What is Auto Healing?

Auto Healing allows you to select which health check service will be used to determine if an Instance needs to be replaced due to unhealthiness.  If an instance fails the health check selected, it is automatically replaced with a new instance. Auto Healing is configurable as part of the “AUTO HEALING” settings in the Compute tab of the Creation Wizard.

Configuring Auto Healing Health Check type

Elastigroup supports the following Health Check types :
None – No replacement will happen – Unhealthy instances will remain in the group!
ELB – Use amazon’s Elastic Load Balancer (Classic load balancer) health check.
TARGET_GROUP – Use the AWS Application Load Balancer health check.
EC2 – Use the AWS EC2 Status check(system/instance).
HCS – Use Spotinst’s health check. For more information please see: Spotinst HCS service.
ECS – Use ECS Cluster Instance Health check. For more information please see: Elastigroup Health Checks For ECS
K8S_NODE – Use the Kubernetes Status check. Requires integration with a Kubernetes cluster. For more information please see: Elastigroup Health Checks For Kubernetes
NOMAD_NODE – Use the Nomad Status check. Requires integration with a Nomad cluster.
MLB – Use the Multai Load Balancer health check service. You can read more about it here: Introduction to Multai Load Balancer

Opsworks – Use the opsworks layer membership verification. For more information see OpsWorks Auto-healing

Additional Health Check Settings

Health Check Grace Period – Specify the time (in seconds)  to allow an instance to boot and applications to fully start. As soon as an instance becomes Healthy the grace period is over and will not be referenced again. If an instance is unhealthy after the given grace period, it will be replaced with a new instance with a blue-green deployment approach.
Unhealthy Duration – Specify the amount of time (in seconds) you want to keep existing instances that are deemed unhealthy before the instance is terminated and replaced with a new one. Only valid after an instance was healthy at least once.

 Note: If during the configured grace period, a healthy instance becomes unhealthy,  the auto scaler will not terminate the instance. Only after the grace period is over, the unhealthy duration period will begin, and the auto scaler will terminate the unhealthy instance.