Do not set the bootstrap_complete flag in all the nodes: do it only in the nodes that had some role assigned.
This is important because, when some nodes are unassigned and we bootstrap the cluster, those nodes will have the bootstrap_complete set anyway.
The problem appears when running the node removal orchestration, as it uses that flag for prioritizing nodes when looking for a replacement (it tries to skip nodes with bootstrap_complete as it assumes that some role is already assigned and running in those machines).
bootstrap_completeflag in all the nodes: do it only in the nodes that had some role assigned. This is important because, when some nodes are unassigned and we bootstrap the cluster, those nodes will have thebootstrap_completeset anyway. The problem appears when running the node removal orchestration, as it uses that flag for prioritizing nodes when looking for a replacement (it tries to skip nodes withbootstrap_completeas it assumes that some role is already assigned and running in those machines).bootstrap_in_progressdespite the result of the orchestration (as we currently do in the node removal orchestration... and we should do in the upddate orchestration too).rebootmngr