Refactor actuators around a context

[vincepri]

/kind feature

Describe the solution you'd like
In the current state of things, the machine and cluster actuator are structured in a way which effectively operates around either a cluster, a machine, or both. Longer term, we should think about refactoring the actuators so they instantiate throw-away objects from a context. During this refactor, a lot of business logic currently present under services/ec2 could potentially move up one level, keeping a clear separation between the ec2 layer and the actuator operations.

CC @detiber @ashish-amarnath @chuckha @randomvariable

[randomvariable]

Do you mean literally a context.Context, or something more typed? Worry with the former is losing some compile time checks. On the other hand, for decent structured logging, particularly when we consider multi-cluster, many-node, it'd be good to pass a context at least for logging.

I generally agree with separating the actual service calls and actuator operations in a refactor now that we've got to a working state.

[vincepri]

I was suggesting of something more typed, I don't think "context.Context" is useful in this context 😛

[detiber]

/priority important-soon

[vincepri]

/close

[k8s-ci-robot]

@vincepri: Closing this issue.

In response to this:

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.