KAFKA:20612: Stitch dlq metrics with ShareGroupDLQStateManager.

smjn · smjn · commit bf53e66adc34 · 2026-05-29T17:29:06.000+05:30
diff --git a/core/src/main/java/kafka/server/share/SharePartitionManager.java b/core/src/main/java/kafka/server/share/SharePartitionManager.java
@@ -170,6 +170,7 @@ public SharePartitionManager(
         long remoteFetchMaxWaitMs,
         Persister persister,
         ShareGroupConfigProvider configProvider,
+        ShareGroupMetrics metrics,
         BrokerTopicStats brokerTopicStats,
         Supplier<Boolean> shareGroupDlqEnableSupplier,
         ShareGroupDLQManager shareGroupDLQManager
@@ -184,7 +185,7 @@ public SharePartitionManager(
             remoteFetchMaxWaitMs,
             persister,
             configProvider,
-            new ShareGroupMetrics(time),
+            metrics,
             brokerTopicStats,
             shareGroupDlqEnableSupplier,
             shareGroupDLQManager
@@ -634,7 +635,6 @@ private void addDelayedShareFetch(DelayedShareFetch delayedShareFetch, List<Dela
     @Override
     public void close() throws Exception {
         this.timer.close();
-        this.shareGroupMetrics.close();
     }
 
     private ShareSessionKey shareSessionKey(String groupId, String memberId) {
diff --git a/core/src/main/scala/kafka/server/BrokerServer.scala b/core/src/main/scala/kafka/server/BrokerServer.scala
@@ -63,6 +63,7 @@ import org.apache.kafka.storage.internals.log.{LogDirFailureChannel, LogManager
 import org.apache.kafka.storage.log.metrics.BrokerTopicStats
 import org.apache.kafka.server.partition.{AlterPartitionManager, DefaultAlterPartitionManager}
 import org.apache.kafka.server.share.dlq.{DefaultShareGroupDLQManager, NoOpShareGroupDLQManager, ShareGroupDLQManager}
+import org.apache.kafka.server.share.metrics.ShareGroupMetrics
 
 import java.time.Duration
 import java.util
@@ -165,6 +166,8 @@ class BrokerServer(
 
   var clientMetricsManager: ClientMetricsManager = _
 
+  var shareGroupMetrics: ShareGroupMetrics = _
+
   var sharePartitionManager: SharePartitionManager = _
 
   var persister: Persister = _
@@ -389,6 +392,9 @@ class BrokerServer(
       /* create persister */
       persister = createShareStatePersister()
 
+      /* create metrics object to be shared with share DLQ manager share partition manager*/
+      shareGroupMetrics = new ShareGroupMetrics(time)
+
       /* create share group DLQ manager */
       shareGroupDLQManager = createShareGroupDLQManager()
 
@@ -468,6 +474,7 @@ class BrokerServer(
         config.remoteLogManagerConfig.remoteFetchMaxWaitMs().toLong,
         persister,
         new ShareGroupConfigProvider(groupConfigManager),
+        shareGroupMetrics,
         brokerTopicStats,
         () => ShareVersion.fromFeatureLevel(metadataCache.features.finalizedFeatures.getOrDefault(ShareVersion.FEATURE_NAME, 0.toShort)).supportsShareGroupDLQ(),
         shareGroupDLQManager
@@ -765,7 +772,8 @@ class BrokerServer(
           NetworkUtils.buildNetworkClient("ShareGroupDLQManager", config, metrics, Time.SYSTEM, new LogContext(s"[ShareGroupDLQManager broker=${config.brokerId}]")),
           new ShareCoordinatorMetadataCacheHelperImpl(metadataCache, key => shareCoordinator.partitionFor(key), config.interBrokerListenerName, groupConfigManager),
           Time.SYSTEM,
-          shareGroupTimer
+          shareGroupTimer,
+          shareGroupMetrics
         )
       } else if (klass.getName.equals(classOf[NoOpShareGroupDLQManager].getName)) {
         info("Using no-op share group DLQ manager")
@@ -917,6 +925,9 @@ class BrokerServer(
       if (shareGroupDLQManager != null)
         Utils.swallow(this.logger.underlying, () => shareGroupDLQManager.stop())
 
+      if (shareGroupMetrics != null)
+        Utils.swallow(this.logger.underlying, () => shareGroupMetrics.close())
+
       Utils.closeQuietly(shareGroupTimer, "share group timer")
 
       if (lifecycleManager != null)
diff --git a/core/src/test/java/kafka/server/share/SharePartitionManagerTest.java b/core/src/test/java/kafka/server/share/SharePartitionManagerTest.java
@@ -1266,7 +1266,6 @@ public void testCloseSharePartitionManager() throws Exception {
         sharePartitionManager.close();
         // Verify that the timer object in sharePartitionManager is closed by checking the calls to timer.close() and shareGroupMetrics.close().
         Mockito.verify(timer, times(1)).close();
-        Mockito.verify(shareGroupMetrics, times(1)).close();
     }
 
     @Test
diff --git a/server/src/main/java/org/apache/kafka/server/share/dlq/DefaultShareGroupDLQManager.java b/server/src/main/java/org/apache/kafka/server/share/dlq/DefaultShareGroupDLQManager.java
@@ -19,6 +19,7 @@
 
 import org.apache.kafka.clients.KafkaClient;
 import org.apache.kafka.common.utils.Time;
+import org.apache.kafka.server.share.metrics.ShareGroupMetrics;
 import org.apache.kafka.server.util.timer.Timer;
 
 import org.slf4j.Logger;
@@ -39,14 +40,26 @@ public class DefaultShareGroupDLQManager implements ShareGroupDLQManager {
 
     private static final Logger log = LoggerFactory.getLogger(DefaultShareGroupDLQManager.class);
 
-    public static ShareGroupDLQManager instance(KafkaClient client, ShareGroupDLQMetadataCacheHelper cacheHelper, Time time, Timer timer) {
-        DefaultShareGroupDLQManager instance = new DefaultShareGroupDLQManager(client, cacheHelper, time, timer);
+    public static ShareGroupDLQManager instance(
+        KafkaClient client,
+        ShareGroupDLQMetadataCacheHelper cacheHelper,
+        Time time,
+        Timer timer,
+        ShareGroupMetrics metrics
+    ) {
+        DefaultShareGroupDLQManager instance = new DefaultShareGroupDLQManager(client, cacheHelper, time, timer, metrics);
         instance.start();
         return instance;
     }
 
-    private DefaultShareGroupDLQManager(KafkaClient client, ShareGroupDLQMetadataCacheHelper cacheHelper, Time time, Timer timer) {
-        this.stateManager = new ShareGroupDLQStateManager(client, cacheHelper, time, timer);
+    private DefaultShareGroupDLQManager(
+        KafkaClient client,
+        ShareGroupDLQMetadataCacheHelper cacheHelper,
+        Time time,
+        Timer timer,
+        ShareGroupMetrics shareGroupMetrics
+    ) {
+        this.stateManager = new ShareGroupDLQStateManager(client, cacheHelper, time, timer, shareGroupMetrics);
     }
 
     private void start() {
diff --git a/server/src/main/java/org/apache/kafka/server/share/dlq/ShareGroupDLQStateManager.java b/server/src/main/java/org/apache/kafka/server/share/dlq/ShareGroupDLQStateManager.java
@@ -45,6 +45,7 @@
 import org.apache.kafka.common.utils.Time;
 import org.apache.kafka.common.utils.internals.ExponentialBackoffManager;
 import org.apache.kafka.server.config.ServerConfigs;
+import org.apache.kafka.server.share.metrics.ShareGroupMetrics;
 import org.apache.kafka.server.util.InterBrokerSendThread;
 import org.apache.kafka.server.util.RequestAndCompletionHandler;
 import org.apache.kafka.server.util.timer.Timer;
@@ -80,6 +81,7 @@ public class ShareGroupDLQStateManager {
     private final Time time;
     private final Timer timer;
     private final ShareGroupDLQMetadataCacheHelper cacheHelper;
+    private final ShareGroupMetrics shareGroupMetrics;
     public static final long REQUEST_BACKOFF_MS = 1_000L;
     public static final long REQUEST_BACKOFF_MAX_MS = 30_000L;
     private static final int MAX_REQUEST_ATTEMPTS = 5;
@@ -91,7 +93,18 @@ public class ShareGroupDLQStateManager {
     private final Map<Node, List<ProduceRequestHandler>> nodeRPCMap = new HashMap<>();
     private final Object nodeMapLock = new Object();
 
-    public ShareGroupDLQStateManager(KafkaClient client, ShareGroupDLQMetadataCacheHelper cacheHelper, Time time, Timer timer) {
+    // Called when the generateRequests method is executed by InterBrokerSendThread, returning requests.
+    // Mainly for testing and introspection purpose to inspect the state of the nodeRPC map
+    // when generateRequests is called.
+    private Runnable generateCallback;
+
+    public ShareGroupDLQStateManager(
+        KafkaClient client,
+        ShareGroupDLQMetadataCacheHelper cacheHelper,
+        Time time,
+        Timer timer,
+        ShareGroupMetrics shareGroupMetrics
+    ) {
         if (client == null) {
             throw new IllegalArgumentException("Kafkaclient must not be null.");
         }
@@ -108,9 +121,14 @@ public ShareGroupDLQStateManager(KafkaClient client, ShareGroupDLQMetadataCacheH
             throw new IllegalArgumentException("Timer must not be null.");
         }
 
+        if (shareGroupMetrics == null) {
+            throw new IllegalArgumentException("ShareGroupMetrics must not be null.");
+        }
+
         this.time = time;
         this.timer = timer;
         this.cacheHelper = cacheHelper;
+        this.shareGroupMetrics = shareGroupMetrics;
         this.sender = new SendThread(
             "ShareGroupDLQSendThread",
             client,
@@ -154,6 +172,16 @@ CompletableFuture<Void> dlq(ShareGroupDLQRecordParameter param, long requestBack
         return future;
     }
 
+    // Visibility for tests
+    void setGenerateCallback(Runnable generateCallback) {
+        this.generateCallback = generateCallback;
+    }
+
+    // Visibility for tests
+    Map<Node, List<ShareGroupDLQStateManager.ProduceRequestHandler>> nodeRPCMap() {
+        return nodeRPCMap;
+    }
+
     private void enqueue(ProduceRequestHandler requestHandler) {
         sender.enqueue(requestHandler);
     }
@@ -333,6 +361,10 @@ public ProduceRequestData.TopicProduceData topicProduceData() {
                 simpleRecords.toArray(new SimpleRecord[]{})
             );
 
+            // Update the metric to say a new request is created to se sent. This might not be the
+            // actual RPC count as we coalesce the requests before sending.
+            shareGroupMetrics.recordDLQProduce(param.groupId());
+
             return new ProduceRequestData.TopicProduceData()
                 .setName(dlqTopicPartitionData.topicName())
                 .setTopicId(dlqTopicPartitionData.topicId().get())
@@ -580,6 +612,7 @@ private void handleProduceResponse(ClientResponse response) {
                     switch (error) {
                         case NONE:
                             LOG.debug("Successfully produced records {} to dlq topic node {}.", this, dlqPartitionLeaderNode());
+                            shareGroupMetrics.recordDLQRecordWrite(param.groupId(), (int) (param.lastOffset() - param.firstOffset() + 1));
                             produceRequestBackoff.resetAttempts();
                             this.result.complete(null);
                             break;
@@ -588,6 +621,7 @@ private void handleProduceResponse(ClientResponse response) {
                             LOG.debug("Received retriable error produce response for {} to dlq topic node {} - {}.", this, dlqPartitionLeaderNode(), errorMessage);
                             if (!produceRequestBackoff.canAttempt()) {
                                 LOG.error("Exhausted max retries to produce {} to  DLQ topic node {}.", this, dlqPartitionLeaderNode());
+                                shareGroupMetrics.recordDLQProduceFailed(param.groupId());
                                 requestErrorResponse(new Exception("Exhausted max retries to produce to DLQ topic without success."));
                                 break;
                             }
@@ -598,6 +632,7 @@ private void handleProduceResponse(ClientResponse response) {
                             LOG.error("Unable to produce {} to DLQ topic node {} - {}.", this, dlqPartitionLeaderNode(), errorMessage);
                             partitionResponse.recordErrors().forEach(recordError ->
                                 LOG.error("Records with errors {} - {}.", recordError.batchIndex(), recordError.batchIndexErrorMessage()));
+                            shareGroupMetrics.recordDLQProduceFailed(param.groupId());
                             requestErrorResponse(error.exception());
                     }
                     break;
@@ -609,6 +644,7 @@ private void handleProduceResponse(ClientResponse response) {
                     if (!produceRequestBackoff.canAttempt()) {
                         LOG.error("Exhausted max retries to produce {} to  DLQ topic node {} due to client response error {}.",
                             param, dlqPartitionLeaderNode(), clientResponseErrorMessage);
+                        shareGroupMetrics.recordDLQProduceFailed(param.groupId());
                         requestErrorResponse(clientResponseError.exception());
                         break;
                     }
@@ -618,6 +654,7 @@ private void handleProduceResponse(ClientResponse response) {
                 default:
                     LOG.error("Unable to produce {} to DLQ topic node {} due to client response error {}.",
                         param, dlqPartitionLeaderNode(), clientResponseErrorMessage);
+                    shareGroupMetrics.recordDLQProduceFailed(param.groupId());
                     requestErrorResponse(clientResponseError.exception());
             }
         }
@@ -634,6 +671,11 @@ private class SendThread extends InterBrokerSendThread {
 
         @Override
         public Collection<RequestAndCompletionHandler> generateRequests() {
+            // Introspection for testing - will be null in prod
+            if (generateCallback != null) {
+                generateCallback.run();
+            }
+
             List<RequestAndCompletionHandler> requests = new ArrayList<>();
 
             if (!queue.isEmpty()) {
@@ -772,13 +814,19 @@ public void run() {
         }
     }
 
-    private record CoalesceResults(
+    // Visibility for tests
+    record CoalesceResults(
         AbstractRequest.Builder<? extends AbstractRequest> request,
         List<ProduceRequestHandler> liveHandlers
     ) {
     }
 
-    private static CoalesceResults coalesceProduceRequests(List<ProduceRequestHandler> handlers) {
+    // Visibility for tests
+    static CoalesceResults coalesceProduceRequests(List<ProduceRequestHandler> handlers) {
+        // Above handlers are destined for the same broker node - it could be for different DLQ topics and partitions
+        // but the same broker node. Now the produce request requires each topic data request to be
+        // scoped to a specific topic/topicId and the partition data could have all the record information
+        // and the destination DLQ partition. To accomplish this, we will map handlers by DLQ topic id.
         Map<Uuid, ProduceRequestData.TopicProduceData> produceHandlerMap = new HashMap<>();
         List<ProduceRequestHandler> liveHandlers = new ArrayList<>(handlers.size());
         handlers.forEach(handler -> {
diff --git a/server/src/test/java/org/apache/kafka/server/share/dlq/ShareGroupDLQStateManagerTest.java b/server/src/test/java/org/apache/kafka/server/share/dlq/ShareGroupDLQStateManagerTest.java

Original file line number	Diff line number	Diff line change
`@@ -1266,7 +1266,6 @@ public void testCloseSharePartitionManager() throws Exception {`
`1266`	`1266`	`sharePartitionManager.close();`
`1267`	`1267`	`// Verify that the timer object in sharePartitionManager is closed by checking the calls to timer.close() and shareGroupMetrics.close().`
`1268`	`1268`	`Mockito.verify(timer, times(1)).close();`
`1269`		`- Mockito.verify(shareGroupMetrics, times(1)).close();`
`1270`	`1269`	`}`
`1271`	`1270`
`1272`	`1271`	`@Test`