Warm pool launch configuration

[AayushbajajCAW]

Warm pool launch configuration

I had a suggestion for adding an autoscaling group for a warm pool to this terraform configuration. This would reduce the boot up time of an ASG without significantly increasing cost. I’m open to feedback and questions on my approach.

Here is how I propose an implementation:
Instead of the lambda calling the EC2 API to create a runner, we introduce an ASG and the lambda calls the ASG for capacity.

[npalm]

I think there are two reasons why I have implemented the pool back in the days with Lambda

1. Simply re-use most of the runner scale up logic, avoid more code / more maintenace, as well complexity
2. The create runners a JIT token needs to be created. This is stored with the instance id. In case moving to an ASG, the scaling group creates the instances. Question is how to distribute the JIT token safely to the instance.

Any thoughts on how to solve those challenges?