A simple setup for your own smart home server (on a Raspberry Pi)

The smart home market is really fragmented. Using different gateways and Apps from different ecosystems can be annoying, expansive and might not work very well together. Instead I suggest using a custom smart home server with some open source software that can replace all your hubs and give you access to one system to control it all.

I am running this on a Raspberry Pi but as these are Docker containers you can run them on pretty much any platform.

I recommend accessing your Pi/server via SSH. If you do make sure to configure it for save usage and change the standard password! It is recommended to use login via SSH keys. I described how to set that up here

Create a folder to hold all your docker data. Then clone this repository and update the .env file. Change the password and IDs and update the path to the folder you just created. Make sure that the data folder belongs to the same user/group ids you are providing in the .env file. If you see any errors errors in the container logs about Permission denied or similar then you have to check if that subfolder in your datafolder belongs to the same user the container is using. If you still run into problems add user: ${PUID}:${PGID} to the container definition to force it to use that user.

Be sure to update the email related settings if you want to use notifications for automatic container updates. If you have done that uncomment the WATCHTOWER_NOTIFICATIONS related variables in the hosting.yml file.

If you want to use Loki/Grafana to see all logs in one place you need to also copy the loki-configuration.yaml file to ${DATADIR}/loki/config/loki-config.yaml. You need to also install the Loki logging driver or remove the logging commands from the compose files. The installation can be done via

docker plugin install grafana/loki-docker-driver:latest --alias loki --grant-all-permissions
// or if you are running this on a Raspberry Pi
docker plugin install grafana/loki-docker-driver:arm-v7 --alias loki --grant-all-permissions

Alternatively if you are going to use promtail to ingest the logs you need to also copy the promtail-config.yaml file to ${DATADIR}/promtail/config/promtail-config.yaml and remove the logging: parts in the compose yaml files.

If you want to use monitoring of your containers and host system via prometheus you should also create a config directory for that and copy the config file from this repository like

mkdir /some/data/dir/prometheus/etc
cp prometheus_config.yaml ${DATADIR}/prometheus/etc/prometheus.yml

If you want to use the Mosquitto MQTT broker make sure to also copy the provided configuration file like

cp mosquitto.conf ${DATADIR}/mosquitto/config

If you want to use Zigbee2MQTT do not forget to copy the config file and update it via

cp zigbee2mqtt_configuration.yaml ${DATADIR}/zigbee2mqtt/data/configuration.yaml

and update the IP of your Zigbee bridge (if you use a networked one) or switch to serial port in the same config file. Also make sure to update the devices: entry in your smarthome.yml file to pass the right serial device to your Docker container in that case.

Then you can start the containers via docker compose.

docker-compose -f hosting.yml up -d
docker-compose -f smarthome.yml up -d
docker-compose -f monitoring.yml up -d

// to see logs (some will only be available via Grafana, see below)
docker-compose -f ...yml logs -f
// to see if containers are running
docker-compose -f ...yml ps

// to stop
docker-compose -f ...yml down

In the smarthome.yml:

In the hosting.yml:

You really only need one way of notifying you about Docker image updates so I would recommend running either Watchtower or Diun, depening on whether you want automatic updates or not.

In the monitoring.yml

You should only use one adblocker (Adguard Home or PiHole) at a time as they use the same ports. If you want to run a second Adguard Home instance somewhere else as a backup and still keep the setting in sync then follow this article. The needed container is already in the hosting.yml file.

If something is not working, check the logs first! Some service logs can only be viewed directly via docker logs containername or docker-compose -f yamlname.yaml logs. The important services are pushing their logs to Loki which collects all of them. You can use Grafana to view them all. I describe this in more detail here. You need to install the Loki logging driver (see installation part above) for this to work or slightly change the compose files by removing the custom logging sections. Loki creates a json file for each services log that is kepts until a restart. Without any restrictions these files can get really big. To avoid this I added a size restriction to the configuration of each service via the max-size argument. More details about this can be found here.

You can find more images and a details in my blog post