-
Story
-
Resolution: Done
-
Major
-
None
-
None
-
None
-
False
-
False
-
Hypershift Sprint 3, Hypershift Sprint 4, Hypershift Sprint 5, Hypershift Sprint 22, Hypershift Sprint 23
-
0
-
0
-
0
DoD:
Propagate labels and taints from NodePool to Nodes
This could be adhoc or levereage by CAPI https://github.com/kubernetes-sigs/cluster-api/issues/493 https://github.com/kubernetes-sigs/cluster-api/issues/7731
Absolute MVP reqs:
- Enable NodePool API to express intent for a set of labels and taints to be propagated to Nodes.
- NodePools should enforce the label/taint sync on Node creation only
- A set of labels should not be constrained to any opinionated prefix, or any prefix at all.
Next step MVP to include:
- Apply full list of labels on Node creation (same as above)
- If labels list changes in NodePools, sync to all existing Nodes
- If label already exists on a Node, leave it alone and assume the user knows what they're doing
- If label doesn't exist on a Node, apply it
- If there are other labels on a Node not in the NodePool list, leave those alone as well
FWIW The above could be implemented with server side apply on the Nodes of the set of labels.