Guacamole is a really useful tool,
but can be difficult to setup properly. The deployment process can be greatly
simplified using docker containers, and orchestrated using docker-compose.
This is a sample configuration for docker-compose that puts together all the components needed to deploy guacamole in a containerized environment. I merely fixed it and updated it to the newer official containers as of November 2017.
This is currently version-pinned to 0.9.12-incubating, since the database Dockerfile requires the client repository to be properly tagged to extract the database schema, and at the time of this writing there wasn't a newer tag.
Assuming you already have a working Docker installation and docker-compose,
setup is really easy.
git clone git@github.com:BrowncoatShadow/compose-guacamole.git
cd docker-compose-guacamole
docker-compose up -d
Give it a few seconds to initialize and then you can access guacamole
at http://docker-host:8080/guacamole/. The default username and password are
both guacadmin (go to your user preferences to change it, and then create another user for regular use).
You definitely want to open the .env file and change the example database passwords
before deploying this to anything resembling a production environment.
docker-compose brings together several components to make this work.
guacamole_data- A persistent data volume that stores all the data even if the containers are destroyed.guacamole_db_1- A MySQL container that acts as the database for all of guacamole's data.guacamole_guacd_1- The guacamole server daemon (guacd) container that handles all the remote connections that guacamole makes.guacamole_guacamole_1- The guacamole client web application container that ties everything together and provides the front-end for the user.
Guacamole does not initialize its own database tables. Per the official documentation, the user is expected to create the database manually and use tools from the client container to generate the needed schema files to initialize the database tables.
That is not very portable for a Docker container, which should be able to be
created and destroyed on the fly. This is remedied in this project with a Dockerfile for the
database container that fetches the needed schema files and uses environment
variables to self-initialize.