You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently our oadm manage-node --evacuate code is filled with all sorts of controller checks to make sure we properly identify pods during node evacuation. We should ensure all controllers have controllerRef set (as proposed in kubernetes/kubernetes#24946) to be able to have a generic code in the evacuation.
@soltysh the old servers will not have that annotation set, so we will always need to carry the "legacy" code in case not all pods respects that annotation.
First we need to make sure ALL controllers set this annotation accordingly, once that is done we'll have to do version dependent logic. Since, old clusters won't have all the resources you were adding in your PR, either.
Currently our
oadm manage-node --evacuate
code is filled with all sorts of controller checks to make sure we properly identify pods during node evacuation. We should ensure all controllers havecontrollerRef
set (as proposed in kubernetes/kubernetes#24946) to be able to have a generic code in the evacuation.@mfojtik fyi
The text was updated successfully, but these errors were encountered: