1. Describe IN DETAIL the feature/behavior/change you would like to see.

Currently when a rolling update is performing cluster validation and an unrelated nodes instancegroup scales up, the rolling update will wait for that node to pass validation. It would be desirable if the rolling update could proceed even when nodes in unrelated instance groups are unhealthy.

The exception is that failures in instance groups of role "master" should probably block the progression of rolling update in other instance groups.

2. Feel free to provide a design supporting your feature request.

The cluster validation API will need to be changed such that failures that are specific to a node are associated with that node. Examples of such failures would be node readiness and unready pods with system-node-critical priority. The "not enough nodes in instance group" failure would need to be associated with the group. Rolling update would then need to ignore validation failures that are in irrelevant groups.