AHeise commented on code in PR #154:
URL:
https://github.com/apache/flink-connector-kafka/pull/154#discussion_r1969172042
##########
flink-connector-kafka/src/main/java/org/apache/flink/connector/kafka/sink/ExactlyOnceKafkaWriter.java:
##########
@@ -136,13 +158,10 @@ public void initialize() {
}
private FlinkKafkaInternalProducer<byte[], byte[]> startTransaction(long
checkpointId) {
+ namingContext.setNextCheckpointId(checkpointId);
FlinkKafkaInternalProducer<byte[], byte[]> producer =
- producerPool.getTransactionalProducer(
- TransactionalIdFactory.buildTransactionalId(
- transactionalIdPrefix,
- kafkaSinkContext.getParallelInstanceId(),
- checkpointId),
- checkpointId);
+
transactionNamingStrategy.getTransactionalProducer(namingContext);
+ namingContext.setLastCheckpointId(checkpointId);
Review Comment:
Not sure how to improve it. There could be an arbitrary number of aborted
transactions between the two invocations of the call-site. For me, these are
two disjunct states that are coupled in the best case but not in the general
case. Remember that we need to `setLastCheckpointId` once already during
recovery.
--
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]
