작업에 대한 실행기 선택

Overview

Use jobs.<job_id>.runs-on to define the type of machine to run the job on.

• The destination machine can be either a GitHub-hosted runner, larger runner, or a self-hosted runner.
• You can target runners based on the labels assigned to them, or their group membership, or a combination of these.

• You can provide runs-on as:

  • A single string
  • A single variable containing a string
  • An array of strings, variables containing strings, or a combination of both
  • A key: value pair using the group or labels keys

• If you specify an array of strings or variables, your workflow will execute on any runner that matches all of the specified runs-on values. For example, here the job will only run on a self-hosted runner that has the labels linux, x64, and gpu:

  runs-on: [self-hosted, linux, x64, gpu]
  

  For more information, see Choosing self-hosted runners.

• You can mix strings and variables in an array. For example:

  on:
    workflow_dispatch:
      inputs:
        chosen-os:
          required: true
          type: choice
          options:
          - Ubuntu
          - macOS
  
  jobs:
    test:
      runs-on: [self-hosted, "${{ inputs.chosen-os }}"]
      steps:
      - run: echo Hello world!
  

• If you would like to run your workflow on multiple machines, use jobs.<job_id>.strategy.

Choosing GitHub-hosted runners

If you use a GitHub-hosted runner, each job runs in a fresh instance of a runner image specified by runs-on.

The value for runs-on, when you are using a GitHub-hosted runner, is a runner label or the name of a runner group. The labels for the standard GitHub-hosted runners are shown in the following tables.

For more information, see About GitHub-hosted runners.

Standard GitHub-hosted runners for public repositories

For public repositories, jobs using the workflow labels shown in the table below will run on virtual machines with the associated specifications. The use of these runners on public repositories is free and unlimited.

Virtual Machine	Processor (CPU)	Memory (RAM)	Storage (SSD)	Workflow label
Linux	4	16 GB	14 GB	ubuntu-latest, ubuntu-24.04, ubuntu-22.04, ubuntu-20.04
Windows	4	16 GB	14 GB	windows-latest, windows-2025[Public preview], windows-2022, windows-2019
macOS	4	14 GB	14 GB	macos-13
macOS	3 (M1)	7 GB	14 GB	macos-latest, macos-14, macos-15 [Public preview]

Standard GitHub-hosted runners for private repositories

For private repositories, jobs using the workflow labels shown in the table below will run on virtual machines with the associated specifications. These runners use your GitHub account's allotment of free minutes, and are then charged at the per minute rates. For more information, see About billing for GitHub Actions.

Virtual Machine	Processor (CPU)	Memory (RAM)	Storage (SSD)	Workflow label
Linux	2	7 GB	14 GB	ubuntu-latest, ubuntu-24.04, ubuntu-22.04, ubuntu-20.04
Windows	2	7 GB	14 GB	windows-latest, windows-2025[Public preview], windows-2022, windows-2019
macOS	4	14 GB	14 GB	macos-13
macOS	3 (M1)	7 GB	14 GB	macos-latest, macos-14, macos-15 [Public preview]

In addition to the standard GitHub-hosted runners, GitHub offers customers on GitHub Team and GitHub Enterprise Cloud plans a range of managed virtual machines with advanced features - for example, more cores and disk space, GPU-powered machines, and ARM-powered machines. For more information, see About larger runners.

Example: Specifying an operating system

runs-on: ubuntu-latest

For more information, see Using GitHub-hosted runners.

Choosing self-hosted runners

To specify a self-hosted runner for your job, configure runs-on in your workflow file with self-hosted runner labels.

Self-hosted runners may have the self-hosted label. When setting up a self-hosted runner, by default we will include the label self-hosted. You may pass in the --no-default-labels flag to prevent the self-hosted label from being applied. Labels can be used to create targeting options for runners, such as operating system or architecture, we recommend providing an array of labels that begins with self-hosted (this must be listed first) and then includes additional labels as needed. When you specify an array of labels, jobs will be queued on runners that have all the labels that you specify.

Note that Actions Runner Controller does not support multiple labels and does not support the self-hosted label.

Example: Using labels for runner selection

runs-on: [self-hosted, linux]

For more information, see About self-hosted runners and Using self-hosted runners in a workflow.

Choosing runners in a group

You can use runs-on to target runner groups, so that the job will execute on any runner that is a member of that group. For more granular control, you can also combine runner groups with labels.

Runner groups can only have larger runners or self-hosted runners as members.

Example: Using groups to control where jobs are run

In this example, Ubuntu runners have been added to a group called ubuntu-runners. The runs-on key sends the job to any available runner in the ubuntu-runners group:

name: learn-github-actions
on: [push]
jobs:
  check-bats-version:
    runs-on: 
      group: ubuntu-runners
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: '14'
      - run: npm install -g bats
      - run: bats -v

Example: Combining groups and labels

When you combine groups and labels, the runner must meet both requirements to be eligible to run the job.

In this example, a runner group called ubuntu-runners is populated with Ubuntu runners, which have also been assigned the label ubuntu-20.04-16core. The runs-on key combines group and labels so that the job is routed to any available runner within the group that also has a matching label:

name: learn-github-actions
on: [push]
jobs:
  check-bats-version:
    runs-on:
      group: ubuntu-runners
      labels: ubuntu-20.04-16core
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version: '14'
      - run: npm install -g bats
      - run: bats -v