RabbitMQ Call Webhooks Plugin

This plugin extends RabbitMQ by forwarding messages from one or more queues to arbitrary HTTP endpoints (often called webhooks). It is inspired by the unmaintained rabbitmq-webhooks plugin by Jon Brisbin and is designed to work with modern RabbitMQ releases (3.10 and newer). The code provided here forms the basis for a complete plugin; it can be built into an .ez file and enabled on a running broker. The plugin connects to the broker using a direct AMQP connection, consumes messages from configured queues and then sends the message payload to the configured URL using the specified HTTP method. Messages are acknowledged only if the HTTP request succeeds (HTTP status 2xx–3xx), otherwise they are requeued for another attempt.

Features

• Per‑queue webhooks: configure one or more queues to be forwarded to different URLs. Each webhook entry defines the queue name, the target URL and the HTTP method to use.
• Direct connection: the plugin uses amqp_client's #amqp_params_direct{} type to connect to the same Erlang node that the broker is running on. No network sockets need to be opened and the plugin can authenticate using the guest user or another account.
• Synchronous HTTP calls: outgoing requests are sent using Erlang’s built‑in inets HTTP client (httpc). Only when a request returns a successful response does the plugin acknowledge the message; otherwise the message is nacked and requeued.
• Simple configuration: define webhook entries in either rabbitmq.conf format or the legacy rabbitmq.config format. Multiple webhooks can be defined under the same plugin.

Installation

1. Clone the plugin source into the RabbitMQ source tree or another directory under your build root. The plugin is structured like other RabbitMQ plugins (see the metronome example).

2. Build the plugin. RabbitMQ plugins are normally built with erlang.mk and the RabbitMQ‑specific rabbitmq-components.mk and rabbitmq-plugins.mk files. An example Makefile is provided in this directory. If you already build RabbitMQ from source, copy or symlink erlang.mk, rabbitmq-components.mk and rabbitmq-plugins.mk from the RabbitMQ source tree into this directory and run:

   make dist

   The resulting .ez archive will be written to the dist/ directory.

3. Install the compiled plugin by copying the .ez file into the broker’s plugins/ directory and enabling it:

   cp dist/rabbit_callwebhooks.ez $RABBITMQ_HOME/plugins/
   rabbitmq-plugins enable rabbit_callwebhooks

4. Configure the plugin (see below) and restart the broker.

Configuration

The plugin reads its configuration from the application environment key rabbit_callwebhooks under the webhooks entry. Each webhook entry must include a queue name, a URL and a method. Optional fields such as username, password and vhost can be provided when the queue lives in a vhost other than / or must be accessed by a different user.

Using the modern rabbitmq.conf format:

rabbit_callwebhooks.webhooks.1.queue = my.queue
rabbit_callwebhooks.webhooks.1.url   = https://example.com/webhook
rabbit_callwebhooks.webhooks.1.method = post

rabbit_callwebhooks.webhooks.2.queue = other.queue
rabbit_callwebhooks.webhooks.2.url   = https://api.example.net/endpoint
rabbit_callwebhooks.webhooks.2.method = put
rabbit_callwebhooks.webhooks.2.username = webhook_user
rabbit_callwebhooks.webhooks.2.password = s3cret
rabbit_callwebhooks.webhooks.2.vhost   = /myvhost

When using the legacy rabbitmq.config format (Erlang terms):

{rabbit_callwebhooks, [
  {webhooks, [
    #{queue => <<"my.queue">>,
      url   => <<"https://example.com/webhook">>,
      method => post},
    #{queue => <<"other.queue">>,
      url    => <<"https://api.example.net/endpoint">>,
      method => put,
      username => <<"webhook_user">>,
      password => <<"s3cret">>,
      vhost   => <<"/myvhost">>}
  ]}
]}.

File overview

Path	Purpose
src/rabbit_callwebhooks.app.src	Defines the OTP application for the plugin and its dependencies.
src/rabbit_callwebhooks_app.erl	Application behaviour implementation; starts the top‑level supervisor.
src/rabbit_callwebhooks_sup.erl	Supervisor that starts a worker process for each configured webhook.
src/rabbit_callwebhooks_worker.erl	Worker process that consumes messages from a queue and delivers them to an HTTP endpoint.
Makefile	Example build file for compiling the plugin using erlang.mk.

Limitations

• The HTTP client used is Erlang’s built‑in httpc from the inets application. For higher performance or advanced features (e.g. HTTPS client certificates), consider replacing it with another client such as hackney.
• Errors during HTTP delivery will result in messages being nacked and requeued. There is no dead‑letter handling built in.
• There is no batching or rate limiting; each message triggers one HTTP call.
• The plugin currently forwards the raw payload as the request body and does not set any additional headers. You can extend send_http/4 in rabbit_callwebhooks_worker.erl to include message headers or properties.

Acknowledgements

This plugin is heavily inspired by the historical rabbitmq-webhooks project【506323775653347†L260-L365】. Many thanks to Jon Brisbin for the original idea.