Implementing health checks

Health checks are a surprisingly long topic. After an outage, I often see action items to "add a health check" or "add [specific check] to the health check." Unfortunately unless the health check signals are handled with a circuit breaker, then a failure that would have turned into a slight degradation can turn into a full outage.

I do need to revisit this article to discuss automatic target weights and target group health thresholds, which are some great tools for avoiding the downsides of health checks.