| meta | content | tags | dates | categories | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
|
|
kubernetes load-balancer loadbalancer service annotations ip expose |
|
|
Kubernetes node images refer to the containerized operating system and runtime environment that run on each node within a Kubernetes cluster.
These images encapsulate the essential components necessary for hosting and managing containers, such as the container runtime (e.g. containerd), the kubelet (Kubernetes agent), and other dependencies.
In Kubernetes, a node is a physical or virtual machine inside a cluster, that serves for deploying and running containerized workloads. A Kubernetes cluster consists of a main node (control plane) and multiple worker nodes. These worker nodes are where the actual containers run.
Scaleway Kubernetes node images are customized Ubuntu Linux-based OS images, specifically built for Scaleway's Kubernetes infrastructure.
These images include containerd, a high-performance container runtime, and Scaleway's Kubernetes Kapsule agent, facilitating communication between the Kubernetes control plane and the underlying infrastructure.
The new node images offers advantages in cluster consistency, management ease, and security. Our images are optimized for performance and compatibility with the Kubernetes control plane and integrate seamlessly with the Kubernetes ecosystem.
The main benefits of using the new node images are:
- Accelerated upgrade timelines
- Uninterrupted service continuity with hot upgrades
- Increased upgrade frequency for large clusters
- Faster provisioning of new nodes
- No more need for cluster reboots
- Expanded possibilities for incorporating new features
If your Kubernetes cluster runs version <=1.29 we have implemented a hybrid roll-out strategy:
- Any newly created node is deployed using the new image system.
- Replaced nodes transition to the new image system.
- Any new node pool is deployed using the new node image
- Nodes created through auto-scaling are deployed using the new system
- Legacy nodes maintain the older system for retro compatibility
- Upgraded nodes follow the same approach, retaining the older system
If your cluster is running Kubernetes =>1.30 it will use the new compute image
- Pool upgrades bring compute images to its nodes