1996fanrui commented on code in PR #28661:
URL: https://github.com/apache/flink/pull/28661#discussion_r3728240568
##########
flink-runtime/src/main/java/org/apache/flink/runtime/checkpoint/channel/SequentialChannelStateReaderImpl.java:
##########
@@ -110,7 +111,7 @@ public Optional<FetchedChannelState> readInputData(
// only signals "there is state to recover". The spilling backend
returns a real,
// file-backed container here.
Review Comment:
This value is only used as a presence flag today: if it's present,
StreamTask requests partitions with needsRecovery=true and builds a drainer; if
it's empty, it does neither. It never carries data — the in-memory handler
pushes the recovered buffers straight into the channels' queues.
The `Spilling*` handlers are introduced here but not selected by the factory
yet, so getProducedChannelState() is always null at this point; it starts
carrying real data in the next PR.
--
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]
