Prometheus Support

[esceeles]

We currently have a forked version of your autoscaler with some additional lines of code to add an endpoint that serves prometheus metrics. This is what we use for monitoring and allows us to receive an alert when a cluster has hit its max number of nodes, so that the configuration can be altered as needed. Is there any interest for us to submit a PR to master to implement this feature?

[bobo]

Hi!
That does sound awesome. But since not everyone uses prometheus, it needs to be configurable in some way. And probably make it easy to add new metric storages.

[esceeles]

Great! Is there any way that you would prefer to set that option? Would an environment variable be ideal? If it's better, I can open a PR with what I have and get feedback there.

[bobo]

@esceeles Il bring it up with the rest of the team on standup on monday and get back to you! :-)

[bobo]

It should be in the config file: https://github.com/spotify/bigtable-autoscaler/blob/master/src/main/resources/bigtable-autoscaler.conf
It would also be nice if any dependencies (if any) needed for prometheus support arent needed if you dont use prometheus. And if it would be easy to add more metric repositories in the future. :-)