The purpose of this repository is to provide a docker-compose.yml to easily create, run and delete common docker containers on a raspberry (for example influxdb, grafana, rpi-monitor, ...).
The hardware you need depends on your field of application. But I recommend a Raspberry 3 B (4 cores, 1 GB RAM) and at least a 16GB sd-card inserted (we are going to download a lot of docker images).
First you have to install docker on your raspberry. There is a real good documentation from official docker sources, so I won't get much into detail. Follow the steps from here. Please note that you need at least raspbian jessie or stretch.
After installation you can try your docker installation by running:
sudo docker run --rm arm32v7/hello-world
The next step is to enable the docker daemon on boot:
sudo systemctl enable docker
sudo systemctl start docker
Enable the user pi to issue docker commands without sudo
sudo usermod -aG docker pi
# You should now be able to run (without sudo)
docker run --rm arm32v7/hello-world
Unfortunately there are no pre-build docker-compose binaries for the ARM-architecture, so we have to install docker-compose via python-pip:
sudo apt-get update
sudo apt-get -yy python python-pip
sudo pip install docker-compose
Gratz! You are now running a docker featured multi-purpose raspberry server.
TODO: Describe the docker-compose wrapper file and the environment variables
I use RPI-Monitor to monitor service and the raspberry overall health.
Unfortunately you have to install the rpimonitor on the host machine itself, cause when running in a docker environment it would collect stats about the guest system and not the host system itself.
To install rpimonitor you can follow the installation guide.
I've included a services.conf inside the template directory. This will enable rpi-monitor to monitor the availability of your services like influxdb and grafana
sudo cp ./templates/rpimonitor.services.conf /etc/rpimonitor/services.conf
Next you have to activate it inside /etc/rpimonitor/data.conf. Just search for services.conf, uncomment it and adjust the path. Then restart the service:
sudo service rpimonitor restart
To run services via docker-compose you only have to run
# Run influxdb; -d stands for daemon and will cause influx to run in background
docker-compose up -d influxdb
It is easy as that.
To view the log output of any running service
# You can Ctrl+C out of it when finished
docker-compose logs --tail=50 -f influxdb
Portainer is a lightweight management ui for managing your docker containers, volumes and images. Further documentation is available here.
docker-compose up -d portainer
Next navigate to the url http://<your-raspi-name-or-ip>:9000 to get into the portainer management ui.
Influx DB is a database to store, monitor and analyze time series data. Further documentation is available here.
docker-compose up -d influxdb
# If you need the influx client and to test your running service
docker-compose exec influxdb influx
If you need further customization you can modify the configuration file ./conf/influxdb.conf before firing up the service.
To change default credentials of root:root and the pre-created databases (so far only smarthome) you have to modify the appropiate settings in docker-compose.yml.
Grafana is a data visualization and monitoring tool for (mostly) time series data. Further documentation is available here.
docker-compose up -d grafana
You will notice that influxdb is firing up as well. That's because influxdb should be available as a data source to grafana.
In my setup I basically write sensor data to influxdb to monitor and analyze it via grafana.
If you need further customization you can modify the configuration file ./conf/grafana.ini before firing up the service.
Grafana is accessable by browser: http://<your-raspi-name-or-ip>:3000. Default credentails are admin:admin.
TODO: Howto create Influx Data Source
MQTT is the short-form of Message Queue Telemtry Transport and is used to transfer sensor data (telemetry) from a source to potential multiple sinks.
The message queue (aka broker) is the man in the middle and receives messages that represent sensor data or complex objects and pusblishes them to a topic (known as publishing).
Consumer can subscribe to topics and receive data when it is available (known as subscribing).
Example: A temperature sensor publishes its data to the message queue and a subscriber takes this data and forwards it to an influx database to visualize those sensor data.
I use mosquitto as my MQTT service. mosquitto is best-suited for the Internet of Things (IoT) and the requirements of a raspberry pi.
# Have to change permissions for certain folders cause container runs code with another user
sudo mkdir /var/mqtt && sudo chown 777 /var/mqtt
sudo mkdir /var/log/mqtt && sudo chown 777 /var/log/mqtt
docker-compose up -d mqtt
You can install the client tools to test your running container
sudo apt-get mosquitto-clients
mosquitto_sub -h localhost -v -t 'home/#' # Subscribe for new messages on any home topic -> home/*
# Open up another terminal window
mosquitto_pub -h localhost -t 'home/sensor' -m "MQTT is running" # Publish message to Topic: home/sensor
# After publishing the message you should see 'MQTT is running' on the first terminal window
RPI-Monitor keeps an eye on your raspberry itself and monitors the global health status (cpu, mem, temperature) as well as service availability.
This service is only integrated for educational purpose and it doesn't make sense to run it inside a container. It will monitor your container environment rather than your actual raspberry.
So I've decided to install the monitor the old fashioned way by running apt-get install ;-) (see instructions above)