Hi, there! I was excited to see changes (in an autoscaler community edition nightly release) for different policy cooldown based on scale up events. I recently changed our underlying Terraform module for our GCP MIGs to incorporate sane defaults (for us) of 10 minutes for cooldown and 3 minutes for cooldown_on_scale_up. The module worked as expected and added these to each MIGs policy file.
However, I do not see evidence that there is a different cooldown after a scale_up (also called scale out in the gce-mig plugin’s source code). It would take the 10 minutes for both despite their being different values in the policy file.
I made some changes to the debugging logs and changes to the handler.go, policy.go, and parse.go in a PR for HashiCorp’s autoscaler source code. I’m new to Golang so would welcome discussion on this. And before I keep going in this direction, I wanted to see if others had success using the cooldown_on_scale_up?
Perhaps it is not intended to be parsed from policy files but from the configuration of the autoscaler.nomad horizontal cluster scaling block itself? Regardless, I’d like to be able to set up different cooldown options in policy files; so hope I’m on the right track.