flink
flink copied to clipboard
[FLINK-35030][runtime] Introduce Epoch Manager under async execution
What is the purpose of the change
This PR introduces Epoch Manager to handle watermark and watermark status processing under async execution. Epoch manager segments inputs into distinct epochs, marked by the arrival of non-records(e.g.watermark, record attributes). Records are assigned to a unique epoch based on their arrival, records within an epoch are allowed to be parallelized, while the non-record of an epoch can only be executed when all records in this epoch have finished.
Brief change log
- Add
SerialEpochManagerandParallelEpochManager - Wire
SerialEpochManagerto AEC - Leverage EpochManager to process watermark/watermark status
Verifying this change
This change added tests and can be verified as follows:
EpochManagerTestAsyncExecutionControllerTest#testSerialEpochManager
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): (yes)
- 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? (yes)
- If yes, how is the feature documented? (JavaDocs)
CI report:
- 41cd07d407c8b7a24107af5b80883d358e9a7dc8 Azure: SUCCESS
Bot commands
The @flinkbot bot supports the following commands:@flinkbot run azurere-run the last Azure build
Thanks all for the detailed review! 🚀🚀