An EC2 instance in an Auto Scaling group is running but its application is stopped.
Replace it when the service is unhealthy, even if the instance itself is still powered on.
Which setting is most appropriate?
Enable the Auto Scaling group's ELB health check so application health can trigger instance replacement.
Detailed explanation
A running instance can still have an unhealthy operating system or application.
A running instance can still have an unhealthy operating system or application.
DNS cache lifetime does not control Auto Scaling instance replacement.
DNS cache lifetime does not control Auto Scaling instance replacement.
The load balancer's target status can detect an unhealthy application and cause replacement.
The load balancer's target status can detect an unhealthy application and cause replacement.
An S3 timestamp is not an application health check.
An S3 timestamp is not an application health check.
Check official documentation
Compare the following points with the official documentation.
EC2 Auto Scaling health checksとELB連携の公式説明を確認する。
What to confirm
インスタンスヘルスとアプリケーションヘルスを分けて監視できる。
Key points
- Auto Scaling
- ELB health check
- Instance replacement
Notes
- Environment: AWS公式Auto Scaling・ELBドキュメントの確認
- Confirm current AWS service behavior in the official documentation.
Foundation review
Instance versus service health
Power state alone does not show whether the application is serving requests.
Managed remediation
Auto Scaling can replace the instance after the ELB reports an unhealthy target.