flinkbot bot supports the following commands:
  • @flinkbot run azure re-run the last Azure build

Split transitions directly from paused to idle and from idle to pause don't make any sense, improving the logs to allow further analysis. It is a convention for flink metrics system to not fail the job though.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants

Comments