README: Kubernetes Autoscaling on Utho-Managed Cluster

Overview

This repository contains comprehensive documentation and guides for enabling Kubernetes autoscaling on a Utho-managed cluster. The Cluster Autoscaler automatically adjusts the number of nodes in your cluster based on resource demand, ensuring optimal performance and cost efficiency.

Contents

This package includes the following files:

📄 Main Documentation

• kubernetes-autoscaling-utho.md - Complete step-by-step guide for setting up Kubernetes autoscaling on Utho clusters

📋 Quick Reference

This README file provides quick access information and troubleshooting tips.

Quick Start

Prerequisites

• Kubernetes cluster v1.20+ on Utho Cloud
• kubectl command-line tool installed and configured
• Helm package manager (optional but recommended)
• Administrator access to Utho Cloud account
• API token from Utho Cloud console

Installation Summary

# 1. Verify nodes
kubectl get nodes

# 2. Deploy Cloud Controller Manager
kubectl apply -f https://raw.githubusercontent.com/uthoplatforms/utho-cloud-controller-manager/main/docs/releases/latest.yml

# 3. Configure and apply autoscaler secret
curl -LO https://raw.githubusercontent.com/uthoplatforms/autoscaler/utho-autoscaler/cluster-autoscaler/cloudprovider/utho/examples/cluster-autoscaler-secret.yaml
nano cluster-autoscaler-secret.yaml  # Add your cluster ID and API token
kubectl apply -f cluster-autoscaler-secret.yaml

# 4. Deploy Cluster Autoscaler
kubectl apply -f https://raw.githubusercontent.com/uthoplatforms/autoscaler/utho-autoscaler/cluster-autoscaler/cloudprovider/utho/examples/cluster-autoscaler-deployment.yaml

# 5. Verify autoscaler is running
kubectl -n kube-system get pods | grep autoscaler

Detailed Steps

For complete, step-by-step instructions with detailed explanations, please refer to the main documentation file: kubernetes-autoscaling-utho.md

Getting Your API Token

1. Log in to your Utho Cloud account
2. Navigate to Account Settings → API Tokens
3. Click Create New Token
4. Name your token (e.g., “Kubernetes Autoscaler”)
5. Select appropriate permissions for cluster management
6. Copy the token immediately (you won’t be able to see it again)
7. Use this token in the Cluster Autoscaler secret configuration

Finding Your Cluster ID

Your cluster ID is typically your cluster name. To find it:

1. Log in to Utho Cloud console
2. Navigate to Kubernetes → Clusters
3. Select your cluster
4. Copy the cluster ID/name from the cluster details page

Verification Commands

Check Cluster Status

kubectl get nodes -o wide

Verify Cloud Controller Manager

kubectl -n kube-system get pods | grep cloud-controller

Monitor Cluster Autoscaler

kubectl -n kube-system get pods | grep autoscaler
kubectl -n kube-system logs -f deployment/cluster-autoscaler

Check Node Scaling

kubectl get nodes -w

Monitor Pending Pods

kubectl get pods -w --all-namespaces

Testing Autoscaling

To verify autoscaling is working:

# Deploy stress test workload
kubectl apply -f https://raw.githubusercontent.com/uthoplatforms/autoscaler/utho-autoscaler/cluster-autoscaler/cloudprovider/utho/examples/stress-test.yaml

# Watch pods (should see some in Pending state)
kubectl get pods -w

# In another terminal, watch nodes scale
kubectl get nodes -w

# After scaling completes, verify pods are running
kubectl get pods

Troubleshooting

Issue: Pods remain in Pending state

Solution:

• Check autoscaler logs: kubectl -n kube-system logs -f deployment/cluster-autoscaler
• Verify secret is applied correctly: kubectl -n kube-system get secrets | grep autoscaler
• Ensure API token has sufficient permissions in Utho Cloud

Issue: Autoscaler pod is not running

Solution:

• Verify the deployment was applied: kubectl -n kube-system get deployment cluster-autoscaler
• Check pod events: kubectl -n kube-system describe pod <pod-name>
• Review deployment logs: kubectl -n kube-system logs deployment/cluster-autoscaler

Issue: Nodes not scaling up

Solution:

• Verify Cloud Controller Manager is running
• Check autoscaler logs for error messages
• Verify API token and cluster ID are correct in the secret
• Ensure your Utho account has sufficient resources/quota available

Issue: Nodes scaling too aggressively

Solution:

• Adjust autoscaler configuration parameters in the deployment
• Set resource requests and limits on your pods for better scheduling decisions
• Review the autoscaler documentation for tuning options

Important Notes

⚠️ Security:

• Never commit API tokens to version control
• Store tokens securely and rotate them regularly
• Limit token permissions to minimum required scope

⚠️ Cost Management:

• Monitor your cluster scaling to understand cost implications
• Set resource requests/limits on pods for better bin packing
• Consider using node auto-repair features

⚠️ Best Practices:

• Always specify resource requests and limits for your pods
• Test autoscaling behavior before production deployment
• Monitor cluster autoscaler logs regularly
• Keep your cluster updated with latest Utho releases

Common Commands Reference

Related Documentation

• Kubernetes Official Documentation
• Kubernetes Autoscaler GitHub
• Utho Cloud Documentation
• kubectl Cheat Sheet

Support & Resources

• Utho Cloud Support: https://support.utho.com
• Kubernetes Community: https://kubernetes.io/community/
• GitHub Issues: Report issues or suggest improvements

Version Information

• Documentation Version: 1.0
• Last Updated: December 31, 2024
• Kubernetes Version: 1.20+
• Utho Autoscaler Version: Latest

Next Steps

1. Follow the complete guide in kubernetes-autoscaling-utho.md
2. Deploy the Utho Cloud Controller Manager
3. Configure and apply the Cluster Autoscaler secret
4. Deploy the Cluster Autoscaler
5. Test with a stress workload
6. Monitor and optimize your scaling configuration

For detailed instructions on each step, please refer to the main documentation file.

Happy Scaling! 🚀