Is ShouldDecrementDesiredCapacity in TerminateInstanceInAutoScalingGroup atomic?

[Anton-Cao]

When we run the following code:

import (
	"github.com/aws/aws-sdk-go-v2/service/autoscaling"
        ...
)

...
autoscaler.TerminateInstanceInAutoScalingGroup(ctx,
  &autoscaling.TerminateInstanceInAutoScalingGroupInput{
	  InstanceId:                     &instanceId,
	  ShouldDecrementDesiredCapacity: aws.Bool(true),
})
...

We observed that the ASG terminates a node, then immediately brings up a new node, then immediately terminates a different node (from 04:47:33 to 04:50:46). It seems like it is terminating the first node before updating the desired count. We wanted to confirm whether or not the termination of the node and the update to the desired count happen atomically?