floci-io/floci · error · AwsException

ConflictException

ConflictException

Error message

Broker already exists: " + name

What it means

Floci keeps a per-instance registry of brokers and rejects CreateBroker when an existing broker already has the same BrokerName, returning 409 ConflictException. Broker names are unique per emulator state (unlike AWS where uniqueness is account+region scoped but effectively the same locally).

Source

Thrown at src/main/java/io/github/hectorvent/floci/services/amazonmq/AmazonMqService.java:99

            throw new AwsException("BadRequestException",
                    "Only SINGLE_INSTANCE DeploymentMode is supported", 400);
        }
        // RabbitMQ brokers require exactly one user at creation; that user becomes the
        // broker's RabbitMQ administrator (seeded into the container). This mirrors AWS,
        // which rejects CreateBroker for RabbitMQ unless exactly one user is supplied.
        List<MqUser> requestedUsers = params.users() == null ? List.of() : params.users();
        if (requestedUsers.size() != 1) {
            throw new AwsException("BadRequestException",
                    "Exactly one broker user is required for a RabbitMQ broker", 400);
        }
        MqUser admin = requestedUsers.get(0);
        if (admin.getUsername() == null || admin.getUsername().isBlank()) {
            throw new AwsException("BadRequestException", "Broker user username is required", 400);
        }
        validateUserPassword(admin.getPassword());

        if (storage.scan(k -> true).stream().anyMatch(b -> name.equals(b.getBrokerName()))) {
            throw new AwsException("ConflictException", "Broker already exists: " + name, 409);
        }

        String brokerId = "b-" + UUID.randomUUID();
        String accountId = regionResolver.getAccountId();
        String brokerArn = AwsArnUtils.Arn.of("mq", config.defaultRegion(), accountId,
                "broker:" + name + ":" + brokerId).toString();
        String engineVersion = (params.engineVersion() == null || params.engineVersion().isBlank())
                ? DEFAULT_ENGINE_VERSION : params.engineVersion();

        Broker broker = new Broker(brokerId, brokerArn, name, ENGINE_RABBITMQ,
                engineVersion, deploymentMode, params.hostInstanceType());
        broker.setAccountId(accountId);
        broker.setVolumeId(String.format("%06x", new SecureRandom().nextInt(0xFFFFFF)));
        broker.setPubliclyAccessible(params.publiclyAccessible());
        broker.setAutoMinorVersionUpgrade(params.autoMinorVersionUpgrade());
        if (params.users() != null) {
            broker.setUsers(new ArrayList<>(params.users()));
        }

View on GitHub (pinned to 62ff490619)

Solutions

  1. Delete the existing broker first: aws mq delete-broker --broker-id <id>, wait for it to disappear, then recreate
  2. Use a unique broker name per test run (e.g. suffix with a UUID or timestamp)
  3. If persistent storage holds stale brokers from earlier runs, clean the storage path or list-and-delete existing brokers in test setup

Example fix

# before
aws mq create-broker --broker-name demo ...   # second run -> ConflictException
# after
aws mq delete-broker --broker-id b-xxxx || true
aws mq create-broker --broker-name demo ...
Defensive patterns

Strategy: try-catch

Validate before calling

boolean nameTaken = mqClient.listBrokers().brokers().stream()
    .anyMatch(b -> name.equals(b.brokerName()));
if (nameTaken) throw new IllegalStateException("Broker name already in use: " + name);

Try / catch

try {
    mqClient.createBroker(req);
} catch (ConflictException e) {
    // idempotent create: treat existing broker as success, or delete-and-retry once
    String id = findBrokerIdByName(name);
    if (id == null) throw e;
    mqClient.deleteBroker(r -> r.brokerId(id));
    mqClient.createBroker(req);
}

Prevention

When it happens

Trigger: Running CreateBroker twice with the same BrokerName without an intervening DeleteBroker; re-running a test script or integration suite against a persistent/hybrid storage mode that retained the previous broker.

Common situations: Test suites that create named brokers and fail before cleanup; switching storage to persistent mode so brokers survive restarts while the setup code assumes a clean slate; retries of a create call that actually succeeded server-side.

Related errors

AI-assisted analysis of floci-io/floci@62ff490619 (2026-08-14). Data as JSON: /api/errors/105fb47dd7fce2fb. Report an issue: GitHub.