How to solve the problem of duplicate notifications when there are multiple instances of the job service?

The problem of duplicate notifications when running multiple instances of a job service occurs because each instance independently processes the same tasks.

Solutions:

1. Using distributed locking — before sending a notification, an instance attempts to acquire a lock in a distributed storage (e.g., Redis with Redlock). Only the instance that acquires the lock sends the notification.

2. Message queue with guaranteed single processing — use a message broker (Kafka, RabbitMQ) configured so that each notification is processed exactly once by a single consumer.

3. Identification and deduplication of notifications — add a unique identifier to each notification and check in the database or cache whether it has already been sent.

4. Leader election — one instance becomes the leader and is responsible for sending notifications, others do not.

Example with Redis Redlock (Java):

// Acquire lock before sending notification
RLock lock = redissonClient.getLock("notification_lock");
boolean isLocked = lock.tryLock(100, 1000, TimeUnit.MILLISECONDS);
if (isLocked) {
    try {
        // Send notification
    } finally {
        lock.unlock();
    }
}