maintainers can restart signalboost w/ signal message + password

behavior

GIVEN a running instance of signalboost with a support channel and a password <PASSWORD> (defined in encrypted/version-controlled config files)

• WHEN an admin of the support channel sends RESTART <PASSWORD>
• THEN signalboost will spin all containers down and back up again

implementation

sadly this is not as simple as just calling bin/restart (because that itself invokes docker-compose down && docker-compose up which cannot be invoked from inside the docker container where signalboost would try to invoke the script. and we can't use the docker api from inside a container to shut down and restart that container...

instead we could try something like:

• create an orchestrator container. this can just run node (or python) on bare metal alpine
• the orchestrator runs a simple http server that only accepts requests from the app container on the same network
• it has a POST /restart endpoint that, when hit, uses the docker API (see: https://github.com/apocas/dockerode) to invoke the equivalent of docker-compose down && docker-compose --env-file .env up -d
• when dispatcher.commands.execute receives RESTART, it:
  • calls channelRepository.isSysadmin(sender.phoneNumber) (to see if the sender is an admin of the support channel and thus a sysadmin)
  • checks to see if the password matches the configured password
  • proceeds to issue POST /restart to orchestrator (and in so doing, dies and is reborn)

or something simpler if we can come up with it?