Getting --url option is mandatory error for ping and upload commands

[mackaybeth]

Describe the bug

I'm trying to upload in a GitHub Action with jfrog rt upload --spec build/upload_spec.json, and I get the --url option is mandatory error. I tried configuring the JFrog CLI with environment variables (JF_URL and JF_TOKEN), and also with the Config Token exported from my local setup (where the command works).

I tried even just running jfrog rt ping, and that also gives the same error.

If I add --url on on the command line, the ping command works, but the upload does not (get 401 Unauthorized).

Current behavior

This is the output from the jfrog/setup-jfrog-cli@v3 step

Run jfrog/setup-jfrog-cli@v3
  with:
    version: [2](https://github.com/REDACTED/actions/runs/4582980103/jobs/8093546472#step:12:2).[3](https://github.com/REDACTED/actions/runs/4582980103/jobs/8093546472#step:12:3)5.0
  env:
    BRANCH_NAME: 2/merge
    CI: true
    CHANGE_ID: 2
    BUILD_NUMBER: 18
    CHANGE_TARGET: master
    GITHUB_TOKEN: ***
    JFROG_CLI_LOG_LEVEL: DEBUG
    AWS_DEFAULT_REGION: us-east-2
    AWS_REGION: us-east-2
    AWS_ACCESS_KEY_ID: ***
    AWS_SECRET_ACCESS_KEY: ***
    AWS_SESSION_TOKEN: ***
    pythonLocation: /opt/actions-runner/_work/_tool/Python/3.7.16/x6[4](https://github.com/REDACTED/actions/runs/4582980103/jobs/8093546472#step:12:4)
    PKG_CONFIG_PATH: /opt/actions-runner/_work/_tool/Python/3.7.16/x64/lib/pkgconfig
    Python_ROOT_DIR: /opt/actions-runner/_work/_tool/Python/3.7.16/x64
    Python2_ROOT_DIR: /opt/actions-runner/_work/_tool/Python/3.7.16/x64
    Python3_ROOT_DIR: /opt/actions-runner/_work/_tool/Python/3.7.16/x64
    LD_LIBRARY_PATH: /opt/actions-runner/_work/_tool/Python/3.7.16/x64/lib
    JF_ENV_1: 
Setup JFrog CLI
  Downloading JFrog CLI from https://releases.jfrog.io/artifactory/jfrog-cli/v2/2.3[5](https://github.com/REDACTED/actions/runs/4582980103/jobs/8093546472#step:12:5).0/jfrog-cli-linux-amd[6](https://github.com/REDACTED/actions/runs/4582980103/jobs/8093546472#step:12:6)4/jfrog

This is the out put from my next step, which runs the JFrog CLI command

Run jfrog rt upload --spec build/upload_spec.json
11:56:56 [Debug] JFrog CLI version: 2.35.0
11:56:56 [Debug] OS/Arch: linux/amd64
11:56:56 [Debug] Locking config file to run config Clear command.
11:56:56 [Debug] Creating lock in: /home/ubuntu/.jfrog/locks/config
11:56:56 [Debug] Releasing lock: /home/ubuntu/.jfrog/locks/config/jfrog-cli.conf.lck.5874.1680350216994187067
11:56:56 [Debug] Config Clear command completed successfully. config file is released.
Error: 6 [Error] the --url option is mandatory
Error: Process completed with exit code 1.

When I run the jfrog rt ping command locally with the JFROG_CLI_LOG_LEVEL="DEBUG" flag set, I see successful HTTP GET commands and 200 responses.

Reproduction steps

I have this in my action:

      - name: Setup Jfrog CLI
        uses: jfrog/setup-jfrog-cli@v3
        env:
          JF_ENV_1: ${{ secrets.JF_SECRET_ENV_1 }}
#          JF_URL: ${{ secrets.ARTIFACTORY_URL }}
#          JF_ACCESS_TOKEN: ${{ secrets.ARTIFACTORY_TOKEN }}
      - name: Upload to Artifactory
        shell: bash
        run: jfrog rt upload --spec build/upload_spec.json

When I run the action, the Upload to Artifactory step fails with the error the --url option is mandatory. If I add the URL, I get 401 Unauthorized errors (which is the same thing that happens when I run the jfrog rt upload command locally with that flag). The same thing happens if I run with the Config Token (as shown above in the snip), or with the commented out JF_URL and JF_ACCESS_TOKEN env vars instead. I do not try to run with all 3 env vars set.

I created the admin-level access token in the Artifactory UI. We run a self-hosted installation of Artifactory, and do not use JFrog Platform.

Expected behavior

I expect to see something like this

{
  "status": "success",
  "totals": {
    "success": 1,
    "failure": 0
  }

And find my artifact in Artifactory.

Setup JFrog CLI version

@v3 (as of today, this is 3.2.0)

JFrog CLI version

I'm using the default, which is 2.35.0 as of today.

Workflow operating system type and version

linux/amd64 self-hosted runner, ubuntu-jammy-22.04

JFrog Artifactory version (if relevant)

self-hosted, 7.41.7

JFrog Xray version (if relevant)

N/A

[sverdlov93]

Hi @mackaybeth , Thanks for reporting this issue. Are you sure that you added the secrets correctly on the secrets section under <github_repo_url>/settings/secrets/actions?

JF_ENV_1: ${{ secrets.JF_SECRET_ENV_1 }} OR

JF_URL: ${{ secrets.ARTIFACTORY_URL }}
JF_ACCESS_TOKEN: ${{ secrets.ARTIFACTORY_TOKEN }}

Can you try running an echo command to print the value of secrets.ARTIFACTORY_URL on runtime? Also, can you add the JFROG_CLI_LOG_LEVEL="DEBUG" env to the Action and provide the full logs? It looks like the JFrog CLI is not getting your Credentials and that's why it asks for url.

[shelling21]

I also experience this issue in my GitHub pipelines. But it doesn't occur always. Usually after a retry it works. If I enable JFROG_CLI_LOG_LEVEL="DEBUG" I get the following output.

Run jf rt ping 13:55:15 [Debug] JFrog CLI version: 2.50.4 13:55:15 [Debug] OS/Arch: windows/amd64 13:55:15 [Debug] Locking config file to run config Clear command. 13:55:15 [Debug] Creating lock in: C:\windows\ServiceProfiles\NetworkService.jfrog\locks\config 13:55:15 [Debug] Releasing lock: C:\windows\ServiceProfiles\NetworkService.jfrog\locks\config\jfrog-cli.conf.lck.1952.1701348915532607900 13:55:15 [Debug] Config Clear command completed successfully. config file is released. Error: 5 [Error] the --url option is mandatory

[yahavi]

@shelling21 It looks like a race condition - the config got cleared before the JFrog CLI read it. Could you please provide an outline of the workflow you used?

[shelling21]

I'm curious where the race condition can take place. We use this in a pipeline that performs a hardware-in-the-loop test on a windows host which take about 3-4 minutes to complete. It looks like:

---
name: Hardware in the loop test

on:
  workflow_call:
    inputs:
      buildNumber:
        description: GitHub build number
        required: true
        type: string
      configuration:
        description: Use a Debug or Release configuration in hil test
        required: true
        type: string
      testTag:
        description: Test tag
        required: true
        type: string

permissions:
  contents: read

env:
  GIT_COMMIT: ${{ github.sha }}
  JFROG_CLI_LOG_LEVEL: DEBUG

jobs:
    steps:
      - name: Checkout repository
        uses: actions/checkout@b4ffde65f46336ab88eb53be808477a3936bae11 # v4.1.1
        with:
          persist-credentials: false
          submodules: false
          token: ${{ secrets.GITHUB_TOKEN }}

      - name: Setup jfrog-cli
        uses: jfrog/setup-jfrog-cli@901bb9632db90821c2d3f076012bdeaf66598555 # v3.4.1
        env:
            JF_ENV_1: ${{ secrets.ARTIFACTORY_TOKEN }}

      - name: Download xxx tool
        run: jf rt download generic-external/xxx.exe "xxx.exe"

      - name: Download embedded builds
        env:
          JFROG_CLI_BUILD_NAME: pkgs-embedded+arm-none-eabi
        run: jf rt download --build pkgs-embedded+arm-none-eabi/${{ inputs.buildNumber }} "*/(*)/${{ inputs.configuration }}/(*)" "install/{1}/${{ inputs.configuration }}/{2}"

      - name: Download tools
        env:
          JFROG_CLI_BUILD_NAME: pkgs-embedded+i686-w64-mingw32
        run: jf rt download --build pkgs-embedded+i686-w64-mingw32/${{ inputs.buildNumber }} "*/(*)/Debug/(*)" "install/tools/{1}/Debug/{2}"

      - name: Program device 1
        run: |
          <Programming commands>

      - name: Program device 2
        run: |
          <Programming commands>

      - name: Reset test setup
        run: |
          <Reset commands>

      - name: Sleep for 3 seconds
        run: Start-Sleep -s 3
        shell: powershell

      - name: Run hardware-in-the-loop tests
        run: >
          <Run hardware-in-the-loop commands>

      - name: Upload test report
        if: success() || failure()
        env:
          DEST_PATH: pkgs-embedded-snapshot/nic/${{ github.head_ref || github.ref_name }}/${{ github.run_number }}/
          JFROG_CLI_BUILD_NAME: pkgs-embedded+smoke-test-output-${{ inputs.testTag }}
        run: |
          jf rt ping
          jf rt build-collect-env
          jf rt upload "integration_test_output/*" "${{ env.DEST_PATH }}"
          jf rt build-publish

(I removed the unrelevant parts) Usually the process fails on the jf commands in the last step.

[shelling21]

@yahavi : I'm really curious what the problem can be and how to resolve it. Currently our CI/CD pipeline is unstable because of this issue and I would like to fix that.

[sverdlov93]

Hi @shelling21 and @mackaybeth, We believe that the issue was fixed on https://github.com/jfrog/setup-jfrog-cli/releases/tag/v3.4.2 so feel free to update to the latest version. Looking forward to your feedback on it.

[shelling21]

Hi @sverdlov93,

I've updated the setup-jfrog-cli action to v3.4.2 but it does not seem to resolve the issue. The same issue happens. The problem seems to be most likely to trigger when there is a longer time between the setup-jfrog-cli action and actually using the jf command in later steps. In our use case we're downloading some binaries with the jf cli, then programming a hw target, run some tests, and then upload the test results to artifactory. Usually the failures are happening in the last upload step. It looks like some token gets expired.

For now I got a workable solution to do a setup-jfrog-cli action before each time I use jf cli. Then when the token seems to get renewed when necessary. If there is already a config with a valid token it passes the action in 0s.

[sverdlov93]

Hi @shelling21 , Are you getting the same debug logs with [Debug] Config Clear command completed successfully. config file is released. that happens before your last command? can you share again the full logs?

[shelling21]

@sverdlov93: I've sent you the logs via e-mail (from your GitHub profile)

[sverdlov93]

Hi @shelling21 , I see from your logs that you are using an old version of jfrog/setup-jfrog-cli. The latest is v3.5.1 and it should use the version v2.52.7 of JFrog CLI.

On your logs it was:

jfrog/setup-jfrog-cli@901bb9632db90821c2d3f076012bdeaf66598555
with:
 version: 2.50.4

[ssplatt]

I'm running into this today with

- uses: jfrog/setup-jfrog-cli@v4
        with:
          version: latest

putting jf config show in the action does show a proper config

[sverdlov93]

@ssplatt , how do you configure the credentials for the task? Can you share the full yaml you use?

here is an example:

- uses: jfrog/setup-jfrog-cli@v4
  env:
    # JFrog platform url (for example: https://acme.jfrog.io)
    JF_URL: ${{ vars.JF_URL }}

    # Basic authentication credentials
    JF_USER: ${{ secrets.JF_USER }}
    JF_PASSWORD: ${{ secrets.JF_PASSWORD }}
    or
    # JFrog Platform access token
    JF_ACCESS_TOKEN: ${{ secrets.JF_ACCESS_TOKEN }}
- run: |
    jf rt ping

[ssplatt]

Jf_url and jf_access_token as env vars

[sverdlov93]

is there a chance that your env URL is empty? can you add another step before the jfrog-cli step that prints ${{ vars.JF_URL }} to the prompt?