Post action hangs and times-out after 5 minutes

[sides-flow]

Welcome

• Yes, I understand that the GitHub action repository is not the repository of golangci-lint itself.
• Yes, I've searched similar issues on GitHub and didn't find any.
• Yes, I've included all information below (version, config, etc).

Description of the problem

We have a lot of github workflows that use golangci-lint-action as part of our CICD system.
Recently, we encounter a problem in the Post action of golangci-lint-action, which seems to occur randomly. Currently, when we encounter it, we just rerun the failed job and the second attempt works.

Here is the workflow's log

Post action log with the error

Post job cleanup.
/usr/bin/tar --posix -cf cache.tgz --exclude cache.tgz -P -C /home/runner/work/flow/flow --files-from manifest.txt -z
Warning: Failed to save: Cache service responded with 429 during upload chunk.
Saved cache for golangci-lint from paths '/home/runner/.cache/golangci-lint, /home/runner/.cache/go-build, /home/runner/go/pkg, gzip, 1.0' in 302032ms
/home/runner/work/_actions/golangci/golangci-lint-action/v3.4.0/dist/post_run/index.js:433
                        throw new Error(`Cache upload failed because file read failed with ${error.message}`);
                        ^

Error: Cache upload failed because file read failed with EBADF: bad file descriptor, read
    at ReadStream.<anonymous> (/home/runner/work/_actions/golangci/golangci-lint-action/v3.4.0/dist/post_run/index.js:433:31)
    at ReadStream.emit (node:events:527:28)
    at emitErrorNT (node:internal/streams/destroy:157:8)
    at errorOrDestroy (node:internal/streams/destroy:220:7)
    at node:internal/fs/streams:258:9
    at FSReqCallback.wrapper [as oncomplete] (node:fs:660:5)

I searched and found similar reports of the underlying error (Error: Cache upload failed because file read failed with EBADF), but all them were reported on other github actions and were (allegedly) solved in their respective repositories.

Version of golangci-lint

1.51.2

Version of the GitHub Action

3.4.0

Workflow file

  golangci:
    if: ${{github.event_name == 'push' || !contains(github.event.pull_request.labels.*.name, 'WIP')}}
    name: Linter
    needs: [resolve-modules, validate-matrix]
    runs-on: ubuntu-latest
    strategy:
      matrix: ${{ fromJson(needs.resolve-modules.outputs.matrix) }}
    steps:
      - name: Setup Go
        uses: actions/setup-go@v3
        with:
          go-version: '1.18'
      - name: Checkout
        uses: actions/checkout@v3
      - name: Checkout Submodules
        uses: ./.github/actions/checkout-submodules
      - name: golangci-lint
        uses: golangci/[email protected]
        env:
          ACTIONS_ALLOW_UNSECURE_COMMANDS: "true"
        with:
          version: latest
          working-directory: ${{ matrix.workdir }}
          args: --out-${NO_FUTURE}format colored-line-number --timeout 5m --verbose

Go version

1.18

Code example or link to a public repository

Unrelated (AFAIK)

[mtibben]

I'm also seeing golangci-ilnt-action consistently hang e.g. https://github.com/99designs/aws-vault/actions/runs/4432647480/jobs/7776906381#step:4:24, but can't reproduce a hang when using the tool locally. It only seems to happen in github actions

[mcblair]

Still happening today.
Go version: 1.21
Action version: 4.0.0
Linter version: 1.56.1

[rdonkin-attest]

Try adding this to disable all Go-related caching in golangci-lint-action:

          skip-pkg-cache: true
          skip-build-cache: true

This still caches the linter data, which is tiny (a few MB).

I found this cut several minutes off CI execution, including both the setup and post phases (where the 600MB or so Go cache item is loaded and saved).

On a small codebase, the whole CI-only job is under a minute now, and most of that is installing Go.

This load and save of the GitHub Actions cache item is the main difference between local execution (blazing fast, usually 10 sec or less with caching) and Actions execution (can be very slow without the right config).

I think it would be good if this was a recommended config in the main README.md - linting was very slow in Actions without this, sometimes up to 10 min.

Lots of projects have done this., such as https://github.com/cloudquery/cloudquery/pull/15273/files

[gdubicki]

In my case I had to use:

skip-cache: true

...param to work around the issue.