Given how machines can take time (or fail to) transition to a powerstate state once cluster deployment spec.powerstate is changed, we would like to add a metric that logs the transition time when the powerstate does not match the expected powerstate (i.e. Resuming, Stopping). This metric can be cleared once the expected powerstate is achieved.

Open questions:

• what should be considered the start time for the calculation of this metric?

• would the metric be logged if condition reflects failed state?