Dabz commented on a change in pull request #10894:
URL: https://github.com/apache/kafka/pull/10894#discussion_r654494364
##########
File path:
streams/src/main/java/org/apache/kafka/streams/processor/internals/GlobalStateManagerImpl.java
##########
@@ -288,6 +288,22 @@ private void restoreState(final StateRestoreCallback
stateRestoreCallback,
// -> don't pass in `task.timeout.ms` to stay responsive if
`KafkaStreams#close` gets called
final ConsumerRecords<byte[], byte[]> records =
globalConsumer.poll(requestTimeoutPlusTaskTimeout);
Review comment:
If we do not change this line, the restoring phase would be stuck for
`requestTimeoutPlusTaskTimeout` (5m30s by default) for every partition not
having any data to returned. Thus, potentially drastically increasing the
required restore phase. Maybe we should invoke `poll()` more aggressively?
--
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
