1996fanrui opened a new pull request, #28661: URL: https://github.com/apache/flink/pull/28661
This PR depends on https://github.com/apache/flink/pull/28651, https://github.com/apache/flink/pull/28652, https://github.com/apache/flink/pull/28659 and https://github.com/apache/flink/pull/28660 Please only review the last 5 commits (prefixed [FLINK-39524]). ## What is the purpose of the change [FLINK-39524] Fetched channel state: spill files, filtered write, reader. It's part 5 of FLINK-38544 (checkpointing during recovery of unaligned checkpoint state). This PR introduces the on-disk subsystem for recovered channel state: an append-only segmented spill-file format written by the recovery read path (optionally through the rescale record filter), a ref-counted file container with snapshot/resume semantics, and a forward-only reader. Runtime behavior is unchanged: the handler factory does not select the new handlers, so no production path produces or consumes spill files. ## Brief change log - [FLINK-39524][checkpoint] Add FetchedChannelState spill-file container - [FLINK-39524][checkpoint] Spill-writing handlers: segmented on-disk format - [FLINK-39524][checkpoint] Rewrite ChannelStateFilteringHandler to emit into a spill segment - [FLINK-39524][checkpoint] Forward-only spill reader with snapshot/resume - [FLINK-39524][checkpoint] SequentialChannelStateReader#readInputData returns Optional<FetchedChannelState> ## Verifying this change - New unit tests: FetchedChannelStateTest, FetchedChannelStateRefCountTest, FetchedChannelStateReaderTest, AbstractSpillingHandlerTest, ChannelStateFilteringHandlerTest, GateFilterHandlerTest, GateFilterHandlerBufferOwnershipTest, RecoveredChannelStateHandlerFilterRoutingTest, SequentialChannelStateReaderImplTest - RecoveredStateFilteringLargeRecordITCase stays green (the in-memory filtering path is untouched) ## 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 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
