documentation/Running-Mastodon/Docker-Guide.md

Docker

The project now includes a Dockerfile and a docker-compose.yml file (which requires at least docker-compose version 1.10.0).

Setting up

Review the settings in docker-compose.yml. Note that it is not default to store the postgresql database and redis databases in a persistent storage location. If you plan on running your instance in production, you must uncomment the volumes directive in docker-compose.yml.

Then, you need to fill in the .env.production file:

cp .env.production.sample .env.production
nano .env.production

Do NOT change the REDIS_* or DB_* settings when running with the default docker configurations.

You will need to fill in, at least: LOCAL_DOMAIN, LOCAL_HTTPS, PAPERCLIP_SECRET, SECRET_KEY_BASE, OTP_SECRET, and the SMTP_* settings. To generate the PAPERCLIP_SECRET, SECRET_KEY_BASE, and OTP_SECRET, you may use:

rake secret

Building the app

Before running the first time, you need to build the images:

docker-compose build

docker-compose run --rm web rake secret

Do this once for each of those keys, and copy the result into the .env.production file in the appropriate field.

Then you should run the db:migrate command to create the database, or migrate it from an older release:

docker-compose run --rm web rake db:migrate

Then, you will also need to precompile the assets:

docker-compose run --rm web rake assets:precompile

before you can launch the docker image with:

docker-compose up

If you wish to run this as a daemon process instead of monitoring it on console, use instead:

docker-compose up -d

Configuration

Then you may login to your new Mastodon instance by browsing to http://localhost:3000/

Following that, make sure that you read the production guide. You are probably going to want to understand how to configure Nginx to make your Mastodon instance available to the rest of the world.

The container has two volumes, for the assets and for user uploads, and optionally two more, for the postgresql and redis databases.

The default docker-compose.yml maps them to the repository's public/assets and public/system directories, you may wish to put them somewhere else. Likewise, the PostgreSQL and Redis images have data containers that you may wish to map somewhere where you know how to find them and back them up.

Note: The --rm option for docker-compose will remove the container that is created to run a one-off command after it completes. As data is stored in volumes it is not affected by that container clean-up.

Running tasks

Running any of these tasks via docker-compose would look like this:

docker-compose run --rm web rake mastodon:media:clear

Updating

This approach makes updating to the latest version a real breeze.

1. git fetch to download updates from the repository.
2. Now you need to tell git to use those updates. You have probably changed your docker-compose.yml file. Check with git status.

• If the docker-compose.yml file is modified, run git stash to stash your changes.

3. git checkout TAG_NAME to use the tag code. (If you have committed changes, use git merge TAG_NAME instead, though this isn't likely.)
4. Only if you ran git stash, now run git stash pop to redo your changes to docker-compose.yml. Double check the contents of this file.
5. docker-compose build to compile the Docker image out of the changed source files.
6. (optional) docker-compose run --rm web rake db:migrate to perform database migrations. Does nothing if your database is up to date.
7. (optional) docker-compose run --rm web rake assets:precompile to compile new JS and CSS assets.
8. docker-compose up -d to re-create (restart) containers and pick up the changes.