[fix][client] Fix failover/exclusive consumer with batch cumulate ack issue. (#18454)

Technoboy- · web-flow · commit 7712aa396bfc · 2022-11-15T11:37:16.000+08:00
diff --git a/pulsar-broker/src/test/java/org/apache/pulsar/client/impl/ConsumerDedupPermitsUpdateTest.java b/pulsar-broker/src/test/java/org/apache/pulsar/client/impl/ConsumerDedupPermitsUpdateTest.java
@@ -116,10 +116,23 @@ public void testConsumerDedup(boolean batchingEnabled, int receiverQueueSize) th
         }
         producer.flush();
 
-        for (int i = 0; i < 30; i++) {
-            Message<String> msg = consumer.receive();
-            assertEquals(msg.getValue(), "new-message-" + i);
-            consumer.acknowledge(msg);
+        if (batchingEnabled) {
+            for (int i = 0; i < 30; i++) {
+                Message<String> msg = consumer.receive();
+                assertEquals(msg.getValue(), "hello-" + i);
+                consumer.acknowledge(msg);
+            }
+            for (int i = 0; i < 30; i++) {
+                Message<String> msg = consumer.receive();
+                assertEquals(msg.getValue(), "new-message-" + i);
+                consumer.acknowledge(msg);
+            }
+        } else {
+            for (int i = 0; i < 30; i++) {
+                Message<String> msg = consumer.receive();
+                assertEquals(msg.getValue(), "new-message-" + i);
+                consumer.acknowledge(msg);
+            }
         }
     }
 
diff --git a/pulsar-broker/src/test/java/org/apache/pulsar/client/impl/NegativeAcksTest.java b/pulsar-broker/src/test/java/org/apache/pulsar/client/impl/NegativeAcksTest.java
@@ -24,6 +24,7 @@
 
 import java.util.HashSet;
 import java.util.Set;
+import java.util.concurrent.CountDownLatch;
 import java.util.concurrent.TimeUnit;
 import lombok.Cleanup;
 import lombok.extern.slf4j.Slf4j;
@@ -35,6 +36,7 @@
 import org.apache.pulsar.client.api.ProducerConsumerBase;
 import org.apache.pulsar.client.api.Schema;
 import org.apache.pulsar.client.api.SubscriptionType;
+import org.testng.Assert;
 import org.testng.annotations.AfterClass;
 import org.testng.annotations.BeforeClass;
 import org.testng.annotations.DataProvider;
@@ -362,4 +364,79 @@ public void testNegativeAcksWithBatchAckEnabled() throws Exception {
         // There should be no more messages
         assertNull(consumer.receive(100, TimeUnit.MILLISECONDS));
     }
+
+    @Test
+    public void testFailoverConsumerBatchCumulateAck() throws Exception {
+        final String topic = BrokerTestUtil.newUniqueName("my-topic");
+        admin.topics().createPartitionedTopic(topic, 2);
+
+        @Cleanup
+        Consumer<Integer> consumer = pulsarClient.newConsumer(Schema.INT32)
+                .topic(topic)
+                .subscriptionName("sub")
+                .subscriptionType(SubscriptionType.Failover)
+                .enableBatchIndexAcknowledgment(true)
+                .acknowledgmentGroupTime(100, TimeUnit.MILLISECONDS)
+                .receiverQueueSize(10)
+                .subscribe();
+
+        @Cleanup
+        Producer<Integer> producer = pulsarClient.newProducer(Schema.INT32)
+                .topic(topic)
+                .batchingMaxMessages(10)
+                .batchingMaxPublishDelay(3, TimeUnit.SECONDS)
+                .blockIfQueueFull(true)
+                .create();
+
+        int count = 0;
+        Set<Integer> datas = new HashSet<>();
+        CountDownLatch producerLatch = new CountDownLatch(10);
+        while (count < 10) {
+            datas.add(count);
+            producer.sendAsync(count).whenComplete((m, e) -> {
+                producerLatch.countDown();
+            });
+            count++;
+        }
+        producerLatch.await();
+        CountDownLatch consumerLatch = new CountDownLatch(1);
+        new Thread(new Runnable() {
+            @Override
+            public void run() {
+                consumer.receiveAsync()
+                        .thenCompose(m -> {
+                            log.info("received one msg : {}", m.getMessageId());
+                            datas.remove(m.getValue());
+                            return consumer.acknowledgeCumulativeAsync(m);
+                        })
+                        .thenAccept(ignore -> {
+                            try {
+                                Thread.sleep(500);
+                                consumer.redeliverUnacknowledgedMessages();
+                            } catch (Exception e) {
+                                throw new RuntimeException(e);
+                            }
+                        })
+                        .whenComplete((r, e) -> {
+                            consumerLatch.countDown();
+                        });
+            }
+        }).start();
+        consumerLatch.await();
+        Thread.sleep(500);
+        count = 0;
+        while(true) {
+            Message<Integer> msg = consumer.receive(5, TimeUnit.SECONDS);
+            if (msg == null) {
+                break;
+            }
+            consumer.acknowledgeCumulative(msg);
+            Thread.sleep(200);
+            datas.remove(msg.getValue());
+            log.info("received msg : {}", msg.getMessageId());
+            count++;
+        }
+        Assert.assertEquals(count, 9);
+        Assert.assertEquals(0, datas.size());
+    }
 }
diff --git a/pulsar-client/src/main/java/org/apache/pulsar/client/impl/ConsumerImpl.java b/pulsar-client/src/main/java/org/apache/pulsar/client/impl/ConsumerImpl.java
@@ -1275,9 +1275,9 @@ void messageReceived(CommandMessage cmdMessage, ByteBuf headersAndPayload, Clien
         final int numMessages = msgMetadata.getNumMessagesInBatch();
         final int numChunks = msgMetadata.hasNumChunksFromMsg() ? msgMetadata.getNumChunksFromMsg() : 0;
         final boolean isChunkedMessage = numChunks > 1;
-
         MessageIdImpl msgId = new MessageIdImpl(messageId.getLedgerId(), messageId.getEntryId(), getPartitionIndex());
-        if (acknowledgmentsGroupingTracker.isDuplicate(msgId)) {
+        if (numMessages == 1 && !msgMetadata.hasNumMessagesInBatch()
+                && acknowledgmentsGroupingTracker.isDuplicate(msgId)) {
             if (log.isDebugEnabled()) {
                 log.debug("[{}] [{}] Ignoring message as it was already being acked earlier by same consumer {}/{}",
                         topic, subscription, consumerName, msgId);
@@ -1541,7 +1541,10 @@ void receiveIndividualMessagesFromBatch(BrokerEntryMetadata brokerEntryMetadata,
                         skippedMessages++;
                         continue;
                     }
-
+                }
+                if (acknowledgmentsGroupingTracker.isDuplicate(message.getMessageId())) {
+                    skippedMessages++;
+                    continue;
                 }
                 executeNotifyCallback(message);
             }
diff --git a/pulsar-client/src/main/java/org/apache/pulsar/client/impl/MultiTopicsConsumerImpl.java b/pulsar-client/src/main/java/org/apache/pulsar/client/impl/MultiTopicsConsumerImpl.java
@@ -261,7 +261,11 @@ private void receiveMessageFromConsumer(ConsumerImpl<T> consumer, boolean batchR
                 return;
             }
             // Process the message, add to the queue and trigger listener or async callback
-            messages.forEach(msg -> messageReceived(consumer, msg));
+            messages.forEach(msg -> {
+                if (isValidConsumerEpoch((MessageImpl<T>) msg)) {
+                    messageReceived(consumer, msg);
+                }
+            });
 
             int size = incomingMessages.size();
             int maxReceiverQueueSize = getCurrentReceiverQueueSize();
