questions about the LKE cluster auto-scaler
Hello,
I have a couple of questions about the LKE cluster auto-scaler:
- I understand that executing
kubectl drain $node
seems to add two taints:
taints:
- effect: NoSchedule
key: node.kubernetes.io/unschedulable
timeAdded: "<datetime as %Y-%m-%dT%H:%M:%SZ>"
- effect: PreferNoSchedule
key: DeletionCandidateOfClusterAutoscaler
value: "<a UNIX timestamp>"
How much time would it take for the node to be removed?
- Is the
scan intervalparameter of the LKE cluster auto-scaler configurable? If yes, how do I go about doing that?
1 Reply
hphillips
Linode Staff
I created a cluster with idle 4 worker nodes and set the Autoscale Pool to a minimum of 3. After 15 minutes, it scaled down to 3 nodes.
The only three user configurable options for the autoscaler in the Linode API are the minimum and maximum node sizes and whether or not autoscaling is enabled.
I will pass your interest in this feature along to the development team.
If you would like more control over autoscaling node pools, you may wish to explore one of following implementations: