Action is not compatible with latest Github base image

[Tolsto]

This action expects the access token to be in ${process.env.HOME}/.azure. However, the latest Github base image now uses /opt/az-config As a result, this action doesn't work anymore when using it together with the azure/login action.

[Tolsto]

As a workaround you can use

- name: 'Run Az login'
  uses: azure/login@v2
  with:
    client-id: ${{ inputs.client-id }}
    tenant-id: ${{ inputs.tenant-id }}
    subscription-id: ${{ inputs.subscription-id }}
  env:
    AZURE_CONFIG_DIR: "/home/runner/.azure"
  
- name: Run Az CLI
  uses: azure/CLI@v2
  with:
    azcliversion: 2.65.0
    inlineScript: |
      ...
  env:
    AZURE_CONFIG_DIR: "/root/.azure"

[mcalthrop]

@Tolsto, @MoChilia What value should be specified for AZURE_CONFIG_DIR when az is run directly from a bash step?

Context: when using nick-fields/retry, and setting AZURE_CONFIG_DIR: "/root/.azure", an error results:

Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/opt/az/lib/python3.12/site-packages/azure/cli/__main__.py", line 30, in <module>
    az_cli = get_default_cli()
             ^^^^^^^^^^^^^^^^^
  File "/opt/az/lib/python3.12/site-packages/azure/cli/core/__init__.py", line 922, in get_default_cli
    from azure.cli.core.azlogging import AzCliLogging
  File "/opt/az/lib/python3.12/site-packages/azure/cli/core/azlogging.py", line 30, in <module>
    from azure.cli.core.commands.events import EVENT_INVOKER_PRE_CMD_TBL_TRUNCATE
  File "/opt/az/lib/python3.12/site-packages/azure/cli/core/commands/__init__.py", line 27, in <module>
    from azure.cli.core.extension import get_extension
  File "/opt/az/lib/python3.12/site-packages/azure/cli/core/extension/__init__.py", line 18, in <module>
    az_config = CLIConfig(config_dir=GLOBAL_CONFIG_DIR, config_env_var_prefix=ENV_VAR_PREFIX)
                ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/opt/az/lib/python3.12/site-packages/knack/config.py", line 40, in __init__
    ensure_dir(config_dir)
  File "/opt/az/lib/python3.12/site-packages/knack/util.py", line 114, in ensure_dir
    raise e
  File "/opt/az/lib/python3.12/site-packages/knack/util.py", line 111, in ensure_dir
    os.makedirs(d)
  File "<frozen os>", line 225, in makedirs
PermissionError: [Errno 13] Permission denied: '/root/.azure'

The YML looks roughly like this:

    - name: 'Login via Azure CLI'
      uses: azure/[email protected]
      env:
        AZURE_CONFIG_DIR: '/home/runner/.azure'
      with:
        client-id: ${{ some.value }}
        tenant-id: ${{ some.value }}
        subscription-id: ${{ some.value }}
     
    - name: Do something
      uses: nick-fields/[email protected]
      env:
        AZURE_CONFIG_DIR: '/root/.azure'
      with:
        timeout_seconds: 360
        max_attempts: 5
        retry_wait_seconds: 10
        command: |
          az some command

This is a blocking issue right now.

[mcalthrop]

Looks like the following works:

AZURE_CONFIG_DIR: '/home/runner/.azure'

[MoChilia]

Thank you for reporting this! We are now looking into this issue.

[MoChilia]

Hi @Tolsto, @mcalthrop, may I know which runner image you are using to trigger this issue? I want to test the fix.

[mcalthrop]

Hi @Tolsto, @mcalthrop, may I know which runner image you are using to trigger this issue? I want to test the fix.

@MoChilia We've narrowed it down to a specific image that always fails: 20250113.1.0.

It never fails with the previous image: 20250105.1.0.

The problem is that the GitHub code seems to choose the image at random. So it's really hard to test: if the good image is chosen, we have to re-run the workflow until it so happens to choose the bad image.

[MoChilia]

The Azure CLI product team demands an immediate revert of this change: actions/runner-images#11173 (comment).

[Tolsto]

Hi @Tolsto, @mcalthrop, may I know which runner image you are using to trigger this issue? I want to test the fix.

We use ubuntu-latest (for x86) and the default image provided by the ARM runner. As already pointed out by @mcalthrop the rollout of the new image seems to be staged, i.e. failures are random.

[MoChilia]

The GitHub runner image team has started the rollback: actions/runner-images#11400. The issue should be resolved soon. Let's wait for them to complete the rollback.

And I would like to send a big thank you to @Tolsto for bringing this issue to our attention. Your immediate report has helped us prevent a large-scale outage. 👍