jolshan commented on code in PR #14088:
URL: https://github.com/apache/kafka/pull/14088#discussion_r1275549290


##########
clients/src/test/java/org/apache/kafka/clients/producer/internals/SenderTest.java:
##########
@@ -3107,6 +3099,45 @@ public void 
testReceiveFailedBatchTwiceWithTransactions() throws Exception {
         txnManager.beginTransaction();
     }
 
+    @Test
+    public void testInvalidTxnStateIsAnAbortableError() throws Exception {
+        ProducerIdAndEpoch producerIdAndEpoch = new 
ProducerIdAndEpoch(123456L, (short) 0);
+        apiVersions.update("0", 
NodeApiVersions.create(ApiKeys.INIT_PRODUCER_ID.id, (short) 0, (short) 3));
+        TransactionManager txnManager = new TransactionManager(logContext, 
"testInvalidTxnState", 60000, 100, apiVersions);
+
+        setupWithTransactionState(txnManager);
+        doInitTransactions(txnManager, producerIdAndEpoch);
+
+        txnManager.beginTransaction();
+        txnManager.maybeAddPartition(tp0);
+        client.prepareResponse(buildAddPartitionsToTxnResponseData(0, 
Collections.singletonMap(tp0, Errors.NONE)));
+        sender.runOnce();
+
+        Future<RecordMetadata> request = appendToAccumulator(tp0);
+        sender.runOnce();  // send request
+        sendIdempotentProducerResponse(0, tp0, Errors.INVALID_TXN_STATE, -1);
+
+        // Return InvalidTxnState error. It should be abortable.
+        sender.runOnce();
+        assertFutureFailure(request, InvalidTxnStateException.class);
+        assertTrue(txnManager.hasAbortableError());
+        TransactionalRequestResult result = txnManager.beginAbort();
+        sender.runOnce();
+
+        // Once the transaction is aborted, we should be able to begin a new 
one.
+        respondToEndTxn(Errors.NONE);
+        sender.runOnce();
+        assertTrue(txnManager::isInitializing);
+        prepareInitProducerResponse(Errors.NONE, 
producerIdAndEpoch.producerId, producerIdAndEpoch.epoch);

Review Comment:
   The places we call:
   ```
    if (canBumpEpoch()) {
                               epochBumpRequired = true;
                               KafkaException exception = new 
KafkaException(unackedMessagesErr + "It is safe to abort " +
                                       "the transaction and continue.");
                               transitionToAbortableError(exception);
   ```
   
   ```
       public synchronized void maybeTransitionToErrorState(RuntimeException 
exception) {
           if (exception instanceof ClusterAuthorizationException
                   || exception instanceof TransactionalIdAuthorizationException
                   || exception instanceof ProducerFencedException
                   || exception instanceof UnsupportedVersionException) {
               transitionToFatalError(exception);
           } else if (isTransactional()) {
               if (canBumpEpoch() && !isCompleting()) {
                   epochBumpRequired = true;
               }
               transitionToAbortableError(exception);
           }
   ```
   and a third that sometimes does and sometimes doesn't.
   
   
   I think the path for this specific produce request error is the second one 
so we may not bump epoch in some cases?
   
                                       



-- 
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

Reply via email to