The current way of providing a replication token is to run docker exec on an existing container which has been started. This requires hacking the DBDefs.pm file with grep/echo and also means that if the container is stopped and removed, this process has to be run again. From time to time it seems like we have people coming in looking for help because they've done this and didn't realise that they need to run set-token again.

It seems like an easier way to provide this information would be to pass the token as an environment variable from the docker-compose file. This means that a user would have to only edit the docker-compose.yml file (or add it to .env if we want to support this file) to add the token once, and then every time they start up the containers, the correct token will be added. We already use env variables in DBDefs.pm for the database username/password, so this shouldn't be a significant change.