K8s Spot Termination Handler
Table of contents
The K8s Spot Termination handler watches the AWS metadata service when running on Spot Instances.
When a Spot Instance is due to be terminated, precisely 2 minutes before it's termination a "termination notice" is issued. The K8s Spot Termination Handler watches for this and then gracefully drains the node it is running on before the node is taken away by AWS.
Deploy to Kubernetes
A docker image is available at
These images are currently built on pushes to master. Releases will be tagged as and when releases are made.
Sample Kubernetes manifests are available in the deploy folder.
To deploy in clusters using RBAC, please apply all of the manifests (Daemonset, ClusterRole, ClusterRoleBinding and ServiceAccount) in the deploy folder but uncomment the
serviceAccountName in the Daemonset.
For the K8s Termination Handler to schedule correctly; you will need an identifying label on your spot instances.
We add a label
node-role.kubernetes.io/spot-worker to our spot instances and hence this is the default value in the node selector of the Daemonset.
nodeSelector: "node-role.kubernetes.io/spot-worker": "true"
To achieve this, add the following flag to your Kubelet:
- K8s Spot Rescheduler: Move nodes from on-demand instances to spot instances when space is available.
- Found a bug? Please open an issue.
- Have a feature request. Please open an issue.
- If you want to contribute, please submit a pull request
Please see our Contributing guidelines.
This project is licensed under Apache 2.0 and a copy of the license is available here.