Sending RabbitMq messages between Docker containers using docker-compose

Question

So my goal is to have several containers that interact among themselves using rabbitmq messages(rabbitmq server is in a separate container)

rabbit.py

class Rabbit:
    host = 'rabbitmq-host'
    userid = 'test'
    password = 'test'


class Consumer(Rabbit):

    def __init__(self, exchange_name):
        self.exchange_name = exchange_name
        self.connection = None
        try:
            credentials = pika.PlainCredentials(self.userid, self.password)
            params = pika.ConnectionParameters(self.host, 5672, '/', credentials)
            self.connection = pika.BlockingConnection(params)
        except Exception as ex:
            print(ex)
            if self.connection is not None and self.connection.is_open:
                self.connection.close()
            raise ex

        self.channel = self.connection.channel()

Credentials test:test exist, I re-checked.

Then from another file(main.py) the Consumer is created

c = Consumer('media')

docker-compose.yml

version: '3'
services:

rabbitmq-server:
  image: "rabbitmq:3-management"
  hostname: "rabbitmq-host"
  environment:
    RABBITMQ_ERLANG_COOKIE: "SWQOKODSQALRPCLNMEQG"
    RABBITMQ_DEFAULT_USER: "test"
    RABBITMQ_DEFAULT_PASS: "test"
    RABBITMQ_DEFAULT_VHOST: "/"
  ports:
    - "15672:15672"
    - "5672:5672"
  labels:
    NAME: "rabbitmq1"


info-getter:
  build: ./info-getter
  depends_on:
    - rabbitmq-server
  tty: true
  ports:
    - "3000:3000"

When testing, I run rabbitmq server in a container, but the app from local mechine, all is working, but when I run 'docker-compose up' I get this exception:

info-getter_1      | 
info-getter_1      | Traceback (most recent call last):
info-getter_1      |   File "main.py", line 10, in <module>
info-getter_1      |     c = Consumer('media')
info-getter_1      |   File "libs/rabbit.py", line 27, in __init__
info-getter_1      |     raise ex
info-getter_1      |   File "libs/rabbit.py", line 22, in __init__
info-getter_1      |     self.connection = pika.BlockingConnection(params)
info-getter_1      |   File "/usr/local/lib/python3.6/site-packages/pika/adapters/blocking_connection.py", line 360, in __init__
info-getter_1      |     self._impl = self._create_connection(parameters, _impl_class)
info-getter_1      |   File "/usr/local/lib/python3.6/site-packages/pika/adapters/blocking_connection.py", line 451, in _create_connection
info-getter_1      |     raise self._reap_last_connection_workflow_error(error)
info-getter_1      | pika.exceptions.AMQPConnectionError
services_info-getter_1 exited with code 1

Where am I going wrong?

Also I would like to add that even though I have 'depends_on' specified, after I run 'docker-compose up' the info-getter log appears before the rabbitmq-server log.

Answer 1

The host needs to be the name you defined in docker-compose.yml:

my-worker:
    image: my-worker-image
    restart: always
    depends_on:
        - my-rabbitmq
my-rabbitmq:
    image: rabbitmq:management
    ports:
    - 5672:5672
    - 15672:15672

then in your consumer (python):

connection = pika.BlockingConnection(pika.ConnectionParameters('my-rabbitmq'))

Answer 2

Usually the error message contains additional status - for example:

pika.exceptions.AMQPConnectionError: Connection to :5672 failed: [Errno 111] Connection refused

This error is caused usually because the RabbitMQ workers/clients failed to connect to the RabbitMQ-server.
This might happen from a number of reasons - I'll name a few:

1) The IP of the RabbitMQ-server wasn't specified correctly.
It might no passed has a environment variable (this is why its important to add a check for that in the code).
When working with docker-compose the IP of the RabbitMQ-server should be replaced with the service DNS (which is the name of the service in the docker-compose.yml file or the value of hostname property if specified).

2) The RabbitMQ workers/clients tried to reach the RabbitMQ-server before its ready.
Notice that depend_on only express dependency between services, and wait for services to start but not to be ready.
So you can't rely on it by adding:

  depends_on:
    - rabbitmq-server

Because the RabbitMQ-server service bootstrap phase take time.
See the reference below regarding the depend_on command.
See the solutions I provided for this runtime dependency problem below.

3) There is a RabbitMQ-server service already running on the host which uses the 5672 port.
In this case you'll receive an explicit error when you try to start the RabbitMQ-server service, but from the RabbitMQ-worker perspective its the same problem.

---

You can solve the runtime dependency problem mentioned in the #2 by:

A) Had a retry logic in the client - Consider using pluggins like Shoval and Federation.

B) If the cause for the problem is #2 - you can use the restart_policy option and the connection will succeed after a few retries.

C) Use a tool such as wait-for-it, dockerize, or sh-compatible wait-for.
These are small wrapper scripts which you can include in your application’s image to poll a given host and port until it’s accepting TCP connections. Read more in here.

D) Execute docker-compose up rabbitmq-server and only after service is ready execute the other services.

E) Use a time interval (e.g sleep 10) in the workers execution command (I won't recommend this approach).

---

Regarding from depends_on from the Docker-Compose docs:

There are several things to be aware of when using depends_on:

depends_on does not wait for db and redis to be “ready” before starting web - only until they have been started. If you need to wait for a service to be ready, see Controlling startup order for more on this problem and strategies for solving it.
Version 3 no longer supports the condition form of depends_on.
The depends_on option is ignored when deploying a stack in swarm mode with a version 3 Compose file.