This topic describes instance management in Auto Scaling.

Instance management operations

In Auto Scaling, you can manually add or remove instances of Alibaba Cloud services, switch Elastic Compute Service (ECS) instances to or remove them from the Protected or Standby state in a scaling group, and query instances.

Impacts of instance interruption

Auto Scaling uses instances of other Alibaba Cloud services. You can manage these instances in their respective consoles. For example, you can delete ECS instances that Auto Scaling automatically creates in the ECS console. However, the deletion operation interrupts the instances. If these instances are interrupted by you or due to system failures, Auto Scaling responds to the instance interruption as described in the following table.

Service Interruption Response
ECS An ECS instance in a scaling group is deleted through the ECS console or API. Auto Scaling determines that the ECS instance is unhealthy through a health check and removes it from the scaling group. The internal IP address of the ECS instance is not automatically removed from the IP address whitelist of a Relational Database Service (RDS) instance attached to the scaling group. If the total number of ECS instances in the scaling group falls below the lower limit after the ECS instance is removed, the scaling group automatically creates an ECS instance to guarantee that the number of instances reaches the lower limit.
ECS The permissions for calling the ECS API are revoked from Auto Scaling. Auto Scaling rejects all scaling requests.
Server Load Balancer (SLB) An ECS instance is detached from an SLB instance through the SLB console or API. Auto Scaling does not automatically detect the action but retains the ECS instance in the scaling group. When a scale-in activity is triggered, Auto Scaling releases the ECS instance if the instance meets the removal policy.
SLB An SLB instance is deleted or the health check feature is disabled for an SLB instance through the SLB console or API. Auto Scaling does not add ECS instances to scaling groups to which the SLB instance is attached. Auto Scaling removes ECS instances from the scaling groups if a scaling task triggers a scale-in rule or the ECS instances are recognized as unhealthy through a health check.
SLB An SLB instance becomes unavailable due to reasons such as overdue payment or failure. All scaling activities fail except for scale-in activities that are manually executed.
SLB The permissions for calling the SLB API are revoked from Auto Scaling. Auto Scaling rejects all scaling requests for scaling groups to which SLB instances are attached.
RDS The IP address of an ECS instance is removed from the IP address whitelist of an RDS instance through the RDS console or API. Auto Scaling does not automatically detect the action but retains the ECS instance in the scaling group. When a scale-in activity is triggered, Auto Scaling releases the ECS instance if the instance meets the removal policy.
RDS An RDS instance is deleted through the RDS console or API. Auto Scaling does not add ECS instances to scaling groups to which the RDS instance is attached. Auto Scaling removes ECS instances from the scaling groups if a scaling task triggers a scale-in rule or the ECS instances are recognized as unhealthy through a health check.
RDS An RDS instance becomes unavailable due to reasons such as overdue payment or failure. All scaling activities fail except for scale-in activities that are manually executed.
RDS The permissions for calling the RDS API are revoked from Auto Scaling. Auto Scaling rejects all scaling requests for scaling groups to which RDS instances are attached.