4_nodeclass_nodepool_rebalance

Example: Advanced Node Management with Custom Classes and Pools

This example demonstrates advanced EKS cluster management using custom nodeclasses and nodepools with CloudPilot AI's intelligent rebalancing and optimization features.

By default, optimization is not enabled. You can enable optimization features by modifying the variables in terraform.tfvars and re-applying.

Features

• Custom nodeclass definitions with specific resource allocations
• Advanced nodepool configuration with instance type filtering
• Comprehensive rebalance and optimization settings
• GPU support and specialized instance configurations
• Zone-specific deployments and disruption controls

Advanced Capabilities

• Custom Node Classes: Define node specifications for different workload types
• Instance Filtering: Control CPU, memory, and instance family selection
• Capacity Management: Mix on-demand and spot instances intelligently
• Disruption Controls: Configure node replacement timing and limits
• Architecture Support: Target specific architectures (amd64, arm64)

Prerequisites

• Terraform - Version 1.0 or later
• AWS CLI - Install and configure the AWS CLI with credentials that have EKS cluster management permissions. Required for EKS-related operations such as updating kubeconfig. If you haven't created an EKS cluster yet, see the example setup: eks-ondemand
• Kubectl - For cluster operations and component management
• CloudPilot AI API key - See CloudPilot AI API Key Documentation for setup instructions

Usage

Step 1: Basic Installation

1. Configure your variables:

   cp terraform.tfvars.example terraform.tfvars

   Edit terraform.tfvars with your values:

   • cloudpilot_api_key: Your CloudPilot AI API key
   • cluster_name: Your EKS cluster name
   • region: AWS region where your cluster is located
   • restore_node_number: Node count for cluster restoration

2. Review and customize node configuration in main.tf:

   • nodeclasses: Customize node specifications
   • nodepools: Adjust instance filtering and priorities

3. Apply the configuration:

   terraform init
   terraform plan
   terraform apply

   This installs the CloudPilot AI agent and configures nodeclasses/nodepools without enabling optimization.

Step 2: Enable Optimization (When Ready)

When you're ready to enable optimization features, edit terraform.tfvars:

enable_rebalance = true

Then re-apply:

terraform plan
terraform apply

Key Configuration Options

Nodeclass Configuration

• instance_tags: Custom tags for provisioned nodes
• system_disk_size_gib: Storage configuration
• extra_cpu_allocation_mcore: Additional CPU for burstable workloads
• extra_memory_allocation_mib: Additional memory allocation

Nodepool Configuration

• provision_priority: Priority levels for different pools
• instance_arch: Target architectures (amd64, arm64)
• instance_family: Specific EC2 instance families
• capacity_type: Mix of on-demand and spot instances
• instance_cpu_min/instance_cpu_max: CPU core filtering
• instance_memory_min/instance_memory_max: Memory filtering
• node_disruption_limit: Maximum concurrent node replacements
• node_disruption_delay: Wait time before replacing underutilized nodes

Use Cases

• Multi-workload Clusters: Different node types for different applications
• Cost Optimization: Intelligent spot instance usage
• Performance Tuning: Specific instance types for specific workloads
• High Availability: Zone distribution and disruption controls

See the main.tf file for detailed configuration with comprehensive inline comments.