AHeise commented on a change in pull request #17019:
URL: https://github.com/apache/flink/pull/17019#discussion_r698539256
##########
File path:
flink-connectors/flink-connector-kafka/src/main/java/org/apache/flink/connector/kafka/sink/KafkaWriter.java
##########
@@ -183,62 +181,83 @@ public void close() throws Exception {
closer.close();
}
- private KafkaWriterState recoverAndInitializeState(List<KafkaWriterState>
recoveredStates) {
- final int subtaskId = kafkaSinkContext.getParallelInstanceId();
- if (recoveredStates.isEmpty()) {
- final KafkaWriterState state =
- new KafkaWriterState(transactionalIdPrefix, subtaskId, 0);
- abortTransactions(getTransactionsToAbort(state, new
ArrayList<>()));
- return state;
+ private void abortLingeringTransactions(
+ List<KafkaWriterState> recoveredStates, long startCheckpointId) {
+ List<String> prefixesToAbort =
Lists.newArrayList(transactionalIdPrefix);
+
+ if (!recoveredStates.isEmpty()) {
+ KafkaWriterState lastState = recoveredStates.get(0);
+ if
(!lastState.getTransactionalIdPrefix().equals(transactionalIdPrefix)) {
Review comment:
Isn't this exactly what's done here?
We usually only try to abort transactions with the current prefix. When we
detect a change, we additionally check for transactions with the old prefix.
--
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]
