KAFKA-19274; Group Coordinator Shards are not unloaded when `__consumer_offsets` topic is deleted (#19713)

Group Coordinator Shards are not unloaded when `__consumer_offsets`
topic is deleted. The unloading is scheduled but it is ignored because
the epoch is equal to the current epoch:

```
[2025-05-13 08:46:00,883] INFO [GroupCoordinator id=1]
Scheduling  unloading of metadata for __consumer_offsets-0 with epoch
OptionalInt[0]
(org.apache.kafka.coordinator.common.runtime.CoordinatorRuntime)
[2025-05-13 08:46:00,883] INFO [GroupCoordinator id=1] Scheduling
unloading of metadata for __consumer_offsets-1 with epoch OptionalInt[0]
(org.apache.kafka.coordinator.common.runtime.CoordinatorRuntime)
[2025-05-13 08:46:00,883] INFO [GroupCoordinator id=1] Ignored unloading
metadata for __consumer_offsets-0 in epoch OptionalInt[0] since current
epoch is 0.
(org.apache.kafka.coordinator.common.runtime.CoordinatorRuntime)
[2025-05-13 08:46:00,883] INFO [GroupCoordinator id=1] Ignored unloading
metadata for __consumer_offsets-1 in epoch OptionalInt[0] since current
epoch is 0.
(org.apache.kafka.coordinator.common.runtime.CoordinatorRuntime)
```

This patch fixes the issue by not setting the leader epoch in this case.
The coordinator expects the leader epoch to be incremented when the
resignation code is called. When the topic is deleted, the epoch is not
incremented. Therefore, we must not use it. Note that this is aligned
with deleted partitions are handled too.

Reviewers: Dongnuo Lyu <dlyu@confluent.io>, José Armando García Sancio <jsancio@apache.org>

core/src/main/scala/kafka/server/metadata/BrokerMetadataPublisher.scala

@@ -287,6 +287,11 @@ class BrokerMetadataPublisher(
   /**
    * Update the coordinator of local replica changes: election and resignation.
    *
+   * When the topic is deleted or a partition of the topic is deleted, {@param resignation}
+   * callback must be called with {@code None}. The coordinator expects the leader epoch to be
+   * incremented when the {@param resignation} callback is called but the leader epoch
+   * is not incremented when a topic is deleted.
+   *
    * @param image latest metadata image
    * @param delta metadata delta from the previous image and the latest image
    * @param topicName name of the topic associated with the coordinator
@@ -307,7 +312,7 @@ class BrokerMetadataPublisher(
       if (topicsDelta.topicWasDeleted(topicName)) {
         topicsDelta.image.getTopic(topicName).partitions.entrySet.forEach { entry =>
           if (entry.getValue.leader == brokerId) {
-            resignation(entry.getKey, Some(entry.getValue.leaderEpoch))
+            resignation(entry.getKey, None)
           }
         }
       }

core/src/test/scala/integration/kafka/api/GroupCoordinatorIntegrationTest.scala

@@ -16,8 +16,8 @@ import org.apache.kafka.common.test.api.{ClusterConfigProperty, ClusterTest, Typ
 import kafka.utils.TestUtils
 import org.apache.kafka.clients.admin.{Admin, ConsumerGroupDescription}
 import org.apache.kafka.clients.consumer.{Consumer, GroupProtocol, OffsetAndMetadata}
-import org.apache.kafka.common.errors.GroupIdNotFoundException
+import org.apache.kafka.common.errors.{GroupIdNotFoundException, UnknownTopicOrPartitionException}
-import org.apache.kafka.common.{ConsumerGroupState, GroupType, KafkaFuture, TopicPartition}
+import org.apache.kafka.common.{ConsumerGroupState, GroupType, KafkaFuture, TopicCollection, TopicPartition}
 import org.junit.jupiter.api.Assertions._
 
 import scala.jdk.CollectionConverters._
@@ -27,11 +27,13 @@ import org.apache.kafka.common.test.ClusterInstance
 import org.apache.kafka.coordinator.group.GroupCoordinatorConfig
 import org.apache.kafka.server.config.ServerConfigs
 import org.apache.kafka.storage.internals.log.UnifiedLog
+import org.apache.kafka.test.{TestUtils => JTestUtils}
 import org.junit.jupiter.api.Timeout
 
 import java.time.Duration
 import java.util.Collections
 import java.util.concurrent.TimeUnit
+import scala.concurrent.ExecutionException
 
 @Timeout(120)
 class GroupCoordinatorIntegrationTest(cluster: ClusterInstance) {
@@ -278,6 +280,58 @@ class GroupCoordinatorIntegrationTest(cluster: ClusterInstance) {
     }
   }
 
+  @ClusterTest(
+    types = Array(Type.KRAFT),
+    serverProperties = Array(
+      new ClusterConfigProperty(key = GroupCoordinatorConfig.OFFSETS_TOPIC_PARTITIONS_CONFIG, value = "1"),
+      new ClusterConfigProperty(key = GroupCoordinatorConfig.OFFSETS_TOPIC_REPLICATION_FACTOR_CONFIG, value = "1")
+    )
+  )
+  def testRecreatingConsumerOffsetsTopic(): Unit = {
+    withAdmin { admin =>
+      TestUtils.createTopicWithAdminRaw(
+        admin = admin,
+        topic = "foo",
+        numPartitions = 3
+      )
+
+      withConsumer(groupId = "group", groupProtocol = GroupProtocol.CONSUMER) { consumer =>
+        consumer.subscribe(List("foo").asJava)
+        TestUtils.waitUntilTrue(() => {
+          consumer.poll(Duration.ofMillis(50))
+          consumer.assignment().asScala.nonEmpty
+        }, msg = "Consumer did not get an non empty assignment")
+      }
+
+      admin
+        .deleteTopics(TopicCollection.ofTopicNames(List(Topic.GROUP_METADATA_TOPIC_NAME).asJava))
+        .all()
+        .get()
+
+      TestUtils.waitUntilTrue(() => {
+        try {
+          admin
+            .describeTopics(TopicCollection.ofTopicNames(List(Topic.GROUP_METADATA_TOPIC_NAME).asJava))
+            .topicNameValues()
+            .get(Topic.GROUP_METADATA_TOPIC_NAME)
+            .get(JTestUtils.DEFAULT_MAX_WAIT_MS, TimeUnit.MILLISECONDS)
+          false
+        } catch {
+          case e: ExecutionException =>
+            e.getCause.isInstanceOf[UnknownTopicOrPartitionException]
+        }
+      }, msg = s"${Topic.GROUP_METADATA_TOPIC_NAME} was not deleted")
+
+      withConsumer(groupId = "group", groupProtocol = GroupProtocol.CONSUMER) { consumer =>
+        consumer.subscribe(List("foo").asJava)
+        TestUtils.waitUntilTrue(() => {
+          consumer.poll(Duration.ofMillis(50))
+          consumer.assignment().asScala.nonEmpty
+        }, msg = "Consumer did not get an non empty assignment")
+      }
+    }
+  }
+
   private def rollAndCompactConsumerOffsets(): Unit = {
     val tp = new TopicPartition("__consumer_offsets", 0)
     val broker = cluster.brokers.asScala.head._2

core/src/test/scala/unit/kafka/server/metadata/BrokerMetadataPublisherTest.scala

@@ -28,9 +28,11 @@ import kafka.server.{BrokerServer, KafkaConfig, ReplicaManager}
 import kafka.utils.TestUtils
 import org.apache.kafka.clients.admin.AlterConfigOp.OpType.SET
 import org.apache.kafka.clients.admin.{Admin, AlterConfigOp, ConfigEntry, NewTopic}
+import org.apache.kafka.common.Uuid
 import org.apache.kafka.common.config.ConfigResource
 import org.apache.kafka.common.config.ConfigResource.Type.BROKER
-import org.apache.kafka.common.metadata.FeatureLevelRecord
+import org.apache.kafka.common.internals.Topic
+import org.apache.kafka.common.metadata.{FeatureLevelRecord, PartitionRecord, RemoveTopicRecord, TopicRecord}
 import org.apache.kafka.common.test.{KafkaClusterTestKit, TestKitNodes}
 import org.apache.kafka.common.utils.Exit
 import org.apache.kafka.coordinator.group.GroupCoordinator
@@ -183,6 +185,70 @@ class BrokerMetadataPublisherTest {
     }
   }
 
+  @Test
+  def testGroupCoordinatorTopicDeletion(): Unit = {
+    val config = KafkaConfig.fromProps(TestUtils.createBrokerConfig(0))
+    val metadataCache = new KRaftMetadataCache(0, () => KRaftVersion.KRAFT_VERSION_1)
+    val logManager = mock(classOf[LogManager])
+    val replicaManager = mock(classOf[ReplicaManager])
+    val groupCoordinator = mock(classOf[GroupCoordinator])
+    val faultHandler = mock(classOf[FaultHandler])
+
+    val metadataPublisher = new BrokerMetadataPublisher(
+      config,
+      metadataCache,
+      logManager,
+      replicaManager,
+      groupCoordinator,
+      mock(classOf[TransactionCoordinator]),
+      mock(classOf[ShareCoordinator]),
+      mock(classOf[DynamicConfigPublisher]),
+      mock(classOf[DynamicClientQuotaPublisher]),
+      mock(classOf[DynamicTopicClusterQuotaPublisher]),
+      mock(classOf[ScramPublisher]),
+      mock(classOf[DelegationTokenPublisher]),
+      mock(classOf[AclPublisher]),
+      faultHandler,
+      faultHandler,
+      mock(classOf[SharePartitionManager])
+    )
+
+    val topicId = Uuid.randomUuid()
+    var delta = new MetadataDelta(MetadataImage.EMPTY)
+    delta.replay(new TopicRecord()
+      .setName(Topic.GROUP_METADATA_TOPIC_NAME)
+      .setTopicId(topicId)
+    )
+    delta.replay(new PartitionRecord()
+      .setTopicId(topicId)
+      .setPartitionId(0)
+      .setLeader(config.brokerId)
+    )
+    delta.replay(new PartitionRecord()
+      .setTopicId(topicId)
+      .setPartitionId(1)
+      .setLeader(config.brokerId)
+    )
+    val image = delta.apply(MetadataProvenance.EMPTY)
+
+    delta = new MetadataDelta(image)
+    delta.replay(new RemoveTopicRecord()
+      .setTopicId(topicId)
+    )
+
+    metadataPublisher.onMetadataUpdate(delta, delta.apply(MetadataProvenance.EMPTY),
+      LogDeltaManifest.newBuilder()
+        .provenance(MetadataProvenance.EMPTY)
+        .leaderAndEpoch(LeaderAndEpoch.UNKNOWN)
+        .numBatches(1)
+        .elapsedNs(100)
+        .numBytes(42)
+        .build())
+
+    verify(groupCoordinator).onResignation(0, OptionalInt.empty())
+    verify(groupCoordinator).onResignation(1, OptionalInt.empty())
+  }
+
   @Test
   def testNewImagePushedToGroupCoordinator(): Unit = {
     val config = KafkaConfig.fromProps(TestUtils.createBrokerConfig(0))