mjsax commented on code in PR #17686:
URL: https://github.com/apache/kafka/pull/17686#discussion_r1828589912
##########
streams/src/main/java/org/apache/kafka/streams/processor/internals/StreamTask.java:
##########
@@ -1073,10 +1073,13 @@ private void
initializeTaskTimeAndProcessorMetadata(final Map<TopicPartition, Of
@Override
public Map<TopicPartition, Long> purgeableOffsets() {
final Map<TopicPartition, Long> purgeableConsumedOffsets = new
HashMap<>();
- for (final Map.Entry<TopicPartition, Long> entry :
consumedOffsets.entrySet()) {
+ for (final Map.Entry<TopicPartition, Long> entry :
committedOffsets.entrySet()) {
final TopicPartition tp = entry.getKey();
if (topology.isRepartitionTopic(tp.topic())) {
- purgeableConsumedOffsets.put(tp, entry.getValue() + 1);
+ // committedOffsets map is initialized at -1 so no purging
until there's a committed offset
+ if (entry.getValue() > -1) {
+ purgeableConsumedOffsets.put(tp, entry.getValue() + 1);
Review Comment:
Do we still need to add `+ 1`? -- The old `consumedOffsets` sounds like the
offset of the last consumed message and thus adding `+ 1` is right, but for
`committedOffsets` I would assume that we did already add 1?
--
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]
