What this PR does / why we need it:
This PR adds changes to avoid any scale-down (scale-down due to underutilization was already stopped earlier using the scale-down-disabled annotation) like removal of longUnregistered nodes or removal of registered but long not Ready nodes.
Which issue(s) this PR fixes:
Fixes #30
Special notes for your reviewer:
Part of fix as per solution proposed here
Logic relies on the Progressing condition of machineDeployment.
Release note:
- Autoscaler would not scale-down if the node-grp(i.e. machineDeployment) is under rolling update.
- Logic of fixing up the node grp size for node grp, whose current size not equals expected size for long time , is also disabled.
What this PR does / why we need it: This PR adds changes to avoid any scale-down (scale-down due to underutilization was already stopped earlier using the
scale-down-disabledannotation) like removal of longUnregistered nodes or removal of registered but long not Ready nodes.Which issue(s) this PR fixes: Fixes #30
Special notes for your reviewer: Part of fix as per solution proposed here
Logic relies on the
Progressingcondition of machineDeployment.Release note: