We are on a migration from Heroku to Render.
Today in Heroku we had a case in which in a production environment of 3 dynos, one of the dynos went into a “continuously crashing” mode. The other 2 dynos are alive and well.
So let’s say something similar happens in Redner and Redner checks for a healthy state against the “Health Check Path”. Is it possible the route check “by change” will hit one of the two live instances, skipping the dead one, and returning the “OK” status?
Or the “Health Check Path” is implemented in such a way that it checks all the instances currently configured?
To give more context, in the incident I’ve mentioned, we were using BetterUpTime, and surprisingly it didn’t report on an incident, probably because “by luck” its monitoring requests hit the 2 live dynos, and missed the crashing one.