p4n4-docs

IoT Stack

The IoT stack (p4n4-iot) is the foundation of every p4n4 deployment. It owns the p4n4-net Docker bridge network that all other stacks attach to.

Services

Service	Container	Image	Port	Role
Mosquitto	`p4n4-mqtt`	`eclipse-mosquitto:2`	1883 (TCP) / 9001 (WebSocket)	MQTT broker
InfluxDB	`p4n4-influxdb`	`influxdb:2`	8086	Time-series database
Node-RED	`p4n4-node-red`	`nodered/node-red:latest`	1880	Flow-based data routing
Grafana	`p4n4-grafana`	`grafana/grafana-oss:latest`	3000	Dashboards

Node-RED and Grafana wait for their dependencies’ healthchecks before starting.

Network

The IoT stack creates p4n4-net:

networks:
  p4n4-net:
    name: p4n4-net
    driver: bridge
    ipam:
      config:
        - subnet: 172.20.0.0/16

All other stacks declare it as external: true.

Topics and storage

Devices publish JSON to sensors/<device-id>/<measurement>, for example sensors/greenhouse-01/temperature with {"value": 23.5, "unit": "C"}.

Topic	Bucket	Measurement	Tags
`sensors/<device-id>/<measurement>`	`INFLUXDB_BUCKET` (`raw_telemetry`)	`sensor_data`	`device`, `sensor`
`inference/<device-id>/result`	`INFLUXDB_BUCKET`	`inference`	`device`, `model` (if in payload)
`sandbox/sensors/…`, `sandbox/inference/…`	`INFLUXDB_SANDBOX_BUCKET` (`sandbox`)	same as above	same as above

Numbers, strings and booleans in the payload become fields. The device comes from the topic, so a device key in the payload is ignored. Messages on sensors/ or inference/ topics with any other shape (for example sensors/temperature) are dropped with a warning in the Node-RED debug sidebar, and failed InfluxDB writes reach a Catch node. The full rules are in §8.6 of the specs.

InfluxDB buckets

Bucket	Retention	Purpose
`raw_telemetry`	30 days	Inbound sensor readings and inference results (primary bucket)
`processed_metrics`	365 days	Downsampled / aggregated data
`ai_events`	Infinite	Edge runner results, AI annotations, anomaly flags
`system_health`	7 days	Stack component health metrics
`sandbox`	30 days	Development and testing

raw_telemetry is created by InfluxDB’s setup; scripts/init-buckets.sh creates the rest. Grafana gets a datasource per bucket.

Node-RED

The editor and Admin API require a login, checked against NODE_RED_USER and NODE_RED_PASSWORD. If NODE_RED_PASSWORD is empty, every login is refused. p4n4 init generates it; projects created before that must add both keys to the IoT .env.

Flows live in config/node-red/flows/flows.json. Compose mounts the flows/ directory rather than the file, because Node-RED saves by renaming a temporary file over flows.json, so deploying from the editor writes straight to the project. Commit the file to keep your changes.

Environment variables

Variable	Default	Description
`TZ`	`UTC`	Container timezone
`INFLUXDB_USERNAME` / `INFLUXDB_PASSWORD`	`admin` / `adminpassword`	InfluxDB admin login
`INFLUXDB_ORG`	`ming`	InfluxDB organisation (shared with other stacks)
`INFLUXDB_TOKEN`	`p4n4-stack-token`	InfluxDB API token (shared with other stacks)
`INFLUXDB_BUCKET`	`raw_telemetry`	Primary bucket
`INFLUXDB_BUCKET_PROCESSED`	`processed_metrics`	Downsampled data
`INFLUXDB_BUCKET_AI_EVENTS`	`ai_events`	AI events
`INFLUXDB_BUCKET_HEALTH`	`system_health`	Stack health
`INFLUXDB_SANDBOX_BUCKET` / `INFLUXDB_SANDBOX_RETENTION`	`sandbox` / `30d`	Sandbox bucket
`GRAFANA_USER` / `GRAFANA_PASSWORD`	`admin` / `adminpassword`	Grafana admin login
`NODE_RED_USER` / `NODE_RED_PASSWORD`	`admin` / (empty)	Node-RED editor login

The defaults are placeholders. p4n4 init generates real values and p4n4 secret rotate replaces them.

Mosquitto authentication

config/mosquitto/mosquitto.conf ships with allow_anonymous true for development. For production, turn it off and use a password file and ACL (templates are in config/mosquitto/passwd.example and acl.example):

allow_anonymous false
password_file /mosquitto/config/passwd
acl_file /mosquitto/config/acl

mosquitto_passwd -c config/mosquitto/passwd <username>

Clients such as the edge runner then need matching MQTT_USER / MQTT_PASSWORD in their .env. See the Security guide.

Make commands

make up / down / restart / status / logs
make start SERVICE=grafana    # start one service and its dependencies
make buckets                  # list InfluxDB buckets
make test-mqtt                # publish test readings
make test-sandbox             # publish test readings to the sandbox bucket
make clean                    # stop and remove data volumes