refine the cluster-machines-ready machine readiness criteria, take status.nodeRef into account

Thomas Morinrequested to merge
refine-cluster-machines-ready into main

I've observed the following case during a rolling update:

$ k get machines                                                                   
NAME                                     CLUSTER              NODENAME                                     PROVIDERID                                          PHASE         AGE    VERSION
management-cluster-control-plane-fjjrd   management-cluster                                                openstack:///1e5574bd-f26d-4acb-8820-861520a4acac   Provisioned   36m    v1.28.8
management-cluster-control-plane-jrrmc   management-cluster   management-cluster-cp-6731d2e4cd-7xdzw       openstack:///7c0b626a-0d8d-472b-ada0-4bafa754f744   Running       166m   v1.28.8
management-cluster-control-plane-ssn88   management-cluster   management-cluster-cp-6731d2e4cd-vfths       openstack:///66402fb4-b146-458b-bf45-5933f800d012   Running       166m   v1.28.8
management-cluster-control-plane-vffnk   management-cluster   management-cluster-cp-6731d2e4cd-dr5ln       openstack:///4be7fb33-8dac-47e6-b332-0de67cb9d027   Running       46m    v1.28

This illustrates that to confirm that a machine is fully ready, we should also check for nodeRef.name.

Edited by Thomas Morin

Merge request reports