Error Installing Wrangler with pnpm about wrangler-action

[1000hz]

@thantos thanks for investigating. It looks like there are two distinct problems here:

• The missing -w flag in the install command causes pnpm to error
• Package manager inference fails when workspaces are used (#198)

I'll keep this issue open to track the first problem.
Installation error stdout being swallowed should be addressed by #171.
I'll also open a new issue to track installing wrangler only when it's not already installed. (#199)

[AdiRishi]

I just ran into this myself. For me I think the special case was that I'm using the workingDirectory property

- name: Deploy apex-gateway
  uses: cloudflare/wrangler-action@v3
  with:
    accountId: ${{ secrets.CF_ACCOUNT_ID || secrets.CLOUDFLARE_ACCOUNT_ID }}
    apiToken: ${{ secrets.CF_API_TOKEN || secrets.CLOUDFLARE_API_TOKEN }}
    workingDirectory: 'workers/apex-gateway'
    command: 'deploy'

My suspicion is that it is looking for pnpm-lock.yaml which doesn't exist when running in the subdirectory.
To be clear, this is running in a monorepo setup (using turborepo).
The exact error message is

Run cloudflare/[email protected]
  with:
    accountId: ***
    apiToken: ***
    workingDirectory: workers/apex-gateway
    command: deploy
    quiet: false
  env:
    TURBO_TOKEN: ***
    TURBO_REMOTE_CACHE_SIGNATURE_KEY: ***
    TURBO_TEAM: team_tiptop
    TURBO_API: ***
    GITHUB_SHA: ***
    CLOUDFLARE_ACCOUNT_ID: ***
    CLOUDFLARE_API_TOKEN: ***
    NODE_ENV: production
    PNPM_HOME: /home/runner/setup-pnpm/node_modules/.bin
📥 Installing Wrangler
  Running command: npm i [email protected]
  npm ERR! code EUNSUPPORTEDPROTOCOL
  Error: Command failed: npm i [email protected]
  npm ERR! code EUNSUPPORTEDPROTOCOL
  npm ERR! Unsupported URL Type "workspace:": workspace:*
  
  npm ERR! A complete log of this run can be found in: /home/runner/.npm/_logs/2023-10-14T09_39_26_137Z-debug-0.log
  
  npm ERR! Unsupported URL Type "workspace:": workspace:*

  npm ERR! A complete log of this run can be found in: /home/runner/.npm/_logs/2023-10-14T09_39_26_137Z-debug-0.log
  Error: 🚨 Action failed

[1000hz]

@AdiRishi yes, please open a new issue for that

[AdiRishi]

@AdiRishi yes, please open a new issue for that

Done - #198

[enfipy]

I was able to work around this issue with a pretty simple change, just by setting packageManager: pnpm near the workingDirectory property:

- uses: pnpm/action-setup@v2
  with:
    version: 8.9.0
- uses: cloudflare/wrangler-action@v3
  with:
    apiToken: ${{ secrets.CLOUDFLARE_API_TOKEN }}
    workingDirectory: ./<PATH_TO_PACKAGE>/
    environment: production
    packageManager: pnpm

The issue indeed is because of getPackageManager function tries to retrieve packageManager from detectPackageManager(workingDirectory), and when workingDirectory is a subdirectory of workspace (e.g. without pnpm-lock.yaml) - it results in an error. So explicitly setting packageManager: pnpm - solves this issue.

P.S: Also, I think it's a good idea to mention this in the docs or troubleshooting, as nowadays many projects use workspaces.

[sam-goodwin]

Update: I was able to get by this with a hack

wranglerVersion: "* -w"

This uses the existing version of wrangler in the root pnpm and injects the -w needed to bypass the PNPM workspace root error.

This workaround isn't working for me.

I have opened a PR with an attempt to fix this: #207

@1000hz let me know if my approach is on the right track.

[1000hz]

It seems like this could be a result of pnpm not being available in the runner environment. Are you including https://github.com/pnpm/action-setup in an earlier step of your workflow before wrangler-action?

[AdiRishi]

Actually I'm now 100% sure that it's a bug because it's searching in the workingDirectory. We can see this in the wrangler-action codebase

function detectPackageManager(
	workingDirectory = ".",
): PackageManagerValue | null {
	if (existsSync(path.join(workingDirectory, "package-lock.json"))) {
		return "npm";
	}
	if (existsSync(path.join(workingDirectory, "yarn.lock"))) {
		return "yarn";
	}
	if (existsSync(path.join(workingDirectory, "pnpm-lock.yaml"))) {
		return "pnpm";
	}
	if (existsSync(path.join(workingDirectory, "bun.lockb"))) {
		return "bun";
	}
	return null;
}

This is likely a seperate issue to the one being reported, should I create a new issue report?

[thantos]

It seems like this could be a result of pnpm not being available in the runner environment. Are you including https://github.com/pnpm/action-setup in an earlier step of your workflow before wrangler-action?

Ran into the same error and I am 100% installing pnpm

      - name: Set up pnpm
        uses: pnpm/action-setup@v2
        with:
          version: 8.6.2

      - name: do a bunch of things with pnpm

      - name: Deploy
        uses: cloudflare/wrangler-action@v3
        with:
          apiToken: ${{ secrets.CF_API_TOKEN }}
          accountId: ${{ secrets.CF_ACCOUNT_ID }}
          command: pages deploy ./packages/portal/out --project-name=${{ vars.CF_PROJECT_NAME }}

[thantos]

Here is the actual error (from running locally since you swallow the error :( )

ERR_PNPM_ADDING_TO_ROOT  Running this command will add the dependency to the workspace root, which might not be what you want - if you really meant it, make it explicit by running this command again with the -w flag (or --workspace-root). If you don't want to see this warning anymore, you may set the ignore-workspace-root-check setting to true.

Going to try with a working directory.

By the way, I already have it installed at the root, would be great if the action didn't need to do it again...

[thantos]

Which of course runs into this: #198

This action doesn't work with pnpm workspaces.

[thantos]

Update: I was able to get by this with a hack

wranglerVersion: "* -w"

This uses the existing version of wrangler in the root pnpm and injects the -w needed to bypass the PNPM workspace root error.

[maxpetretta]

Another solution to this issue here: #198 (comment)