CAST AI Kubernetes Agent

A component that connects your Kubernetes cluster to the CAST AI platform to enable Kubernetes automation and cost optimization features.

Getting started

Visit the docs to connect your cluster.

Helm chart

The helm chart for the CAST AI Kubernetes agent is published in the castai/helm-charts repo.

Reading configuration from file

You can pass configuration to agent via YAML file.

Example file

api:
  key: "api key"
  url: "api.cast.ai"

or

api.key: "api key"
api.url: "api.cast.ai"

Shell example

CONFIG_PATH=<PATH_TO_CONFIG> ./castai-agent

Contributing

Run the agent in your IDE

You must provide the these environment variables:

API_KEY=your-castai-api-key
API_URL=api.cast.ai
KUBECONFIG=/path/to/kubeconfig

Then, based on the Kubernetes provider, you need to provide additional environment variables.

AKS

PROVIDER=aks
AKS_LOCATION=your-cluster-location
AKS_SUBSCRIPTION_ID=your-cluster-subscription-id
AKS_NODE_RESOURCE_GROUP=your-cluster-resource-group

EKS

PROVIDER=eks
EKS_ACCOUNT_ID=your-aws-account-id
EKS_REGION=your-cluster-region
EKS_CLUSTER_NAME=your-cluster-name

kOps

PROVIDER=kops

GKE

PROVIDER=gke
GKE_PROJECT_ID=your-gke-project-id
GKE_CLUSTER_NAME=your-cluster-name
GKE_REGION=your-cluster-region
GKE_LOCATION=your-cluster-location

note, when using zonal GKE_REGION and GKE_LOCATION is often the same, i.e. europe-west3-a

If you want to test changes with existing cluster already onboarded to CAST AI console, you can set environment variable

STATIC_CLUSTER_ID=your-cluster-id

Issues

If you encounter "Error: no Auth Provider found for name "gcp"", add a discard import to the main fn:

import (
    _ "k8s.io/client-go/plugin/pkg/client/auth/gcp"
)

Test agent in a containerized/packaged environment

• Prepare a repository to work with your cluster, e.g. Google Artifact Repository if working with a GKE cluster
• Create and push agent image with TAG=path/to/repository/k8s-agent:1.1-your-test-tag make build. This will build the multi-architecture agent docker image and push it to the repository of your choice. TAG in this case should refer to the full image path and will be used for docker push command as-is. Increment the version when building more test versions.
• Clone castai/helm-charts repo;
• Add necessary changes to the chart (if needed)
• Clone values.yaml and tweak these chart values:
  • image.repository: for image build above, set this to path/to/repository/k8s-agent
  • image.tag: for above example, this would be 1.1-your-test-tag
  • clusterVPA.enabled: turn off for testing setups
  • apiURL: override if using non-production CAST AI environment (CAST AI developers only)
  • apiKey: API key to use for the agent
• Deploy the chart:

  helm template castai-agent . -n castai-agent -f values.ignore.yaml | kubectl apply -f -
  

Release procedure (with automatic release notes)

Head to the GitHub new release page, create a new tag at the top, and click Generate Release Notes at the middle-right.

Licence

Apache 2.0 License