Current best practice for autoscaling CI runner on AWS (EC2, ECS, ...)



we currently use the CI runner with the Docker+Machine executor on AWS EC2 with docker images from our own GitLab-backed registry. Docker Machine is now unmaintained for some years and I fear that AWS just might introduce breaking changes one day that result in it no longer be usable from one day to the next.

I also found the AWS ECS Fargate instructions, but the fargate executor also seems a bit unmaintained.

What is the current best practice for getting autoscaling CI runners on AWS with self-hosted docker images?