Issue with folder option?

[seadeep42]

Hi,

I have a monorepo setup with the folder at packages/dashboard/build. But it seems to be deploying packages/dashboard for some reason since yesterday. And I haven't updated anything in the past few days

[Reggionick]

Which version are you using?

[seadeep42]

Thank you for the quick response.

reggionick/s3-deploy@v3

[Reggionick]

We haven't made any change on the v3 of the action. Could you double check on your side?

[callum-hyland]

Same issue for me, no code changed in the action for months but randomly it's started pushing the root folder. Could be an internal Github change?

- name: Deploy
        uses: reggionick/s3-deploy@v3
        with:
          folder: ./interface/build

[ajainvivek]

Same issue for me. no code changes pushing the root folder.

[ajainvivek]

@Reggionick Any updates on this issue?

[Reggionick]

No, I can't work on this right now.. But any PR is welcome!

[palashkaria]

This caused our prod dashboard to go down, we have migrated away from using GH actions to amplify for this reason. ~~Would recommend anyone else using this to pin the version that works.~~ Update: looks like it's not even a version issue; The code has had no changes for months, but our builds worked on 31st March, broke on 14th April. That's odd, and I'm unable to figure out what actually happened

[seadeep42]

I'm not sure if this is the cause but in my case, the only change I notice is the Github job runner version change from 2.289.1 to 2.289.2. It's currently at version 2.290.1 but the issue seems to persist

[tanmoyAtb]

Same issue here. We had to move to manual deploys for the time being.

[seadeep42]

Hi,

For anyone else who came across this. I couldn't solve it with the action but found an alternate to fix it. Edited my action yml file to replace this with aws cli.

      - name: Configure AWS Credentials
        uses: aws-actions/configure-aws-credentials@v1
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ secrets.AWS_REGION }}

      - name: Deploy dashboard to bucket
        run: aws s3 sync ./packages/dashboard/build s3://<bucket-name>/<folder-path>/

      - name: Invalidate dashboard cloudfront
        run: aws cloudfront create-invalidation --distribution-id <dist-id> --paths "/<invalidation-path>/*"

[Akuukis]

The root cause

In simplified form, this library calls npx --cwd ./ via exec.exec(command, [], { cwd }), where's npx "walks" upwards to first package.json and uses that folder as cwd instead, partially ignoring the set cwd within exec. For a reason the upstream docs mention setting cwd: s3-deploy './dist/**' --cwd './dist/' ....

See also https://github.com/npm/cli/issues/1975

Solution

don't use npx or don't hardcode npx --cwd ./.

Workaround

None I can think of, use either upstream instead or aws cli directly as in the comment above.