apache/rocketmq · error · MQClientException

The producer group[{producerGroup}] has been created before,

Error message

The producer group[{producerGroup}] has been created before, specify another name please.

What it means

MQClientException thrown from DefaultMQProducerImpl.start() when registerProducer(group, this) returns false, i.e. a producer with the same group name is already registered in this JVM's MQClientInstance (keyed by clientId). RocketMQ requires producer group names to be unique within a client instance because the group maps 1:1 to the producer implementation inside the shared factory. The companion FAQ link points at the duplicate group name page.

Source

Thrown at client/src/main/java/org/apache/rocketmq/client/impl/producer/DefaultMQProducerImpl.java:261

    public void start(final boolean startFactory) throws MQClientException {
        switch (this.serviceState) {
            case CREATE_JUST:
                this.serviceState = ServiceState.START_FAILED;

                this.checkConfig();

                if (!this.defaultMQProducer.getProducerGroup().equals(MixAll.CLIENT_INNER_PRODUCER_GROUP)) {
                    this.defaultMQProducer.changeInstanceNameToPID();
                }

                this.mQClientFactory = MQClientManager.getInstance().getOrCreateMQClientInstance(this.defaultMQProducer, rpcHook);

                defaultMQProducer.initProduceAccumulator();

                boolean registerOK = mQClientFactory.registerProducer(this.defaultMQProducer.getProducerGroup(), this);
                if (!registerOK) {
                    this.serviceState = ServiceState.CREATE_JUST;
                    throw new MQClientException("The producer group[" + this.defaultMQProducer.getProducerGroup()
                        + "] has been created before, specify another name please." + FAQUrl.suggestTodo(FAQUrl.GROUP_NAME_DUPLICATE_URL),
                        null);
                }

                if (startFactory) {
                    mQClientFactory.start();
                }

                this.initTopicRoute();

                this.mqFaultStrategy.startDetector();

                log.info("the producer [{}] start OK. sendMessageWithVIPChannel={}", this.defaultMQProducer.getProducerGroup(),
                    this.defaultMQProducer.isSendMessageWithVIPChannel());
                this.serviceState = ServiceState.RUNNING;
                break;
            case RUNNING:
            case START_FAILED:

View on GitHub (pinned to 293f588571)

Solutions

Use one shared DefaultMQProducer per group name per JVM (cache/singleton it) instead of creating new instances
If you truly need multiple producers, give each a distinct producerGroup name
Ensure producer.shutdown() fully ran (it unregisters the group) before creating a replacement with the same group
Set a distinct instanceName/unitName on the ClientConfig so each producer gets a different MQClientInstance if isolation is intended

Example fix

// before
DefaultMQProducer p = new DefaultMQProducer("order-producer");
p.start(); // called again elsewhere with same group -> exception
// after
private static volatile DefaultMQProducer shared;
if (shared == null) { shared = new DefaultMQProducer("order-producer"); shared.start(); }

Defensive patterns

Strategy: validation

Validate before calling

 private final Set<String> registeredGroups = ConcurrentHashMap.newKeySet();
public DefaultMQProducer getOrCreate(String group) throws MQClientException {
    if (!registeredGroups.add(group)) {
        return existingProducers.get(group); // reuse, don't create duplicate
    }
    DefaultMQProducer p = new DefaultMQProducer(group);
    p.start();
    existingProducers.put(group, p);
    return p;
}

Try / catch

try {
    producer.start();
} catch (MQClientException e) {
    if (e.getMessage() != null && e.getMessage().contains("has been created before")) {
        return existingProducerFor(group); // reuse the already-registered instance
    }
    throw e;
}

Prevention

Cache producers by group name in a registry/map; never construct a second producer for the same group in one JVM
Wrap producers in a singleton bean (Spring) so lifecycle is managed once
Always call shutdown() in @PreDestroy / finally so group names are released before re-creation
Give producers distinct instanceName via setInstanceName when intentional isolation is needed

When it happens

Trigger: Creating and starting two DefaultMQProducer instances with the same producerGroup and the same clientId (same unit name / instance settings) in one JVM; starting a producer, shutting it down incompletely, and starting a new one with the same group before unregistration completes; re-start() after a partially failed start.

Common situations: Singleton wrappers (Spring beans, connection pools) that lazily create a producer per request without caching by group name; code that calls producer.start() in a loop or in @PostConstruct of a prototype-scoped bean; tests that build a new producer per test method but share the JVM and instance name.

Related errors

AI-assisted analysis of apache/rocketmq@293f588571 (2026-08-14). Data as JSON: /api/errors/8b4070fab9bc242b. Report an issue: GitHub.