Github user revans2 commented on a diff in the pull request:
https://github.com/apache/storm/pull/1919#discussion_r99868716
--- Diff:
external/storm-kafka-client/src/main/java/org/apache/storm/kafka/spout/KafkaSpout.java
---
@@ -319,8 +320,10 @@ private boolean
emitTupleIfNotEmitted(ConsumerRecord<K, V> record) {
} else {
collector.emit(tuple, msgId);
}
- emitted.add(msgId);
- numUncommittedOffsets++;
+ if(!consumerAutoCommitMode){//only need to track in
none-AutoCommitMode
+ emitted.add(msgId);
--- End diff --
I think we still want to track emitted even in auto commit mode. Replaying
a failed tuple will not work unless it is added to emitted and even with auto
commit some people may want to replay tuples.
numUncommittedOffsets is only ever decremented in auto commit mode so that
part of the change is fine. Although, it would be good to add a javadoc to
numUncommittedOffsets indicating that it is not modified in auto commit mode.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
