On 4/3/24 around 8:30a pacific, a new version of the Firezone Relays was deployed to production which contained a logic bug that caused them to fail to be used by Clients after a certain period of time.
This logic bug was not caught in automated testing or our staging environment because it only occurred in Relays with a certain access pattern.
We rolled back the afflicted Relays around 5pm pacific time which resolved the issue.
We’ve since added better load testing to catch issues like this in the future.