Difference between revisions of "Karpenter.sh/do-not-evict: true"
Jump to navigation
Jump to search
| Line 17: | Line 17: | ||
== Related == | == Related == | ||
* <code>[[ttlSecondsUntilExpired]]</code> | * <code>[[ttlSecondsUntilExpired]]</code> | ||
| − | [[controller.node]] Triggering termination for expired node after 168h0m0s .../... | + | [[controller.node]] [[Triggering termination for expired node after]] 168h0m0s .../... |
== See also == | == See also == | ||
Revision as of 19:27, 19 January 2023
Pods without an ownerRef (also called “controllerless” or "naked pods") will be evicted during voluntary node disruption, such as expiration or consolidation. A pod with the annotations: karpenter.sh/do-not-evict: true will cause its node to be opted out from voluntary node disruption workflows.
kubectl annotate pod your-pod-name karpenter.sh/do-not-evict='true' -n yournamespace your-pod-name annotated
kubectl annotate
Eviction
ConsolidationEmptinessExpiration
Related
controller.node Triggering termination for expired node after 168h0m0s .../...
See also
- Kubernetes annotations,
kubectl annotate, eks.amazonaws.com, annotations.kubernetes.io, annotations:, ingress.annotations - Eviction, Eviction API,
kind: Eviction, kubectl cordon - Karpenter,
karpenter.sh, provisioners.karpenter.sh, Karpenter releases, best practices,karpenter.sh/capacity-type, karpenter.sh/discovery,kind: Provisioner, kind: AWSNodeTemplate,kubectl provisioner,TopologyKey, FailedDraining, Evict, DisruptionBlocked, Karpenter logs,controller., ttlSecondsUntilExpired, KEDA, NodePools, Kind: NodePool, Workload Consolidation, Disruption controls
Advertising:
