jolshan commented on code in PR #17402:
URL: https://github.com/apache/kafka/pull/17402#discussion_r1821581624
##########
core/src/main/scala/kafka/coordinator/transaction/TransactionMetadata.scala:
##########
@@ -346,14 +346,14 @@ private[transaction] class TransactionMetadata(val
transactionalId: String,
// Since the state change was successfully written to the log, unset the
flag for a failed epoch fence
hasFailedEpochFence = false
- val (updatedProducerId, updatedProducerEpoch) =
+ val (updatedProducerId, updatedProducerEpoch, updatedLastProducerEpoch) =
// If we overflowed on epoch bump, we have to set it as the producer ID
now the marker has been written.
if (clientTransactionVersion.supportsEpochBump() && nextProducerId !=
RecordBatch.NO_PRODUCER_ID) {
- (nextProducerId, 0.toShort)
+ (nextProducerId, 0.toShort, producerEpoch)
Review Comment:
Retry for this case is doesn't look at lastProducerEpoch, so either way is
fine there. I think the big question is if we get a benefit from initProducerId
by not updating the epoch here. The valid epoch check can be changed if we
decide not to update between prepare and complete.
Let me know if I'm missing something
--
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: jira-unsubscr...@kafka.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
