prefer-no-schedule taint sticks forever if oldMS --> newMS --> oldMS #989
Description
How to categorize this issue?
/area robustness
/kind bug
/priority 3
What happened:
a machineSet keeps the taint prefer-no-schedule forever
What you expected to happen:
the taint is removed
How to reproduce it (as minimally and precisely as possible):
have a machienset with rolling update strategy. have a worker definition like so
- cri:
name: containerd
name: standard-16
machine:
type: n2-standard-16
image:
name: ubuntu
version: 22.0.4
architecture: amd64
maximum: 10
minimum: 1
maxSurge: 1
maxUnavailable: 0
volume:
type: pd-standard
size: 800Gi
zones:
- europe-west1-b
systemComponents:
allow: true
change the worker image
- cri:
name: containerd
name: standard-16
machine:
type: n2-standard-16
image:
name: flatcar
version: 9999.9.9
architecture: amd64
maximum: 10
minimum: 1
maxSurge: 1
maxUnavailable: 0
volume:
type: pd-standard
size: 800Gi
zones:
- europe-west1-b
systemComponents:
allow: true
Wait for the new machineset to be created.
Change the worker definition back to what it was before:
- cri:
name: containerd
name: standard-16
machine:
type: n2-standard-16
image:
name: ubuntu
version: 22.0.4
architecture: amd64
maximum: 10
minimum: 1
maxSurge: 1
maxUnavailable: 0
volume:
type: pd-standard
size: 800Gi
zones:
- europe-west1-b
systemComponents:
allow: true
The "new" machineset will be deleted, the old one will be kept, but tainted, resulting in the machines to be tainted.
Environment:
:
- Others:
machine-controller-manager:v0.53.0