README
¶
Mosquitto Exporter
Prometheus exporter for the Mosquitto MQTT message broker.
There is a docker image available:
docker run \
-p 9234:9234 jryberg/mosquitto-exporter:v0.8.0 \
--endpoint tcp://mosquitto:1883
Usage:
NAME:
mosquitto-exporter - Prometheus exporter for broker metrics
USAGE:
mosquitto-exporter [global options] command [command options] [arguments...]
VERSION:
v0.8.0 (2bbf83770521c99ccd126b4c28ae76332754ff22), go1.27.3
AUTHORS:
Johan Ryberg <johan@securit.se>
Arturo Reuschenbach Puncernau <a.reuschenbach.puncernau@sap.com>
Fabian Ruff <fabian.ruff@sap.com>
COMMANDS:
help, h Shows a list of commands or help for one command
GLOBAL OPTIONS:
--endpoint value, -e value Endpoint for the Mosquitto message broker (default: "tcp://127.0.0.1:1883") [$BROKER_ENDPOINT]
--bind-address value, -b value Listen address for metrics HTTP endpoint (default: "0.0.0.0:9234") [$BIND_ADDRESS]
--user value, -u value Username for the Mosquitto message broker [$MQTT_USER]
--pass value, -p value Password for the User on the Mosquitto message broker [$MQTT_PASS]
--cert value, -c value Location of a TLS certificate .pem file for the Mosquitto message broker [$MQTT_CERT]
--key value, -k value Location of a TLS private key .pem file for the Mosquitto message broker [$MQTT_KEY]
--ca value, -a value Location of a CA certificate .pem file used to verify the Mosquitto message broker [$MQTT_CA]
--insecure-skip-verify Skip verification of the Mosquitto message broker TLS certificate (insecure) [$MQTT_INSECURE_SKIP_VERIFY]
--client-id value, -i value Client id to be used to connect to the Mosquitto message broker [$MQTT_CLIENT_ID]
--reset-metrics, -r Reset metrics when loosing connection to broker (default: true) [$RESET_METRICS]
--help, -h show help
--version, -v print the version
Grafana dashboard:
grafana/mosquitto-dashboard.json is a ready-made dashboard for these metrics.
To use it, open Dashboards → New → Import in Grafana and upload the file, or put it in a provisioning folder. It picks the Prometheus data source from a dashboard variable, so it needs no editing.
The exporter's metrics have no labels of their own, so a broker is picked by its scrape labels:
- Job and Instance select the scrape target. When several brokers are selected, the panels show their total.
- Filters takes any other label your Prometheus adds, for example
kubernetes_namespace = mosquitto.
The rows are:
- Overview: exporter state, uptime, clients, subscriptions, message rates, retained messages and dropped publishes.
- Clients and subscriptions: connected and persistent sessions, new connections per second (shows reconnect loops), subscriptions and expired sessions.
- Throughput: messages, bytes and PUBLISH traffic with sent above zero and received below, dropped messages, average payload size and payload share of traffic.
- Broker load averages: the broker's own 1, 5 and 15 minute averages.
- Message store and memory: heap, stored and retained messages, and the outgoing packet queue.
- Exporter process: the exporter's own memory, CPU and scrape health (collapsed).
With --reset-metrics (the default) the exporter sets every metric to 0 when it loses the broker.
The dashboard marks that period with an annotation. Expect a one-off spike in the rate panels when
the connection comes back, because the counters jump from 0 to their old totals.
Documentation
¶
There is no documentation for this package.