curcur commented on a change in pull request #16531:
URL: https://github.com/apache/flink/pull/16531#discussion_r677941509
##########
File path:
flink-streaming-java/src/main/java/org/apache/flink/streaming/runtime/tasks/StreamTask.java
##########
@@ -338,6 +338,9 @@ protected StreamTask(
Executors.newCachedThreadPool(
new ExecutorThreadFactory("AsyncOperations",
uncaughtExceptionHandler));
+ environment.setMainMailboxExecutor(mainMailboxExecutor);
+ environment.setAsyncOperationsThreadPool(asyncOperationsThreadPool);
Review comment:
> I think it's not exactly true - asyncOperationsThreadPool only waits
for snapshot completion, but upload can happen in other threads (which is the
with RocksDB).
That's exactly happening here: `asyncOperationsThreadPool` is waiting for a
snapshot completion. The difference is that for the normal state backend, async
happens during checkpointing; while in materialization(changelog), it happens
during periodic materialization.
> Merging these pools is out of scope (if it happens), but should be taken
into account IMO.
That part I totally agree, that why `asyncOperationsThreadPool` is reused
here.
--
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]
