Skip to content

azlin github-runner

Deploy self-hosted GitHub Actions runners on azlin VMs for CI/CD pipelines.

Description

Provision and manage self-hosted GitHub Actions runners on azlin VMs. Scale CI/CD capacity, use custom VM sizes (including GPU), and reduce GitHub Actions costs.

Usage

azlin github-runner [COMMAND] [OPTIONS]

Commands

Command Description
create Create runner on VM
list List all runners
delete Remove runner
logs View runner logs

Examples

Create GitHub Runner

# Create runner on new VM
azlin github-runner create \
  --repo owner/repo \
  --name my-runner \
  --vm-size Standard_D4s_v3

Output: 

Creating GitHub Actions runner...
  Repository: owner/repo
  Runner name: my-runner
  VM: azlin-runner-001

Provisioning VM...
✓ VM created: azlin-runner-001

Installing GitHub Runner...
✓ Runner installed and configured

Registering with GitHub...
✓ Runner registered: my-runner

Runner is ready!
  Status: Online
  VM: azlin-runner-001
  URL: https://github.com/owner/repo/settings/actions/runners

Use in workflow:
  runs-on: self-hosted

Create GPU Runner

# ML/AI workflows with GPU
azlin github-runner create \
  --repo owner/ml-project \
  --name gpu-runner \
  --vm-size Standard_NC6 \
  --labels gpu,ml,cuda

List All Runners

azlin github-runner list

Output: 

GitHub Actions Runners:

NAME         VM               STATUS   LABELS           REPO
my-runner    azlin-runner-001 Online   self-hosted      owner/repo
gpu-runner   azlin-runner-002 Online   gpu,ml,cuda      owner/ml-project
ci-runner    azlin-runner-003 Idle     self-hosted,ci   owner/webapp

Scale Runner Fleet

# Create multiple runners for parallel jobs
for i in {1..5}; do
  azlin github-runner create \
    --repo owner/repo \
    --name ci-runner-$i \
    --labels ci,parallel
done

Delete Runner

# Remove runner and VM
azlin github-runner delete my-runner

Common Workflows

CI/CD Pipeline

# .github/workflows/ci.yml
name: CI Pipeline
on: [push]

jobs:
  test:
    runs-on: self-hosted  # Uses azlin runner
    steps:
      - uses: actions/checkout@v3
      - name: Run tests
        run: pytest

GPU-Accelerated Builds

# .github/workflows/ml-training.yml
name: ML Training
on: [push]

jobs:
  train:
    runs-on: [self-hosted, gpu, ml]
    steps:
      - uses: actions/checkout@v3
      - name: Train model
        run: python train.py --gpu

Cost Optimization

# Use spot instances for cost savings
azlin github-runner create \
  --repo owner/repo \
  --name spot-runner \
  --vm-size Standard_D4s_v3 \
  --spot-instance

# Auto-stop idle runners
azlin github-runner auto-stop --idle-minutes 30

Security

Runners are isolated: - Dedicated VM per runner - Secure token registration - Network isolation - Automatic cleanup

Pricing

GitHub-hosted vs. self-hosted comparison:

Runner Type Cost (per minute) Notes
GitHub-hosted $0.008 2-core, 7GB RAM
azlin Standard_D2s_v3 $0.0016 2-core, 8GB RAM (5x cheaper)
azlin Standard_NC6 (GPU) $0.015 1x K80 GPU (not available on GitHub)

See Also