Rabbit is a set of tools for building applications with RabbitMQ.
The package can be installed by adding rabbit
to your list of dependencies in mix.exs
:
def deps do
[
{:rabbit, "~> 0.20"}
]
end
Please see HexDocs for additional documentation.
Connections form the basis of any application that is working with RabbitMQ. A connection module is needed by all the other modules included with Rabbit.
Upon start, a connection will automatically create a pool of RabbitMQ connections to utilize.
defmodule MyConnection do
use Rabbit.Connection
def start_link(opts \\ []) do
Rabbit.Connection.start_link(__MODULE__, opts, name: __MODULE__)
end
# Callbacks
@impl Rabbit.Connection
def init(:connection_pool, opts) do
# Perform runtime pool config
{:ok, opts}
end
def init(:connection, opts) do
# Perform runtime connection config
uri = System.get_env("RABBITMQ_URI") || "amqp://guest:guest@127.0.0.1:5672"
opts = Keyword.put(opts, :uri, uri)
{:ok, opts}
end
end
MyConnection.start_link()
Consumers are the "workers" of your application. They must be provided a connection
module and queue to consume. Every message received is then passed along to your
handle_message/1
callback within its own process.
You can optionally implement the handle_setup/2
callback to perform any work
needed to declare queues/exchanges/bindings.
defmodule MyConsumer do
use Rabbit.Consumer
def start_link(opts \\ []) do
Rabbit.Consumer.start_link(__MODULE__, opts, name: __MODULE__)
end
# Callbacks
@impl Rabbit.Consumer
def init(:consumer, opts) do
# Perform runtime config
{:ok, opts}
end
@impl Rabbit.Consumer
def handle_setup(state) do
# Optional callback to perform any exchange or queue setup
AMQP.Queue.declare(state.channel, state.queue)
:ok
end
@impl Rabbit.Consumer
def handle_message(message) do
# Handle message consumption
IO.inspect(message.payload)
{:ack, message}
end
@impl Rabbit.Consumer
def handle_error(message) do
# Handle message errors
{:nack, message}
end
end
MyConsumer.start_link(connection: MyConnection, queue: "my_queue", prefetch_count: 10)
Consumer supervisors provide an easy way to start and supervise multiple consumer processes. Rather than creating a module for each consumer and implementing repetitive logic - the same callbacks are used across all consumers.
defmodule MyConsumerSupervisor do
use Rabbit.ConsumerSupervisor
def start_link(consumers \\ []) do
Rabbit.ConsumerSupervisor.start_link(__MODULE__, consumers, name: __MODULE__)
end
# Callbacks
@impl Rabbit.ConsumerSupervisor
def init(:consumer_supervisor, _consumers) do
# Perform runtime config for the consumer supervisor
consumers = [
[connection: MyConnection, queue: "my_queue", prefetch_count: 5],
[connection: MyConnection, queue: "my_queue_2", prefetch_count: 10],
]
{:ok, consumers}
end
def init(:consumer, opts) do
# Perform runtime config per consumer
{:ok, opts}
end
@impl Rabbit.ConsumerSupervisor
def handle_setup(state) do
# Optional callback to perform any exchange or queue setup per consumer
AMQP.Queue.declare(state.channel, state.queue)
:ok
end
@impl Rabbit.ConsumerSupervisor
def handle_message(message) do
# Handle message consumption per consumer
IO.inspect(message.payload)
{:ack, message}
end
@impl Rabbit.ConsumerSupervisor
def handle_error(message) do
# Handle message errors per consumer
{:nack, message}
end
end
MyConsumerSupervisor.start_link()
In order to publish messages to RabbitMQ, we must create a producer module. They must be provided a connection module.
Upon start, a producer will automatically create a pool of RabbitMQ channels to publish from.
You can optionally implement the handle_setup/1
callback to perform any work
needed to declare queues/exchanges/bindings.
defmodule MyProducer do
use Rabbit.Producer
def start_link(opts \\ []) do
Rabbit.Producer.start_link(__MODULE__, opts, name: __MODULE__)
end
# Callbacks
@impl Rabbit.Producer
def init(:producer_pool, opts) do
# Perform runtime config for the producer pool
{:ok, opts}
end
def init(:producer, opts) do
# Perform runtime config per producer
{:ok, opts}
end
end
MyProducer.start_link(connection: MyConnection)
Rabbit.Producer.publish(MyProducer, "", "my_queue", "hello")
Topology provides a way to centralize any RabbitMQ setup required by your application. In that sense, it should be started BEFORE any of your producers or consumers.
Using a topology, you can automatically setup queues, exchanges and bindings with simple keyword lists.
defmodule MyTopology do
use Rabbit.Topology
def start_link(opts \\ []) do
Rabbit.Topology.start_link(__MODULE__, opts, name: __MODULE__)
end
# Callbacks
@impl Rabbit.Topology
def init(:topology, opts) do
# Perform runtime config
{:ok, opts}
end
end
MyTopology.start_link(
connection: MyConnection,
queues: [
[name: "my_queue", durable: true],
[name: "my_queue_2", durable: true],
],
exchanges: [
[name: "my_exchange"],
[name: "my_exchange_2", type: :fanout, durable: true],
],
bindings: [
[type: :queue, source: "my_exchange", destination: "my_queue", routing_key: "my_key"],
[type: :exchange, source: "my_exchange_2", destination: "my_exchange_1"]
]
)
Brokers encapsulate all of the above components into a single easy-to-use module. It provides a single place to handle your RabbitMQ connections, topology, producers and consumers.
defmodule MyBroker do
use Rabbit.Broker
def start_link(opts \\ []) do
Rabbit.Broker.start_link(__MODULE__, opts, name: __MODULE__)
end
# Callbacks
@impl Rabbit.Broker
# Perform runtime configuration per component
def init(:connection_pool, opts), do: {:ok, opts}
def init(:connection, opts), do: {:ok, opts}
def init(:topology, opts), do: {:ok, opts}
def init(:producer_pool, opts), do: {:ok, opts}
def init(:producer, opts), do: {:ok, opts}
def init(:consumer_supervisor, opts), do: {:ok, opts}
def init(:consumer, opts), do: {:ok, opts}
@impl Rabbit.Broker
def handle_message(message) do
# Handle message consumption per consumer
IO.inspect(message.payload)
{:ack, message}
end
@impl Rabbit.Broker
def handle_error(message) do
# Handle message errors per consumer
{:nack, message}
end
end
MyBroker.start_link(
connection: [uri: "amqp://guest:guest@127.0.0.1:5672"],
topology: [
queues: [
[name: "my_queue", durable: true],
[name: "my_queue_2", durable: true]
]
],
producer: [pool_size: 10],
consumers: [
[queue: "my_queue"],
[queue: "my_queue_2", prefetch_count: 10]
]
)
Rabbit.Broker.publish(MyBroker, "", "my_queue", "hello")