Investigate: terraformer-only workflow for 0-sized nodepools
Autoscaled nodepools allow min: 0, but the manager schedules the full pipeline (terraformer → ansibler → kube-eleven → kuber) for any nodepool diff regardless of node count — NodePoolsDiff reports an added nodepool even with zero nodes, so e.g. adding an autoscaled nodepool that starts empty runs whole-cluster ansible playbooks and kubeone apply for nothing.
Investigate whether operations whose diff only touches 0-sized nodepools (new autoscaled nodepool at zero, rolling update of a scaled-to-zero nodepool) can schedule just the terraformer stage, and which flows must keep the full pipeline.
Outcome: a written summary of affected flows plus a follow-up implementation issue if feasible.
Investigate: terraformer-only workflow for 0-sized nodepools
Autoscaled nodepools allow
min: 0, but the manager schedules the full pipeline (terraformer → ansibler → kube-eleven → kuber) for any nodepool diff regardless of node count —NodePoolsDiffreports an added nodepool even with zero nodes, so e.g. adding an autoscaled nodepool that starts empty runs whole-cluster ansible playbooks and kubeone apply for nothing.Investigate whether operations whose diff only touches 0-sized nodepools (new autoscaled nodepool at zero, rolling update of a scaled-to-zero nodepool) can schedule just the terraformer stage, and which flows must keep the full pipeline.
Outcome: a written summary of affected flows plus a follow-up implementation issue if feasible.