Deployment Notes

There are any number of ways to deploy Errbit, but official support is limited to Heroku.

See specific notes on deployment via:

• heroku
• dokku
• docker
• kubernetes (experimental)

You can use a process manager to deploy Errbit, but Errbit doesn’t maintain support for any specific process manager. But if you use systemd, @nofxx has been kind enough to share:

• systemd config

HTTPS

Errbit can be deployed with HTTPS in a couple different ways.

Nginx or Apache can be used as a reverse-proxy to Errbit’s Puma. In this scenario, Nginx or Apache is configured to serve using HTTPS. The user sends an HTTPS request to Nginx or Apache. Nginx or Apache decrypts the request and passes it on to Errbit using plain HTTP.

Alternatively, Errbit’s Puma can be configured to serve HTTPS directly. Instead of starting Errbit with the command:

bundle exec puma -C config/puma.default.rb

start it with:

bundle exec puma -b "ssl://0.0.0.0:443?key=server.key&cert=server.crt" -C config/puma.default.rb

Where server.key is a path to the TLS private key and server.crt is the path to the TLS certificate.

Health Checks

If deploying with a system that can check if the app is running as expected then there are two endpoints that can be used:

• /health/readiness - suitable for checking if app is ready to receive requests. If response status is 200 and body contains { "ok": true, "details": [etc...] } then the app is ready.
• /health/liveness - suitable for pinging periodically to check if app is still alive. Expected result is { "ok": true }.