As of today cluster-api-provider references very old versions of k8s packages and machine-api packages, leading to bugs like https://bugzilla.redhat.com/show_bug.cgi?id=1831780. Moreover, the latest machine-api does health checking now [1] which would mean a change like [2] is needed here too, but this cannot be done unless the k8s packages are updated which in turn means the machine-api packages are updated too. Without this update, the machine controller keeps restarting as it is shot down by the machine-api-operator because of the inability to report readiness and liveness.
As of today cluster-api-provider references very old versions of k8s packages and machine-api packages, leading to bugs like https://bugzilla.redhat.com/show_bug.cgi?id=1831780. Moreover, the latest machine-api does health checking now [1] which would mean a change like [2] is needed here too, but this cannot be done unless the k8s packages are updated which in turn means the machine-api packages are updated too. Without this update, the machine controller keeps restarting as it is shot down by the machine-api-operator because of the inability to report readiness and liveness.
[1] https://github.com/openshift/machine-api-operator/pull/602 [2] https://github.com/openshift/cluster-api-provider-azure/pull/139