Skip to content
Elastic Fruit Runner

Getting Started

This tutorial walks you through installing elastic-fruit-runner on a macOS Apple Silicon machine, configuring it, and running your first GitHub Actions workflow on an ephemeral Tart VM.

By the end, you will have a working self-hosted runner that automatically scales up VMs for incoming jobs and destroys them after completion.

Prerequisites

Section titled “Prerequisites”
  • A Mac with Apple Silicon (M1 or later)
  • Homebrew installed
  • Tart installed (brew install cirruslabs/cli/tart)
  • A GitHub organization or repository where you have admin access
  • A GitHub Personal Access Token (PAT) with Organization > Self-hosted runners: Read and write scope

Step 1: Install elastic-fruit-runner

Section titled “Step 1: Install elastic-fruit-runner”
Terminal window
brew install boring-design/tap/elastic-fruit-runner

Verify the installation:

Terminal window
elastic-fruit-runner --help

Step 2: Create a configuration file

Section titled “Step 2: Create a configuration file”
Terminal window
mkdir -p ~/.elastic-fruit-runner

Create ~/.elastic-fruit-runner/config.yaml with the following content. Replace your-org with your GitHub organization name and ghp_xxx with your PAT:

orgs:
  - org: your-org
    auth:
      pat_token: ghp_xxx
    runner_group: Default
    runner_sets:
      # macOS runners via Tart VMs (Apple Silicon only)
      - name: efr-macos-arm64
        backend: tart
        image: ghcr.io/cirruslabs/macos-tahoe-xcode:26.3
        labels: [self-hosted, macos, arm64]
        max_runners: 2


      # Linux arm64 runners via Docker
      - name: efr-linux-arm64
        backend: docker
        image: ghcr.io/actions-runner-controller/actions-runner-controller/actions-runner-dind:latest
        labels: [self-hosted, linux, arm64]
        max_runners: 4
        platform: linux/arm64


      # Linux amd64 runners via Docker (Rosetta 2 emulation on Apple Silicon)
      - name: efr-linux-amd64
        backend: docker
        image: ghcr.io/actions-runner-controller/actions-runner-controller/actions-runner-dind:latest
        labels: [self-hosted, linux, amd64]
        max_runners: 4
        platform: linux/amd64


idle_timeout: 15m
log_level: info

Step 3: Start the service

Section titled “Step 3: Start the service”

Start elastic-fruit-runner as a background service that auto-starts on login:

Terminal window
brew services start elastic-fruit-runner

Check that it is running:

Terminal window
brew services info elastic-fruit-runner

You should see elastic-fruit-runner listed as started.

Step 4: Verify in the logs

Section titled “Step 4: Verify in the logs”

Watch the logs to confirm the daemon connected to GitHub:

Terminal window
tail -f /opt/homebrew/var/log/elastic-fruit-runner.log

You should see log lines indicating the scale set was registered and the daemon is polling for jobs.

Step 5: Run a workflow

Section titled “Step 5: Run a workflow”

In your GitHub repository, create a workflow file .github/workflows/test-efr.yaml:

name: Test elastic-fruit-runner
on: workflow_dispatch


jobs:
  hello:
    runs-on: efr-macos-arm64
    steps:
      - run: |
          echo "Hello from elastic-fruit-runner!"
          sw_vers
          uname -m

Go to your repository on GitHub, navigate to Actions, select the Test elastic-fruit-runner workflow, and click Run workflow.

Switch back to the log output. You should see:

  1. The daemon detecting the new job
  2. A Tart VM being cloned and started
  3. The runner registering and executing the job
  4. The VM being destroyed after completion

Step 6: Check the workflow result

Section titled “Step 6: Check the workflow result”

Go back to GitHub Actions. The workflow run should show a green checkmark, with output displaying the macOS version and arm64 architecture.

Next steps

Section titled “Next steps”