From 01e9918530edd32986215ad016b678822d7bb131 Mon Sep 17 00:00:00 2001 From: "Kuan-Po (Cooper) Tseng" Date: Wed, 5 Jun 2024 07:56:18 +0800 Subject: [PATCH] KAFKA-16814 KRaft broker cannot startup when `partition.metadata` is missing (#16165) When starting up kafka logManager, we'll check stray replicas to avoid some corner cases. But this check might cause broker unable to startup if partition.metadata is missing because when startup kafka, we load log from file, and the topicId of the log is coming from partition.metadata file. So, if partition.metadata is missing, the topicId will be None, and the LogManager#isStrayKraftReplica will fail with no topicID error. The partition.metadata missing could be some storage failure, or another possible path is unclean shutdown after topic is created in the replica, but before data is flushed into partition.metadata file. This is possible because we do the flush in async way here. When finding a log without topicID, we should treat it as a stray log and then delete it. Reviewers: Luke Chen , Gaurav Narula --- .../src/main/scala/kafka/log/LogManager.scala | 11 +- .../server/LogManagerIntegrationTest.java | 137 ++++++++++++++++++ .../scala/unit/kafka/log/LogManagerTest.scala | 7 + 3 files changed, 152 insertions(+), 3 deletions(-) create mode 100644 core/src/test/java/kafka/server/LogManagerIntegrationTest.java diff --git a/core/src/main/scala/kafka/log/LogManager.scala b/core/src/main/scala/kafka/log/LogManager.scala index d7599e569ab25..45d6ab7908db5 100755 --- a/core/src/main/scala/kafka/log/LogManager.scala +++ b/core/src/main/scala/kafka/log/LogManager.scala @@ -1604,11 +1604,16 @@ object LogManager { newTopicsImage: TopicsImage, log: UnifiedLog ): Boolean = { - val topicId = log.topicId.getOrElse { - throw new RuntimeException(s"The log dir $log does not have a topic ID, " + - "which is not allowed when running in KRaft mode.") + if (log.topicId.isEmpty) { + // Missing topic ID could result from storage failure or unclean shutdown after topic creation but before flushing + // data to the `partition.metadata` file. And before appending data to the log, the `partition.metadata` is always + // flushed to disk. So if the topic ID is missing, it mostly means no data was appended, and we can treat this as + // a stray log. + info(s"The topicId does not exist in $log, treat it as a stray log") + return true } + val topicId = log.topicId.get val partitionId = log.topicPartition.partition() Option(newTopicsImage.getPartition(topicId, partitionId)) match { case Some(partition) => diff --git a/core/src/test/java/kafka/server/LogManagerIntegrationTest.java b/core/src/test/java/kafka/server/LogManagerIntegrationTest.java new file mode 100644 index 0000000000000..709454beccd6b --- /dev/null +++ b/core/src/test/java/kafka/server/LogManagerIntegrationTest.java @@ -0,0 +1,137 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one or more + * contributor license agreements. See the NOTICE file distributed with + * this work for additional information regarding copyright ownership. + * The ASF licenses this file to You under the Apache License, Version 2.0 + * (the "License"); you may not use this file except in compliance with + * the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ +package kafka.server; + +import kafka.test.ClusterInstance; +import kafka.test.annotation.ClusterTest; +import kafka.test.annotation.Type; +import kafka.test.junit.ClusterTestExtensions; +import kafka.test.junit.RaftClusterInvocationContext; +import org.apache.kafka.clients.admin.Admin; +import org.apache.kafka.clients.admin.NewTopic; +import org.apache.kafka.clients.consumer.Consumer; +import org.apache.kafka.clients.consumer.ConsumerConfig; +import org.apache.kafka.clients.consumer.ConsumerRecord; +import org.apache.kafka.clients.consumer.ConsumerRecords; +import org.apache.kafka.clients.consumer.KafkaConsumer; +import org.apache.kafka.clients.producer.KafkaProducer; +import org.apache.kafka.clients.producer.Producer; +import org.apache.kafka.clients.producer.ProducerConfig; +import org.apache.kafka.clients.producer.ProducerRecord; +import org.apache.kafka.common.TopicPartition; +import org.apache.kafka.common.TopicPartitionInfo; +import org.apache.kafka.common.serialization.StringDeserializer; +import org.apache.kafka.common.serialization.StringSerializer; +import org.apache.kafka.storage.internals.checkpoint.PartitionMetadataFile; +import org.apache.kafka.test.TestUtils; +import org.junit.jupiter.api.Tag; +import org.junit.jupiter.api.extension.ExtendWith; + +import java.io.IOException; +import java.time.Duration; +import java.util.ArrayList; +import java.util.Collections; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.Optional; +import java.util.UUID; +import java.util.concurrent.ExecutionException; + +import static org.junit.jupiter.api.Assertions.assertDoesNotThrow; +import static org.junit.jupiter.api.Assertions.assertEquals; +import static org.junit.jupiter.api.Assertions.assertFalse; +import static org.junit.jupiter.api.Assertions.assertTrue; + +@ExtendWith(value = ClusterTestExtensions.class) +@Tag("integration") +public class LogManagerIntegrationTest { + private final ClusterInstance cluster; + + public LogManagerIntegrationTest(ClusterInstance cluster) { + this.cluster = cluster; + } + + @ClusterTest(types = {Type.KRAFT, Type.CO_KRAFT}, brokers = 3) + public void testRestartBrokerNoErrorIfMissingPartitionMetadata() throws IOException, ExecutionException, InterruptedException { + RaftClusterInvocationContext.RaftClusterInstance raftInstance = + (RaftClusterInvocationContext.RaftClusterInstance) cluster; + + try (Admin admin = cluster.createAdminClient()) { + admin.createTopics(Collections.singletonList(new NewTopic("foo", 1, (short) 3))).all().get(); + } + cluster.waitForTopic("foo", 1); + + Optional partitionMetadataFile = Optional.ofNullable( + raftInstance.getUnderlying().brokers().get(0).logManager() + .getLog(new TopicPartition("foo", 0), false).get() + .partitionMetadataFile().getOrElse(null)); + assertTrue(partitionMetadataFile.isPresent()); + + raftInstance.getUnderlying().brokers().get(0).shutdown(); + try (Admin admin = cluster.createAdminClient()) { + TestUtils.waitForCondition(() -> { + List partitionInfos = admin.describeTopics(Collections.singletonList("foo")) + .topicNameValues().get("foo").get().partitions(); + return partitionInfos.get(0).isr().size() == 2; + }, "isr size is not shrink to 2"); + } + + // delete partition.metadata file here to simulate the scenario that partition.metadata not flush to disk yet + partitionMetadataFile.get().delete(); + assertFalse(partitionMetadataFile.get().exists()); + raftInstance.getUnderlying().brokers().get(0).startup(); + // make sure there is no error during load logs + assertDoesNotThrow(() -> raftInstance.getUnderlying().fatalFaultHandler().maybeRethrowFirstException()); + try (Admin admin = cluster.createAdminClient()) { + TestUtils.waitForCondition(() -> { + List partitionInfos = admin.describeTopics(Collections.singletonList("foo")) + .topicNameValues().get("foo").get().partitions(); + return partitionInfos.get(0).isr().size() == 3; + }, "isr size is not expand to 3"); + } + + // make sure topic still work fine + Map producerConfigs = new HashMap<>(); + producerConfigs.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, cluster.bootstrapServers()); + producerConfigs.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class.getName()); + producerConfigs.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class.getName()); + + try (Producer producer = new KafkaProducer<>(producerConfigs)) { + producer.send(new ProducerRecord<>("foo", 0, null, "bar")).get(); + producer.flush(); + } + + Map consumerConfigs = new HashMap<>(); + consumerConfigs.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, cluster.bootstrapServers()); + consumerConfigs.put(ConsumerConfig.GROUP_ID_CONFIG, UUID.randomUUID().toString()); + consumerConfigs.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class.getName()); + consumerConfigs.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class.getName()); + + try (Consumer consumer = new KafkaConsumer<>(consumerConfigs)) { + consumer.assign(Collections.singletonList(new TopicPartition("foo", 0))); + consumer.seekToBeginning(Collections.singletonList(new TopicPartition("foo", 0))); + List values = new ArrayList<>(); + ConsumerRecords records = consumer.poll(Duration.ofMinutes(1)); + for (ConsumerRecord record : records) { + values.add(record.value()); + } + assertEquals(1, values.size()); + assertEquals("bar", values.get(0)); + } + } +} diff --git a/core/src/test/scala/unit/kafka/log/LogManagerTest.scala b/core/src/test/scala/unit/kafka/log/LogManagerTest.scala index 13ce4d28e9d3d..6e92094007152 100755 --- a/core/src/test/scala/unit/kafka/log/LogManagerTest.scala +++ b/core/src/test/scala/unit/kafka/log/LogManagerTest.scala @@ -1286,6 +1286,13 @@ class LogManagerTest { onDisk.foreach(log => assertEquals(expectedStrays.contains(log.topicPartition), LogManager.isStrayKraftReplica(0, image, log))) } + @Test + def testIsStrayKraftMissingTopicId(): Unit = { + val log = Mockito.mock(classOf[UnifiedLog]) + Mockito.when(log.topicId).thenReturn(Option.empty) + assertTrue(LogManager.isStrayKraftReplica(0, topicsImage(Seq()), log)) + } + @Test def testFindStrayReplicasInEmptyLAIR(): Unit = { val onDisk = Seq(foo0, foo1, bar0, bar1, baz0, baz1, baz2, quux0)