How does rabbitmq sync messages?

I use this simple code using pika:

conn = pika.BlockingConnection(pika.ConnectionParameters(host='localhost'))
channel = conn.channel()

def on_message(ch, method, properties, message):
     print("Received message: %r" % message)
     channel.basic_ack(delivery_tag=method.delivery_tag)

channel.queue_declare(queue='my_queue', durable=True)
channel.basic_qos(prefetch_count=1) 
channel.basic_consume(on_message, queue='my_queue')
channel.start_consuming() 

Question:
Once I got a message, and my on_message callback is running, how can rabbitmq make sure not to give the same message to other worker, until I call basic_ack ? Since in my tests, if I won't call basic_ack the message will be sent again if I run the code multiple times. Does rabbitmq has some timeout from the moment it send the message, until it gets an ack?

how can rabbitmq make sure not to give the same message to other worker, until I call basic_ack ?

Delivering a message to only one consumer is a core feature of RabbitMQ. The message will remain un-acked until the consumer to which it was delivered acks it, or the channel on which it was delivered closes.

in my tests, if I won't call basic_ack the message will be sent again if I run the code multiple times.

When you run your code multiple times, here is what happens:

• RabbitMQ delivers the message to your consumer, moving it to the un-acked state.
• You stop your consumer application without calling basic_ack, which closes the channel and connection without acknowledging the message.
• RabbitMQ detects this and re-enqueues the message.
• You re-start your consumer, and it is re-delivered the message.
• Lather, rinse, repeat.

All of this and much, much more is covered in the RabbitMQ documentation.