follow up the discussion in OTA-1292, currently node conditions with Unknown state are simply treated equally to a bad state. there is a place for evaluation whatever this treatment can be improved. as Petr commented, alternative treatment besides conservative, could be optimistic, or a special treatment.

if certain Unknown state with specific message is known for not interfering with update, should we treat it positively?

do states with known messages like "Kubelet stopped posting node status" deserve a better representation, than the current Disk pressure | Not ready ?

as well, as pointed in OTA-1292, memory/pid pressure conditions (currently ignored by upgrade status) are taints, that can interfere with scheduling, yet they are not considered by MCO as node unavailable. why MCO is ignoring them? is it even intentional? should we keep ignoring them as well?