instance is unhealthy yet doesn't get replaced

Question

I have a load balanced + auto scaled .NET Windows application configured with 2 hosts minimum. But when 1 of the hosts is "unhealthy", it doesn't get replaced automatically. I have to kill it and then it will be replaced.  
  
The load balancer also knows it is unhealthy. So why the disconnect with Elastic Beanstalk?  
  
Any ideas what could be configured wrong here?  
  
And FWIW, I'm pretty sure this used to work.

Accepted Answer

Hello,  
  
So I get that when the ELB healthchecks fails for the instances, the instances are not replaced. By default the ASG healthcheck type is set to EC2 in Beanstalk. So even if the ELB healthcheck fails, the instances wont be replaced.  
  
In order to replace the instances based on ELB healthchecks, we would have to set the HealthCheckType to ELB using the .ebextions config file:

```
Resources:
  AWSEBAutoScalingGroup:
    Type: AWS::AutoScaling::AutoScalingGroup
    Properties:
      # Example syntax of a property that takes in the actual value
      HealthCheckType: ELB
      HealthCheckGracePeriod: 600
```

More info:  
https://docs.aws.amazon.com/elasticbeanstalk/latest/dg/environmentconfig-autoscaling-healthchecktype.html

Answer

Awesome, thank you!

instance is unhealthy yet doesn't get replaced

Relevant content