Kubernetes support & Cluster size

This document describes the maximum version skew supported between Keptn and Kubernetes.

Supported Versions

Keptn versions are expressed as x.y.z, where x is the major version, y is the minor version, and z is the patch version, following Semantic Versioning terminology. Please refer to the table below to determine what Keptn version is compatible with your cluster.

Keptn Installations:

• Control plane: Keptn components to run a Keptn and to manage projects, stages, and services, to handle events, and to provide integration points. Install option: keptn install

• Control & Execution plane: Keptn control plane including all Keptn-services for continuous delivery and automated operations. Install option: keptn install --use-case=continuous-delivery

Remarks:

• (1): Requires sufficient resources (e.g., >= 8 vCPUs and 14 GB memory for deploying sockshop in multiple stages) depending on your use-case and workloads.

Notes:

• It is not recommended to use Keptn with a version of Kubernetes that is newer than the version it was tested against, as Keptn does not make any forward-compatibility guarantees.
• If you choose to use Keptn with a version of Kubernetes that it does not support, you are using it at your own risk.
• Installing Keptn on lightweight Kubernetes platforms such as KIND, Minikube, Microk8s, etc… might work (we recommend using K3s), but is not tested nor guaranteed. Please refer to the column called Kubernetes for the basic supported Kubernetes version.

Abbreviations:

• AKS … Azure Kubernetes Service
• EKS … Amazon Elastic Kubernetes Service
• GKE … Google Kubernetes Engine
• K3s … A certified Kubernetes distribution built for IoT & Edge computing: k3s.io
• K3d … A lightweight wrapper that runs k3s (Rancher Lab’s minimal Kubernetes distribution) in Docker: k3d.io

Test Strategy for Kubernetes support:

• With a new Keptn release, Keptn is tested based on the default Kubernetes version of each Cloud Provider: AKS, EKS and GKE available at the release date.

• Internally, a test pipeline with newer Kubernetes versions is verifying the master branch of Keptn. Known-limitations identified by these tests are referenced at the corresponding Keptn release.

Cluster size

The size of the Keptn control plane has been derived automatically and is also reported at the latest release; see Kubernetes Resource Data at: https://github.com/keptn/keptn/releases.

The predefined resource values for the Keptn services are available in the Helm Charts.

As a rule of thumb, Keptn control plane will run with 2 vCPUs, 4 GB of memory and roughly 10 GB of additional disk space (Docker Images + Persistent Volumes). For execution plane services with continuous-delivery support, your Kubernetes cluster requires additional resources. This depends on the number of projects and workloads you deploy, as well as the number of stages in your Shipyard, and whether you are using direct or blue-green deployments.

In particular, as you grow the number of projects and their size, it is recommended to increase:

• The CPU limits of shipyard-controller and resource-service, and
• The RAM limits of resource-service.

As an example, we ran a stress test of a Keptn installation with 50 projects, each project having 5 stages and 50 services. Every second we ran an evaluation sequence, for a total of 5000 sequences. During the execution, we observed that the shipyard-controller required 1 full CPU and roughly 180MB of RAM. Similarly, the resource-service required around 3 CPUs and 240MB of RAM.