RabbitMQ smart home integration, free and open source

The health of your RabbitMQ broker in Gladys: queues, messages, consumers, alarms, outages.

• Published by guim31
• Devices
• Version 1.0.0
• Requires Gladys >=5.1.0
• ★ 0

View on GitHubAll integrations

Watches the health of your RabbitMQ broker in Gladys: connections, channels, consumers, messages ready and unacknowledged, publish and delivery rates, memory and disk alarms, running nodes, uptime and outages, plus the queues you choose to follow (messages waiting, consumers).

This integration is not affiliated with Broadcom or the RabbitMQ project.

Overview

RabbitMQ carries the messages of many self-hosted services (Home Assistant add-ons, Node-RED flows, Celery, n8n, custom scripts, MQTT through its plugin). When a consumer stops, its queue fills up silently; when memory or disk runs low, RabbitMQ blocks every publisher. This integration tells Gladys: every 30 seconds it reads the management API of RabbitMQ and turns what it says into Gladys devices, seven scene triggers, two scene actions and a dashboard widget.

It only reads: it never publishes, purges or deletes anything, never opens an AMQP connection and never changes the configuration of RabbitMQ. Everything stays on your local network.

Requirements

• Gladys 5.1 or later.
• RabbitMQ 3.x or 4.x (tested with 4.4.0) with the management plugin enabled (rabbitmq-plugins enable rabbitmq_management; it is already in the rabbitmq:*-management Docker images), its port (15672 by default) reachable from Gladys.
• A RabbitMQ user and its password. The monitoring tag is enough: no administrator needed.

Create a read-only user

On the RabbitMQ host (or docker exec into its container):

rabbitmqctl add_user gladys 'a-long-password'
rabbitmqctl set_user_tags gladys monitoring
## Access to the virtual host whose queues you follow ("/" here), without any right on its messages:
rabbitmqctl set_permissions -p / gladys '^$' '^$' '^$'

The monitoring tag gives the overview, the nodes and their alarms. A queue is only visible in a virtual host the user may access: the permissions above (empty patterns: no configure, no write, no read) grant that access and nothing else. Repeat the last line for every other virtual host of a followed queue, otherwise RabbitMQ answers as if the queue did not exist. The same can be done in the management interface: Admin → Users → Add a user, tag Monitoring, then Set permission.

Configuration

Field	Description
Address of RabbitMQ	The address of the management interface: http://192.168.1.20:15672, or a bare host (15672 is added).
User	The user created above.
Password	Its password.
Accept a self-signed certificate	For an https:// address whose certificate Gladys cannot check.
Queues to follow	Separated by commas, 20 at most. Each becomes a device. Leave empty to follow none.
Language of the texts	Feature names of the devices when added, scene sentences and messages of the actions (French by default).

How to write a queue:

• zigbee: the queue zigbee of the default virtual host /;
• iot/sensors: the queue sensors of the virtual host iot;
• /home/kitchen: the queue home/kitchen of the virtual host / (a leading / for a name holding a slash).

The Check the connection button reads RabbitMQ at once and tells its version, its key numbers, its alarms in progress and which followed queues exist.

Then open the Discovery tab of the integration and add the broker and the queues you want.

The broker device

Feature	Meaning
Reachable	1 while the API answers, 0 during an outage.
Connections	Client connections (AMQP, MQTT, STOMP…) open right now.
Channels	Channels open in those connections.
Consumers	Consumers of all the queues.
Queues	Queues of all the virtual hosts the user sees.
Messages ready	Messages waiting in all the queues, not yet delivered.
Messages unacknowledged	Messages delivered to a consumer that has not confirmed them yet.
Messages published (/min)	Messages received by the broker, per minute.
Messages delivered (/min)	Messages delivered to consumers (or fetched), per minute.
Memory alarm	1 while a node is above its memory watermark: RabbitMQ blocks the publishers.
Disk alarm	1 while a node is below its free disk limit: RabbitMQ blocks the publishers.
Running nodes	Nodes of the cluster running (1 for a single broker).
Uptime	In days, since the youngest node started.
Version	Version of RabbitMQ.

The rates are computed by RabbitMQ itself over its last 5-second sample. During an outage only Reachable changes: the other values keep their last reading.

Followed queues

Each queue of the Queues to follow field becomes a device named after it, with five features: Messages ready, Messages unacknowledged, Consumers, Messages published (/min) and Messages delivered (/min).

When a followed queue does not exist (not declared yet, deleted, an auto-delete queue whose consumer left), its values keep their last reading, the logs say so once, and Check the connection lists it as not found.

Dashboard widget

The RabbitMQ broker widget shows the state of the broker, its version, its uptime, the nodes of a cluster, the alarms in progress, one row per followed queue added to Gladys (messages ready and consumers, in orange when nobody consumes it), and four live tiles: messages ready, published and delivered per minute, connections. Pick the broker in its settings.

Scene triggers

Trigger	When
Broker unreachable	The API has not answered for 1, 5, 15, 30 or 60 minutes (your choice).
Broker back	The API answers again after an outage of at least one minute.
Broker restarted	The uptime of a node went back to zero, even without a visible outage.
Alarm raised	Memory alarm, disk alarm, network partition or a node of the cluster stopped (filterable).
Alarm cleared	One of those alarms is no longer in progress. Variable: its duration in minutes.
Queue without consumer	A followed queue that had consumers has none for a minute.
Queue above a threshold	More messages ready in a followed queue than the chosen number; at most every 30 minutes.

The alarm variable is memory, disk, partition or node_down. After a restart of the integration, the first reading is a reference: an alarm already in progress, a queue already without consumer or already above a threshold fires nothing.

Scene actions

Broker summary gives a sentence ready for a message ("RabbitMQ 4.4.0 has been running for 3 days, 5 connections, 12 messages ready and 0 unacknowledged, 40 messages published and 38 delivered per minute, no alarm.") and its numbers: reachable, connections, queues, messages ready and unacknowledged, published and delivered per minute, alarms in progress and their names, uptime in hours, version.

Queue summary does the same for a followed queue you pick ("Queue zigbee: 7 messages ready, consumers: 1, 2 unacknowledged.") with its numbers and whether it exists.

Scene ideas

• Queue without consumer on the queue of your Node-RED flow → send yourself "Nobody reads {{queue}} any more, {{messages_ready}} messages waiting".
• Alarm raised (filter: disk) → warn that the disk of the broker is almost full: every publisher is blocked.
• Queue above a threshold (1,000 messages) → turn a light orange.
• Every morning, Broker summary → send the summary sentence to Telegram.

Troubleshooting

• RabbitMQ cannot be reached: check the address and the port (15672 by default, not the AMQP port 5672), and that Gladys reaches it (firewall, Docker network).
• RabbitMQ refused the user: check its name and password, and that it has the monitoring tag. A user without any tag cannot use the management API.
• No management API at this address: the plugin is not enabled, or the address points to something else (AMQP port, another service).
• A followed queue is not found: check its exact name and virtual host on the Queues and Streams page (names are case-sensitive), and that the user has a permission on that virtual host (rabbitmqctl list_user_permissions gladys).
• The rates stay at 0: the metrics collector of RabbitMQ may be disabled (management_agent.disable_metrics_collector).
• More details: set LOG_LEVEL=debug and read the logs of the integration.

Limits

• Read every 30 seconds: a consumer that leaves for less than a minute fires nothing.
• The node statistics are refreshed every 5 seconds by RabbitMQ: an alarm shows a few seconds late.
• 20 followed queues at most; a queue name holding a comma cannot be followed.
• Only reading: no purge, no deletion, no message published.
• Exchanges, bindings, shovels and federation links are not watched.

Configuration settings

These are the settings RabbitMQ asks for in its configuration screen in Gladys.

Setting	Type	Required	Description
Connect RabbitMQ	section	No	The integration reads the management API of RabbitMQ every 30 seconds. It never changes anything. A user with the monitoring tag is enough: rabbitmqctl set_user_tags <user> monitoring.
Address of RabbitMQ	string	Yes	The address of the management interface, port 15672 by default (http:// or https://).
User	string	Yes	Read access is enough: a user with the monitoring tag, no administrator.
Password	secret	Yes
Accept a self-signed certificate	boolean	No	https:// only: the certificate of RabbitMQ is then not checked.
Queues to follow	string	No	Separated by commas, 20 at most: each one becomes a device (messages, consumers). Write name for the virtual host /, vhost/name for another one. The user needs a permission on that virtual host.
Language of the texts	select	No	Feature names of the devices when they are added, scene sentences and action messages.

How to install RabbitMQ in Gladys

1. In Gladys, open Integrations: RabbitMQ appears in the catalog, next to the native integrations, with a community badge.
2. Click Install. Gladys pulls the Docker image (ghcr.io/guim31/gladys-rabbitmq:1.0.0), starts it in a sandbox isolated from the core, and generates the integration's interface (devices, discovery and configuration).
3. Open the Configuration screen of the integration, fill in the settings, and save.
4. You can also install it directly from its repository URL: https://github.com/guim31/gladys-rabbitmq.

RabbitMQ requires Gladys >=5.1.0. The catalog inside Gladys refreshes every hour, so a new version becomes available at most one hour after its release.

Not running Gladys yet? It is free and open source: follow the installation guide to get started.

About external integrations

RabbitMQ is an external integration: a community integration packaged as a Docker container and published on GitHub, that Gladys installs in one click and runs in a sandbox isolated from its core. It is published and maintained by guim31, not by the Gladys core team.

• Browse all integrations, native and community
• Build and publish your own external integration
• Source code on GitHub — source of this documentation

Subscribe to the Gladys Assistant newsletter

A few emails per month about new releases and project news. Sent by Pierre-Gilles Leymarie, founder of the project. Unsubscribe anytime 🙂