Make payjo.in more robust
Now that payjoin is in production we should ensure the directory at payjo.in is more robust. Some areas of improvement:
- Redundancy
- Monitoring/alerts
- Auto-renew TLS certificates
- Documentation
A easy win could be to just get prometheus running on the server. Get some alerts on resource usage. If a service was down today would we have anyway of knowing?
A corollary to this would be to make it easy to spin up directories. Both to help normalize running these servers, allow for local development, and reduce single points of failure. If/when payjo.in goes down it'll stop development as of now.
All of this applies to OHTTP servers as well.
Closing this issue as prometheus metrics were introduced in #848, and there are much more in-depth issues/discussions on this topic in #941, #735, #775