Autohealing is also part of the managed instance group settings but merits its own section in order to discuss it further. GCPs managed instance groups are responsible for validating whether each VM instance in our group is running and ready to accept client requests. To perform this validation, it needs a health check, which is basically a probe that contacts each member of the instance group to check their current health. The policy can be based on certain protocols, namely HTTP(S), TCP, or SSL.
We also need to configure the criteria to inform the health check how often to check the instance, the acceptable amount of time that it can get a no response, and the number of consecutive failures to its probe. These settings define when a VM would be classified as unhealthy. If that specific instance is no longer healthy, the autoscaler will add a new instance:
Within...
