Cache node_modules in GitHub Actions

Question

I have a Yarn monorepo (workspace) with 2 packages: backend (Node.js/TypeScript) and frontend (React/Typescript).

/package.json (trimmed)

{
  "workspaces": [
    "backend",
    "frontend"
  ],
}

and I'm trying to add continuous integration with GitHub Actions and trying to use actions/cache@v2 to cache the Yarn cache dir and all of the projects' node_modules dirs

.github/workflows/CI.yml (trimmed)

    steps:
      - uses: actions/checkout@v2

      - name: Get yarn cache directory path
        id: yarn-cache-dir-path
        run: |
          echo "::set-output name=dir::$(yarn cache dir)"
          echo "::set-output name=version::$(yarn -v)"

      - name: Use Node.js ${{ matrix.node-version }}
        uses: actions/setup-node@v2
        with:
          node-version: ${{ matrix.node-version }}

      - uses: actions/cache@v2
        with:
          path: |
            ${{ steps.yarn-cache-dir-path.outputs.dir }}
            '**/node_modules'
            '**/.eslintcache'
          key: ${{ runner.os }}-yarn-${{ steps.yarn-cache-dir-path.outputs.version }}-${{ hashFiles('**/yarn.lock') }}

      - name: Install packages
        run: yarn install --frozen-lockfile

I receive that the cache is stored and re-used for the consecutive runs:

key: Linux-yarn-1.22.10-143fef95c7228810cf502305eff3be1cbc468dc8a3e0b153a4311c0250aaef6f
Received 158645465 of 175422681 (90.4%), 151.3 MBs/sec
Received 175422681 of 175422681 (100.0%), 138.1 MBs/sec
Cache Size: ~167 MB (175422681 B)
/usr/bin/tar --use-compress-program zstd -d -xf /home/runner/work/_temp/08363700-9a23-447e-a80e-6f3dbec6068f/cache.tzst -P -C /home/runner/work/path
Cache restored successfully
Cache restored from key: Linux-yarn-1.22.10-143fef95c7228810cf502305eff3be1cbc468dc8a3e0b153a4311c0250aaef6f

but yarn still tries to resolve the dependencies:

yarn install --frozen-lockfile
shell: /usr/bin/bash -e {0}
yarn install v1.22.10
[1/4] Resolving packages...
[2/4] Fetching packages...
info [email protected]: The platform "linux" is incompatible with this module.
info "[email protected]" is an optional dependency and failed compatibility check. Excluding it from installation.
info [email protected]: The platform "linux" is incompatible with this module.
info "[email protected]" is an optional dependency and failed compatibility check. Excluding it from installation.
info [email protected]: The platform "linux" is incompatible with this module.
info "[email protected]" is an optional dependency and failed compatibility check. Excluding it from installation.
[3/4] Linking dependencies...
.....
[4/4] Building fresh packages...
Done in 40.07s.

My expectations are that it should be working as on my local machine:

$ yarn --frozen-lockfile
yarn install v1.22.10
[1/4] 🔍  Resolving packages...
success Already up-to-date.
✨  Done in 0.72s.

Can I somehow improve my configuration to meet the expected result, or this GitHub Actions behaviour expected?

---

Update: When tried with the following paths:

          path: |
            '**/node_modules'
            '**/.eslintcache'

or:

          path: |
            'node_modules'
            '*/node_modules'
            '**/.eslintcache'

the cache size is 22 B. Probably not matching any of the node_modules dirs

Answer 1

After a lot of trial and error, removing the quotes from the paths seems to have fixed the problem. And the size of the cache have increased almost twice

  - uses: actions/cache@v2
    id: yarn-cache
    with:
      path: |
        **/node_modules
        **/.eslintcache
        ${{ steps.yarn-cache-dir-path.outputs.dir }}

      key: ${{ runner.os }}-yarn-${{ hashFiles('**/yarn.lock') }}
      restore-keys: |
        ${{ runner.os }}-yarn-

Run actions/cache@v2

Received 213909504 of 305669200 (70.0%), 203.6 MBs/sec
Received 305669200 of 305669200 (100.0%), 185.6 MBs/sec
Cache Size: ~292 MB (305669200 B)
/usr/bin/tar --use-compress-program zstd -d -xf /home/runner/work/_temp/2e2d2a1d-04d7-44c3-829e-ec4e8faf394b/cache.tzst -P -C /home/runner/work/path
Cache restored successfully
Cache restored from key: Linux-yarn-143fef95c7228810cf502305eff3be1cbc468dc8a3e0b153a4311c0250aaef6f

Run yarn install --frozen-lockfile

yarn install v1.22.10
[1/4] Resolving packages...
success Already up-to-date.
Done in 0.96s.

• permalink to workflow
• master version

Answer 2

For anyone else looking for yarn caching implementation in gha, the following code is working best for me:

      - name: Get yarn cache directory path
        id: yarn-cache-dir-path
        run: echo "::set-output name=dir::$(yarn cache dir)" 

      - uses: actions/cache@v2
        id: yarn-cache
        with:
          path: ${{ steps.yarn-cache-dir-path.outputs.dir }}
          key: ${{ runner.os }}-yarn-${{ hashFiles('**/yarn.lock') }}
          restore-keys: |
            ${{ runner.os }}-yarn-

      - name: Install dependencies
        run: yarn --prefer-offline
        if: steps.yarn-cache.outputs.cache-hit != 'true'

It is best practice to NOT cache the node_modules.

Caching node_modules to determine which packages to install could led to improper cache busting.

Answer 3

You shouldn't be caching node_modules!
setup-node GitHub action, comes with built-in caching on its own. As written in docs, just add cache property to GitHub Action, where you define your package manager npm | yarn | pnpm

steps:
- uses: actions/checkout@v3
- uses: actions/setup-node@v3
  with:
    node-version: 16
    cache: 'npm' # or 'yarn' | 'pnpm'
- run: npm ci
- run: npm test