Skip to content
cloud

GitHub Action

Anka VM GitHub Action

v1.3.3 Latest version

Anka VM GitHub Action

cloud

Anka VM GitHub Action

A GitHub Action for starting and running commands within Anka macOS VMs

Installation

Copy and paste the following snippet into your .yml file.

              

- name: Anka VM GitHub Action

uses: veertuinc/anka-vm-github-action@v1.3.3

Learn more about this action in veertuinc/anka-vm-github-action

Choose a version

Using the Anka VM GitHub Action

IMPORTANT: This is not able to run other actions inside of the Anka macOS VMs. We recommend our anka-actions (up/down) instead which will instead run everything inside of the VM itself.

  1. Install the Anka Build Virtualization Software onto a macOS host machine.
  2. Install and ensure you have registered a shared (org level; found under org settings/actions) or project specific self-hosted runner (found under repo settings/actions) with GitHub. These runners need to be running on the host machines you run your Anka Virtualization CLI.

    If you want ephemeral VMs to spin up for your CI/CD job, install the github actions runner inside of the node running the Anka Build Virtualization software. For persistent VMs, install the runner inside of the VM and don't use this action.

  3. Include a .github/workflows/{whatever}.yml in your repo
  4. Make sure to set your mapping key uses: to veertuinc/anka-vm-github-action@vX.X.X
  5. There are a few required key/values you need to include under with:: anka-vm-template-name and commands (see the Inputs section for more information)
name: My Project's CI/CD
on:
  push:
    branches: [ master ]
jobs:
  ios:
    runs-on: [self-hosted, macOS]
    steps:
      - uses: actions/checkout@v2
      - name: build
        id: build
        uses: veertuinc/anka-vm-github-action@v1.3.3
        with:
          anka-vm-template-name: "11.4"
          anka-vm-tag-name: "vanilla+port-forward-22+brew-git"
          anka-run-options: "--env"
          vm-commands: |
            echo \"Starting build process on \$(hostname)\"
            ./build.sh && \
            ./cleanup.sh
          anka-cp-host-paths: |
            ./
            /Users/nathanpierce/cache
          artifact-files: |
            log.txt
            build/binaryfile-v1

Using the above yaml, the Anka GitHub Action will:

  1. Clone your project repo to the github action runner's working directory (on the host) (uses: actions/checkout@v2)
  2. Pull the Template 11.4 and Tag vanilla+port-forward-22+brew-git from the Registry into the host
  3. Prepare an Anka VM using that Template and Tag
  4. Upload the entire host working directory (./) and a /Users/nathanpierce/cache from the user root on the host into the VM using anka cp

    If you don't specify anything, ./ will always be uploaded

  5. Execute the vm-commands inside of the VM (using anka run), ensuring Environment Variables (secrets, or anything else your CI/CD builds/tests need) on the host are passed in with anka-run-options: "--env"
  6. Pull the artifact-files out of the VM using anka cp and then upload an artifact/archive with both ./log.txt and ./build/binaryfile-v1 inside

Transferring Host files into the VM

If you need to move files from the host into the VM, see the below points:

  • Anka Virtualization versions >= 2.3: Starting in Anka 2.3, the automatic mounting of the current directory in the VM was removed. Instead, we now use anka cp to upload the cloned working directory contents in by default, or any of the files or folders you specify in anka-cp-host-paths.

  • Anka Virtualization versions <= 2.2.3: When using the legacy mounting method, build and test time can be significantly impacted by the default host -> guest mount performance.

    • By default, anka run mounts the current working directory on the host into the VM. So, any commands that you run will have access to the cloned working directory files.
    • You can cd out of the mounted directory (/private/var/tmp/ankafs.0) inside of the VM and then do a git clone of your project's repo. This allows you to then move the files you want (using vm-commands) to upload as an artifact back into the mounted directory (/private/var/tmp/ankafs.0) so they are available on the host.

Inputs

These are defined under the with: mapping key inside of your workflow yaml.

anka-vm-template-name (string) (required)

  • Name or UUID of your Anka VM Template

vm-commands (multi-line string or regular string) (required)

  • Commands you wish to run inside of the Anka VM
  • You can use vm-commands: | for multi-line input, OR, you can just use a single line string vm-commands: "echo 123"
  • You need to escape nested/inner double quotes \": vm-commands: "echo \"123\""
  • You need to escape any dollar signs \$ so that it doesn't interpolate from the host side. Unless of course you wish to pass in something from the host into the VM.
  • When interpolating, be sure to use the proper amount of escapes for the desired effect:
    \\\$(echo $HOME)
    \\\$(echo \$HOME)
    \\\$(echo \\\$HOME)
    will result in...
    $(echo /Users/nathanpierce) # HOST level env was interpolated
    $(echo /Users/anka)         # GUEST level env was interpolated
    $(echo $HOME)               # No interpolation

anka-vm-tag-name (string) (optional)

  • Name of Anka VM Template Tag
  • Defaults to latest tag

anka-custom-vm-label (string) (optional)

  • Label for the cloned VM that will execute your code
  • Defaults to github-actions-${GITHUB_REPOSITORY}-${GITHUB_RUN_NUMBER}-${GITHUB_JOB}-${GITHUB_ACTION}
  • Your custom label will have a random number added to the end of it to prevent collisions when two VMs are running on the same node with the same label. You can interpolate the ENV we create with $(eval echo \$$(echo ${GITHUB_ACTION}_vmLabel)) in your host-post-commands to obtain the specific VM name.

host-pre-commands (string) (optional)

  • Commands you wish to run outside on the node (host) BEFORE preparation of and execution inside the VM
  • You need to escape double quotes \"

host-post-commands (string) (optional)

  • Commands you wish to run outside on the node (host) AFTER preparation of and execution inside the VM
  • You need to escape double quotes \"

anka-start-options (string) (optional)

  • Options set for the anka start execution

anka-run-options (string) (optional)

  • Options set for anka run execution

anka-registry-pull-options (string) (optional)

  • Options set for anka registry pull execution

host-command-options (string; js object or JSON) (optional)

lock-file-location (string) (optional)

  • Location where the pull/clone lock file exists
  • Defaults to "/tmp"

artifact-files (multi-line string) (optional)

  • Each file or folder path you wish to upload and include in the final artifact, newline separated
  • Symlinks and directories are not supported. You need to archive your directories in order to include them in the artifact.
  • Anka Virtualization versions <= 2.2.3: Requires the default host <-> guest mounted volume so that the artifact creation/upload code, running on the host, can see the files you specify and are created inside of the VM.

artifact-archive-file-name (string) (optional)

  • Name of the artifact (archive) that contains all of the files specified in the artifact-files input.
  • Defaults to "artifact"

artifacts-directory-on-host (string) (optional)

  • An absolute or relative file path that denotes the root parent directory of the files being uploaded.
  • Defaults to "./"

skip-registry-pull (boolean) (optional)

  • Skip the registry pull; useful if you do not have a registry or it is down but the Template + Tag are already on the node.

anka-cp-disable (boolean) (optional)

  • Disables the use anka cp command to move files from VM to host (a requirement to upload them as artifacts).
  • Only useful for Anka Virtualization versions >= 2.3
  • Defaults to false

anka-cp-host-paths (multi-line string) (optional)

  • Each file you wish to upload from the host into the VM, newline separated.
  • You can use ./ in order to upload the entire current host directory contents into the VM
  • If uploading a symlink to a directory into the VM, the contents of the linked directory will be copied into the VM with the same name as the symlink.

anka-cp-destination-directory (string) (optional)

  • Destination directory on VM you wish to upload the anka-cp-host-paths into.
  • Defaults to "./"
  • We will create the destination directory in the VM if it does not exist

Outputs

These are returned to your workflow.yaml so that subsequent steps can use them.

std

  • The STDOUT and STDERR from the executed commands
  • Includes artifact upload output

Usage:

jobs:
  functional-tests-second-agent:
    runs-on: [self-hosted, macOS]
    needs: prep
    steps:
      - uses: actions/checkout@v2
      - name: sleep
        run: "sleep 20"
      - name: pull test 2
        id: pull-test-2
        uses: veertuinc/anka-vm-github-action@v1.3.2
        with:
          anka-vm-template-name: "11.4"
          anka-vm-tag-name: "base:port-forward-22"
          vm-commands: |
            env
            ls -laht ./
            ls -laht ../
            pwd
            echo \"HERE\" && \
            echo \"THERE HERE WHERE\"

      - name: Check for output
        run: |
          PULL_TEST_STD="${{ steps.pull-test-2.outputs.std }}"
          printf "pull test std ========================\n$PULL_TEST_STD"
          [[ ! -z "$(echo \\"$PULL_TEST_STD\\" | head -n 1)" ]] || exit 50
          [[ ! -z "$(echo \\"$PULL_TEST_STD\\" | grep 'Lock file /tmp/registry-pull-lock-10.15.6 found')" ]] || exit 51
          true

Developing

Prepare your environment

npm install

Testing

There are two types of tests we perform:

  1. Unit tests (testing functions)

    npm run test
  2. Functional testing using a workflow yaml (not in this repo)

Building

npm run package

TO-DO

  • Dynamically generate tests.yml for template version to update without having to manually do it in multiple places.
  • yaml lists: actions/toolkit#184
  • Figure out how to handle agent lost situations (steps just run indefinitely)
  • Support multiple artifacts and files for those artifacts
  • Better tests with mocks so we can avoid so much functional testing