Github user tzulitai commented on a diff in the pull request:
https://github.com/apache/flink/pull/3314#discussion_r104311996
--- Diff:
flink-connectors/flink-connector-kafka-0.8/src/main/java/org/apache/flink/streaming/connectors/kafka/internals/SimpleConsumerThread.java
---
@@ -381,6 +381,10 @@ else if (partitionsRemoved) {
partitionsIterator.remove();
continue
partitionsLoop;
}
+
+ if (value == null) {
+ continue;
+ }
--- End diff --
Would it make sense to do the `null` checking inside `emitRecord(...)`?
Otherwise, we wouldn't be updating the state for skipped records, and
therefore not accounting it as "already processed".
I don't think it really matters, since we aren't outputting anything
anyway, but I see at least one minor advantage that might deserve changing it:
If we fail during a series of continuous skipped records, we won't be wasting
any overhead re-processing them on restore.
---
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.
---
