vcluster-resource-quota-controller

README

Kubernetes Admission Controller for vCluster Resource Quotas

Overview

This repository contains a Kubernetes Admission Controller designed to enforce resource quotas on Pods in a vCluster environment. The controller ensures that resource limits are properly adhered to for Pods that are synced to the main cluster via the syncer and not for the vCluster system itself. Currently, Kubernetes supports only PriorityClass based quotas, but this controller provides label-based quotas, making it flexible for various use cases.

Features

• Label-based Resource Quotas: Controls Pods based on specific labels, enabling resource management for various scenarios.
• Dynamic Configuration: Resource limits and mandatory resource requirements are configurable via Kubernetes ConfigMaps.
• Seamless Integration: Works seamlessly with Kubernetes' admission controller interfaces and resource management systems.

Configuration

The admission controller uses a ConfigMap for dynamic configuration. Below is an example of the ConfigMap used to set the resource limits and mandatory resource requirements:

apiVersion: v1
kind: ConfigMap
metadata:
  name: vcluster-resource-quota-controller-config
  namespace: default
data:
  limitCPU: "500m"
  limitMemory: "500Mi"

ConfigMap Fields

• limitCPU: Maximum CPU limit for a pod.
• limitMemory: Maximum memory limit for a pod.

How It Works

The admission controller intercepts Pod creation requests and validates them against predefined resource limits and requirements. If a Pod does not comply with the specified resource policies, it is rejected, ensuring that resource quotas are enforced consistently across vCluster environments.

Usage

Prerequisites

• Kubernetes cluster
• kubectl configured to interact with your cluster
• Certificate and key for TLS communication

Deploying the Admission Controller

1. Create ConfigMap:

   Apply the ConfigMap configuration file.

   kubectl apply -f webhook-config.yaml
   

2. Deploy the Admission Controller:

   Apply the deployment, service, and MutatingWebhookConfiguration files. Ensure these files are correctly configured for your environment.

   Example deployment file:

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: vcluster-resource-quota-controller
     namespace: default
   spec:
     replicas: 1
     selector:
       matchLabels:
         app: vcluster-resource-quota-controller
     template:
       metadata:
         labels:
           app: vcluster-resource-quota-controller
       spec:
         containers:
         - name: webhook
           image: your-image-repository/webhook:latest
           ports:
           - containerPort: 8443
           volumeMounts:
           - name: webhook-certs
             mountPath: "/etc/webhook/certs"
             readOnly: true
         volumes:
         - name: webhook-certs
           secret:
             secretName: webhook-server-cert
   

3. Create the TLS Secret:

   Ensure you have a valid TLS certificate and key stored in a Kubernetes Secret named webhook-server-cert.

4. Register the Webhook:

   Apply the MutatingWebhookConfiguration to register your webhook with the API server.

   Example configuration:

   apiVersion: admissionregistration.k8s.io/v1
   kind: MutatingWebhookConfiguration
   metadata:
     name: vcluster-resource-quota-controller
   webhooks:
   - name: webhook.vcluster.example.com
     clientConfig:
       service:
         name: vcluster-resource-quota-controller
         namespace: default
         path: "/validate"
       caBundle: <base64-encoded-CA-cert>
     rules:
     - operations: ["CREATE"]
       apiGroups: [""]
       apiVersions: ["v1"]
       resources: ["pods"]
     admissionReviewVersions: ["v1"]
   

Limitations

• This admission controller specifically targets pods with the label vcluster.loft.sh/managed-by syncing to the main cluster via the syncer mechanism of vCluster.
• Policies are enforced based on labels. Ensure that the correct labels are applied to the desired pods.

Extending for Other Use Cases

Although this admission controller is designed for vCluster environments, it can be adapted to enforce label-based resource quotas for other use cases. By changing the label key and value in the configuration, different resource management policies can be applied.

Kubernetes Support

Kubernetes currently supports quotas based on PriorityClass, but this admission controller adds the flexibility of label-based quotas, making it useful for more granular resource management scenarios.

Acknowledgments

This project uses the k8s.io client libraries for interacting with Kubernetes APIs and the admissionv1 library for handling admission controller webhooks.

Contact

For any questions or support, please contact alex@prepare.sh