Performance opportunity for yarn 3 cache

Thanks for the new cache feature. Much easier.

After few weeks, I realized that while it supports yarn, there’s some improvements that can be made.

Yarn 3 (probably yarn 2+ too) manages downloaded archives pretty well (.yarn/cache/*.zip) and invalidating on yarn.lock changes does not take that into account and I saw a lot of cache misses.

As an example I converted back to action-cache to illustrate and test.

I’m wondering if a similar approach could be done with setup-node ?

Updated example with action cache

Updated on Nov 23th, taking into account @merceyz comment.

Setup action

      # Get the yarn cache path.
      - name: Get yarn cache directory path
        id: yarn-cache-dir-path
        run: echo "::set-output name=dir::$(yarn config get cacheFolder)"
    
      - name: Restore yarn cache
        uses: actions/cache@v2
        id: yarn-cache # use this to check for `cache-hit` (`steps.yarn-cache.outputs.cache-hit != 'true'`)
        with:
          path: ${{ steps.yarn-cache-dir-path.outputs.dir }}
          key: yarn-cache-folder-${{ hashFiles('**/yarn.lock', '.yarnrc.yml') }}
          restore-keys: |
            yarn-cache-folder-

Testing a cache hit after adding a dependency

With setup node cache, as the yarn.lock have changed all packages would be fetched again (1m 28s) rather than (1s 124ms). Environmentally friendlier 🌳

yarn install --immutable
➤ YN0000: ┌ Fetch step
  ➤ YN0013: │ 1719 packages were already cached, one had to be fetched (superjson@npm:1.7.5)
➤ YN0000: └ Completed in 1s 124ms

Cache Size: ~127 MB (133261279 B)
Cache saved successfully
Cache saved with key: yarn-cache-folder-os-Linux-node--f118ea4bee07eada9df36ad2e83fd6febcbaf06b5b8962689c7650659e872ad3

---

PS: Key points

~Example with action cache~ (old version, before @merceyz improvements)

      - name: Get yarn cache directory path
        id: yarn-cache-dir-path
        run: echo "::set-output name=dir::$(yarn config get cacheFolder)"
      - name: Restore yarn cache
        uses: actions/cache@v2
        id: yarn-cache 
        with:
          path: ${{ steps.yarn-cache-dir-path.outputs.dir }}       
          key: yarn-cache-folder-os-${{ runner.os }}-node-${{ env.node-version }}-${{ hashFiles('**/yarn.lock', '.yarnrc.yml') }}      
          restore-keys: |
            yarn-cache-folder-os-${{ runner.os }}-node-${{ env.node-version }}-
            yarn-cache-folder-os-${{ runner.os }}-

[belgattitude]

Just to share, here’s my updated composite action for

1. yarn 3+ with node-modules linker: https://gist.github.com/belgattitude/042f9caf10d029badbde6cf9d43e400a
2. pnpm 7+: https://gist.github.com/belgattitude/838b2eba30c324f1f0033a797bab2e31

[merceyz]

It can be even more efficient by not including the OS and Node version in the cache key as well - https://github.com/actions/setup-node/pull/272#issuecomment-873564091