Today, when an update run is completed, if any clusters have been skipped by either the user electing to skip them, or failure threshold configuration means the update run has continued past the cluster, there is no way to apply the exact same configuration (specifically node image, but applies to entire update run) to those skipped or failed clusters.
A retry mechanism is proposed that would provide a method to address these clusters in the same context as the original update run, allowing for a consistent state to be applied to all clusters.
Today, when an update run is completed, if any clusters have been skipped by either the user electing to skip them, or failure threshold configuration means the update run has continued past the cluster, there is no way to apply the exact same configuration (specifically node image, but applies to entire update run) to those skipped or failed clusters.
A retry mechanism is proposed that would provide a method to address these clusters in the same context as the original update run, allowing for a consistent state to be applied to all clusters.