Skip to content

README.md

Latest commit

 

History

History
70 lines (54 loc) · 2.23 KB

README.md

File metadata and controls

70 lines (54 loc) · 2.23 KB
Raw

Compute Options Example

This example demonstrates how to configure CPU, memory, and instance types for different provider types to match your workload requirements.

Note: This example creates a VPC for self-containment and testing purposes. Creating a VPC is not required—providers can use the default VPC or an existing VPC.

What This Example Shows

  • How to configure CPU and memory for Fargate providers
  • How to configure compute types for CodeBuild providers
  • How to configure memory (which determines CPU) for Lambda providers
  • How to configure instance types for EC2 providers
  • How to configure instance types and task CPU/memory for ECS providers

Provider-Specific Options

Fargate

  • CPU: 256 (.25 vCPU), 512 (.5 vCPU), 1024 (1 vCPU), 2048 (2 vCPU), 4096 (4 vCPU)
  • Memory: Must match valid combinations with CPU (e.g., 2 vCPU supports 4-16 GB)
  • CPU and memory must be valid Fargate combinations

CodeBuild

  • Compute Types:
    • SMALL: 2 vCPU, 3 GB RAM
    • MEDIUM: 4 vCPU, 7 GB RAM
    • LARGE: 8 vCPU, 15 GB RAM
    • X2_LARGE: 72 vCPU, 145 GB RAM

Lambda

  • Memory: 128 MB to 10 GB (determines CPU proportionally)
  • Ephemeral Storage: Up to 10 GB for /tmp directory
  • More memory = more CPU power automatically

EC2

  • Instance Types: Any EC2 instance type (e.g., m6i.large, c6i.xlarge)
  • Choose based on CPU, memory, network, and storage needs

ECS

  • Instance Type: For cluster instances (e.g., m6i.large)
  • Task CPU: 1024 units = 1 vCPU (fractions supported)
  • Task Memory: In MiB (e.g., 2048 = 2 GB)

Usage

After deploying, use the appropriate provider label in your workflows based on your compute needs:

name: Build
on: [push]
jobs:
  build-small:
    runs-on: [self-hosted, lambda]
    steps:
      - uses: actions/checkout@v5
      - name: Build
        run: npm run build
  
  build-large:
    runs-on: [self-hosted, codebuild]
    steps:
      - uses: actions/checkout@v5
      - name: Build
        run: npm run build

Setup

  1. Deploy the stack: cdk deploy
  2. Follow the setup instructions in the main README.md to configure GitHub integration
  3. Use the appropriate provider label in your workflows based on your compute requirements