Upgrading your mastodon instance

As mastodon v3.3.0 has been released for over two months now, I figured it was time to upgrade my family instance from v3.2.1 to v3.3.0 (thereby skipping v3.2.2).

I wrote about hosting your own instance, via docker-compose, in a previous post. The past week I’ve been tinkering with mastodon development. Considering what a pain it has been to setup a local development environment, I remain convinced that a containerized setup is the way to go for hosting mastodon. Let’s get started!


Before starting the upgrade, I would suggest going through the documentation on ‘Upgrading to a new release‘ and through the v3.3.0 release notes. Together with the notes here, you should be all set for a successful upgrade.

Database backup

Prior to attempting an upgrade, you should ensure you have a recent backup of the mastodon database. I used pg_dumpall inside the running postgres container (whose docker-compose service is named db) as follows:

docker-compose exec db /bin/bash
pg_dumpall -U mastodon -f mastodon_db_pg_dumpall_output_20210309.sql
tail mastodon_db_pg_dumpall_output_20210309.sql

If the backup was successful, tail should output:

-- PostgreSQL database dump complete
-- PostgreSQL database cluster dump complete

Make sure you store the resulting sql file in a location that is persisted on the host system, so that the file isn’t lost when the container is shutdown and removed. In the container, I moved the backup file to /var/lib/postgresql/data, which is mounted on the host at /home/mastodon/mastodon/postgres. On the host I moved the file to /home/mastodon.

Upgrading to new release

After completing the database backup, the steps necessary to upgrade to a new version are roughly as follows:

  • Update your local copy of mastodon to the latest release
  • Rebuild the container images
  • Apply pre-install migrations of the mastodon database
  • Shutdown containers running old release of mastodon
  • Clear mastodon cache and finish DB migration
  • Start newly built containers
  • Prune old containers

In shell commands, make sure you point your working directory to the root folder of the mastodon repo:

git fetch --tags  # fetch all tags from the remote, leaving the working directory intact
git stash # stash any changes made to the working directory (typically only docker-compose.yml file)
git checkout v3.3.0  # checkout working directory to the v3.3.0 release
git stash pop # restore any changes you made to the working directory (i.e. passwords in docker-compose.yml files, verify via git diff)
docker-compose build  # build the web, streaming and sidekiq services, they all share the same container image. This may take a while.
docker images  # you should see the freshly build image at the top, something like: tootsuite/mastodon   latest              aa82ba57a3f4        About a minute ago   1.98GB
docker-compose run --rm -e SKIP_POST_DEPLOYMENT_MIGRATIONS=true web rails db:migrate  # Run the pre-deployment database migrations
# You can check the return code via echo $?, if the rc is zero than the migration was successful
docker-compose down # shutdown all docker-compose services
docker-compose run --rm web bin/tootctl cache clear # clear cache
docker-compose run --rm web rails db:migrate # Run post-deployment database migrations, again verify that it succeeded via echo $?. 0 means success
docker-compose up -d  # create and start containers, detaching from their stdout
docker system prune -a # remove unused docker data (images, containers, volumes)

Verify that the containers are up and running via docker-compose ps and docker ps. You should see something like:

[mastodon@sarch mastodon]$ docker-compose ps
        Name                      Command                  State                     Ports               
mastodon_db_1          docker-entrypoint.sh postgres    Up (healthy)                                     
mastodon_redis_1       docker-entrypoint.sh redis ...   Up (healthy)                                     
mastodon_sidekiq_1     /tini -- bundle exec sidekiq     Up             3000/tcp, 4000/tcp                
mastodon_streaming_1   /tini -- node ./streaming        Up (healthy)   3000/tcp,>4000/tcp
mastodon_web_1         /tini -- bash -c rm -f /ma ...   Up (healthy)>3000/tcp, 4000/tcp
[mastodon@sarch mastodon]$ docker ps
CONTAINER ID        IMAGE                 COMMAND                  CREATED             STATUS                   PORTS                                NAMES
3f7ea048f4f7        tootsuite/mastodon    "/tini -- bash -c 'r…"   4 minutes ago       Up 4 minutes (healthy)>3000/tcp, 4000/tcp   mastodon_web_1
c673502e6afa        tootsuite/mastodon    "/tini -- node ./str…"   4 minutes ago       Up 4 minutes (healthy)   3000/tcp,>4000/tcp   mastodon_streaming_1
1bba8a9e1934        tootsuite/mastodon    "/tini -- bundle exe…"   4 minutes ago       Up 4 minutes             3000/tcp, 4000/tcp                   mastodon_sidekiq_1
6c623213428d        redis:6.0-alpine      "docker-entrypoint.s…"   7 minutes ago       Up 7 minutes (healthy)                                        mastodon_redis_1
f65c1f822600        postgres:9.6-alpine   "docker-entrypoint.s…"   7 minutes ago       Up 7 minutes (healthy)                                        mastodon_db_1

In case a container fails to start, you can troubleshoot using its logs via docker-compose logs -ft <service-name>.

Finally, point your browser to your mastodon instance and rejoice! It should be listing 3.3.0 in the footer.

Great success!

As for new features in the 3.3.0 release, I hope to be exploring these in the coming weeks. Many thanks to everyone that contributed to the release!

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.