Skill v1.0.1
currentAutomated scan100/100+4 new
name: azure-eventhub-java description: "Build real-time streaming applications with Azure Event Hubs SDK for Java. Use when implementing event streaming, high-throughput data ingestion, or building event-driven architectures." risk: critical source: community date_added: "2026-02-27" metadata: version: "1.0.1"
Azure Event Hubs SDK for Java
Build real-time streaming applications using the Azure Event Hubs SDK for Java.
Installation
<dependency><groupId>com.azure</groupId><artifactId>azure-messaging-eventhubs</artifactId><version>5.19.0</version></dependency><!-- For checkpoint store (production) --><dependency><groupId>com.azure</groupId><artifactId>azure-messaging-eventhubs-checkpointstore-blob</artifactId><version>1.20.0</version></dependency>
Client Creation
EventHubProducerClient
import com.azure.messaging.eventhubs.EventHubProducerClient;import com.azure.messaging.eventhubs.EventHubClientBuilder;// With connection stringEventHubProducerClient producer = new EventHubClientBuilder().connectionString("<connection-string>", "<event-hub-name>").buildProducerClient();// Full connection string with EntityPathEventHubProducerClient producer = new EventHubClientBuilder().connectionString("<connection-string-with-entity-path>").buildProducerClient();
With DefaultAzureCredential
import com.azure.identity.DefaultAzureCredentialBuilder;EventHubProducerClient producer = new EventHubClientBuilder().fullyQualifiedNamespace("<namespace>.servicebus.windows.net").eventHubName("<event-hub-name>").credential(new DefaultAzureCredentialBuilder().build()).buildProducerClient();
EventHubConsumerClient
import com.azure.messaging.eventhubs.EventHubConsumerClient;EventHubConsumerClient consumer = new EventHubClientBuilder().connectionString("<connection-string>", "<event-hub-name>").consumerGroup(EventHubClientBuilder.DEFAULT_CONSUMER_GROUP_NAME).buildConsumerClient();
Async Clients
import com.azure.messaging.eventhubs.EventHubProducerAsyncClient;import com.azure.messaging.eventhubs.EventHubConsumerAsyncClient;EventHubProducerAsyncClient asyncProducer = new EventHubClientBuilder().connectionString("<connection-string>", "<event-hub-name>").buildAsyncProducerClient();EventHubConsumerAsyncClient asyncConsumer = new EventHubClientBuilder().connectionString("<connection-string>", "<event-hub-name>").consumerGroup("$Default").buildAsyncConsumerClient();
Pin compatible SDKs including identity/storage and add java.util/time imports; client declarations are alternatives. Scope producer/consumer/checkpoint roles and redact bodies/connection strings. Checkpoint only after durable effects and handle failure; it does not provide exactly-once processing. Preserve subscription handles for cancellation and close clients/stop processors on every exit. ErrorContext can lack partition context: guard it before reading partition ID. Batch retries must preserve partition options, and uncertain sends can duplicate events; use event IDs/reconciliation.
Core Patterns
Send Single Event
import com.azure.messaging.eventhubs.EventData;EventData eventData = new EventData("Hello, Event Hubs!");producer.send(Collections.singletonList(eventData));
Send Event Batch
import com.azure.messaging.eventhubs.EventDataBatch;import com.azure.messaging.eventhubs.models.CreateBatchOptions;// Create batchEventDataBatch batch = producer.createBatch();// Add events (returns false if batch is full)for (int i = 0; i < 100; i++) {EventData event = new EventData("Event " + i);if (!batch.tryAdd(event)) {// Batch is full, send and create new batchproducer.send(batch);batch = producer.createBatch();if (!batch.tryAdd(event)) throw new IllegalArgumentException("Event exceeds empty batch capacity");}}// Send remaining eventsif (batch.getCount() > 0) {producer.send(batch);}
Send to Specific Partition
CreateBatchOptions options = new CreateBatchOptions().setPartitionId("0");EventDataBatch batch = producer.createBatch(options);if (!batch.tryAdd(new EventData("Partition 0 event"))) throw new IllegalArgumentException("Event too large");producer.send(batch);
Send with Partition Key
CreateBatchOptions options = new CreateBatchOptions().setPartitionKey("customer-123");EventDataBatch batch = producer.createBatch(options);if (!batch.tryAdd(new EventData("Customer event"))) throw new IllegalArgumentException("Event too large");producer.send(batch);
Event with Properties
EventData event = new EventData("Order created");event.getProperties().put("orderId", "ORD-123");event.getProperties().put("customerId", "CUST-456");event.getProperties().put("priority", 1);producer.send(Collections.singletonList(event));
Receive Events (Simple)
import com.azure.messaging.eventhubs.models.EventPosition;import com.azure.messaging.eventhubs.models.PartitionEvent;// Receive from specific partitionIterable<PartitionEvent> events = consumer.receiveFromPartition("0", // partitionId10, // maxEventsEventPosition.earliest(), // startingPositionDuration.ofSeconds(30) // timeout);for (PartitionEvent partitionEvent : events) {EventData event = partitionEvent.getData();System.out.println("Body: " + event.getBodyAsString());System.out.println("Sequence: " + event.getSequenceNumber());System.out.println("Offset: " + event.getOffset());}
EventProcessorClient (Production)
import com.azure.messaging.eventhubs.EventProcessorClient;import com.azure.messaging.eventhubs.EventProcessorClientBuilder;import com.azure.messaging.eventhubs.checkpointstore.blob.BlobCheckpointStore;import com.azure.storage.blob.BlobContainerAsyncClient;import com.azure.storage.blob.BlobContainerClientBuilder;// Create checkpoint storeBlobContainerAsyncClient blobClient = new BlobContainerClientBuilder().connectionString("<storage-connection-string>").containerName("checkpoints").buildAsyncClient();// Create processorEventProcessorClient processor = new EventProcessorClientBuilder().connectionString("<eventhub-connection-string>", "<event-hub-name>").consumerGroup("$Default").checkpointStore(new BlobCheckpointStore(blobClient)).processEvent(eventContext -> {EventData event = eventContext.getEventData();System.out.println("Processing: " + event.getBodyAsString());// Checkpoint after processingeventContext.updateCheckpoint();}).processError(errorContext -> {System.err.println("Error: " + errorContext.getThrowable().getMessage());System.err.println("Partition: " + errorContext.getPartitionContext().getPartitionId());}).buildEventProcessorClient();// Start processingprocessor.start();// Keep running...Thread.sleep(Duration.ofMinutes(5).toMillis());// Stop gracefullyprocessor.stop();
Batch Processing
EventProcessorClient processor = new EventProcessorClientBuilder().connectionString("<connection-string>", "<event-hub-name>").consumerGroup("$Default").checkpointStore(new BlobCheckpointStore(blobClient)).processEventBatch(eventBatchContext -> {List<EventData> events = eventBatchContext.getEvents();System.out.printf("Received %d events%n", events.size());for (EventData event : events) {// Process each eventSystem.out.println(event.getBodyAsString());}// Checkpoint after batcheventBatchContext.updateCheckpoint();}, 50) // maxBatchSize.processError(errorContext -> {System.err.println("Error: " + errorContext.getThrowable());}).buildEventProcessorClient();
Async Receiving
asyncConsumer.receiveFromPartition("0", EventPosition.latest()).subscribe(partitionEvent -> {EventData event = partitionEvent.getData();System.out.println("Received: " + event.getBodyAsString());},error -> System.err.println("Error: " + error),() -> System.out.println("Complete"));
Get Event Hub Properties
// Get hub infoEventHubProperties hubProps = producer.getEventHubProperties();System.out.println("Hub: " + hubProps.getName());System.out.println("Partitions: " + hubProps.getPartitionIds());// Get partition infoPartitionProperties partitionProps = producer.getPartitionProperties("0");System.out.println("Begin sequence: " + partitionProps.getBeginningSequenceNumber());System.out.println("Last sequence: " + partitionProps.getLastEnqueuedSequenceNumber());System.out.println("Last offset: " + partitionProps.getLastEnqueuedOffset());
Event Positions
// Start from beginningEventPosition.earliest()// Start from end (new events only)EventPosition.latest()// From specific offsetEventPosition.fromOffset(12345L)// From specific sequence numberEventPosition.fromSequenceNumber(100L)// From specific timeEventPosition.fromEnqueuedTime(Instant.now().minus(Duration.ofHours(1)))
Error Handling
import com.azure.messaging.eventhubs.models.ErrorContext;.processError(errorContext -> {Throwable error = errorContext.getThrowable();String partitionId = errorContext.getPartitionContext().getPartitionId();if (error instanceof AmqpException) {AmqpException amqpError = (AmqpException) error;if (amqpError.isTransient()) {System.out.println("Transient error, will retry");}}System.err.printf("Error on partition %s: %s%n", partitionId, error.getMessage());})
Resource Cleanup
// Always close clientstry {producer.send(batch);} finally {producer.close();}// Or use try-with-resourcestry (EventHubProducerClient producer = new EventHubClientBuilder().connectionString(connectionString, eventHubName).buildProducerClient()) {producer.send(events);}
Environment Variables
EVENT_HUBS_CONNECTION_STRING=Endpoint=sb://<namespace>.servicebus.windows.net/;SharedAccessKeyName=...EVENT_HUBS_NAME=<event-hub-name>STORAGE_CONNECTION_STRING=<for-checkpointing>
Best Practices
- Use EventProcessorClient: For production, provides load balancing and checkpointing
- Batch Events: Use
EventDataBatchfor efficient sending - Partition Keys: Use for ordering guarantees within a partition
- Checkpointing: Checkpoint after durable processing; replay remains possible, so consumers must be idempotent
- Error Handling: Handle transient errors with retries
- Close Clients: Always close producer/consumer when done
Trigger Phrases
- "Event Hubs Java"
- "event streaming Azure"
- "real-time data ingestion"
- "EventProcessorClient"
- "event hub producer consumer"
- "partition processing"
When to Use
This skill is applicable to execute the workflow or actions described in the overview.
Limitations
- Use this skill only when the task clearly matches the scope described above.
- Do not treat the output as a substitute for environment-specific validation, testing, or expert review.
- Stop and ask for clarification if required inputs, permissions, safety boundaries, or success criteria are missing.