azure / azure-python-devtools Goto Github PK

Not, I think, for the same reason as #16, but it might make sense to fix both simultaneously.

This issue is actually two related problems with slightly different solutions:

• The subscription ID isn't removed from the request body because process_request only replaces the uri field.
• It's not removed from the response body because _replace_subscription_id's regex substitution is case sensitive: it searches for /subscriptions/([^/]+/ but sometimes it appears in the URI as /Subscriptions/....

Hi!

While updating the azure-python-devtools Debian package, I came across some Python3-only syntax in the ci_tools submodule (the keyword-only argument lone *).

I'll exclude that submodule from the Python2 Debian package, but I guess the packaging metadata should be updated, and/or the submodule excluded from setup.py on Python2.

Thanks!

As it stands, the two consumers of azure-devtools both override ReplayableTest.create_random_name, which is good because I think the default implementation might not work.

In particular, due to this line in azure_devtools.scenario_tests.utilities.create_random_name (which IntegrationTestBase.create_random_name calls):

random_bytes = os.urandom(int(math.ceil(float(padding_size) / 8) * 5))

the names it generates will be actually random and not consistent between test runs. (I haven't yet tested this but I can't see how it would be otherwise. This would also be a prime candidate for #5 as well.)

The logic of header value fetching is flawed. Although the given header name is case-lowered, the header dictionary is not processed to be lower cased:
https://github.com/Azure/azure-python-devtools/blob/master/src/azure_devtools/scenario_tests/recording_processors.py#L22

Sample recording content:

- request:
    body: null
    headers:
      Accept: [application/json]
      Accept-Encoding: ['gzip, deflate']
      CommandName: [functionapp create]
      Connection: [keep-alive]
      Content-Length: ['0']
      Content-Type: [application/json; charset=utf-8]
      User-Agent: [python/3.5.4 (Windows-10-10.0.16299-SP0) requests/2.18.4 msrest/0.4.29
          msrest_azure/0.4.29 azure-mgmt-storage/1.5.0 Azure-SDK-For-Python AZURECLI/2.0.33]
      accept-language: [en-US]
    method: POST
    uri: https://management.azure.com/subscriptions/00000000-0000-0000-0000-000000000000/resourceGroups/clitest.rg000001/providers/Microsoft.Storage/storageAccounts/functionappplanstorage/listKeys?api-version=2017-10-01
  response:
    body: {string: '{"keys":[{"keyName":"key1","value":"...","permissions":"FULL"},{"keyName":"key2","value":"...","permissions":"FULL"}]}'}

The "SkuNotAvailable" response for VM creation has a body that looks like this:

    body: {string: '{"error": {"code": "SkuNotAvailable", "message": "The requested
        tier for resource ''/subscriptions/11111111-1111-1111-1111-111111111111/resourceGroups/azure-sample-group-virtual-machines/providers/Microsoft.Compute/virtualMachines/VmName''
        is currently not available in location ''westus'' for subscription ''ACTUAL-SUBSCRIPTION-ID-SHOWN-HERE''.
        Please try another tier or deploy to a different location."}, "access_token":
        "fake_token"}'}

and where the string ACTUAL-SUBSCRIPTION-ID-SHOWN-HERE appears in that example, SubscriptionRecordingProcessor fails to remove it.

azure_devtools.scenario_tests.recording_processors.OAuthRequestResponsesFilter doesn't recognize MSAL requests due to the URL change in AAD v2.0.

This results in all https://login.microsoftonline.com requests being recorded, including

• Tenant discovery request: GET https://login.microsoftonline.com/54826b22-38d6-4fb2-bad9-b7b93a3e9c5a/v2.0/.well-known/openid-configuration
• Get access token request: POST https://login.microsoftonline.com/54826b22-38d6-4fb2-bad9-b7b93a3e9c5a/oauth2/v2.0/token

Actually, all requests to https://login.microsoftonline.com should be ignored as it is related to authentication which is patched in playback mode.

1. Support serializing request/response payload with binary content. To reproduce, just try out the webapp log download test
2. For large payload, suggest throw during the recording instead of injecting !!! The request body has been omitted from the recording because its.... You know, the test will still fail during the playback but becomes harder for me to figure out the cause since the test did pass during the recording

This appears in the Jenkins output for the 0.5.0 release, which was not successfully pushed to PyPI:

Uploading distributions to https://pypi.python.org/pypi
Note: you are uploading to the old upload URL. It's recommended to use the new URL
"https://upload.pypi.org/legacy/" or to leave the URL unspecified and allow twine to choose.
Uploading azure_devtools-0.5.0-py2.py3-none-any.whl
HTTPError: 410 Client Error: Gone (This API has been deprecated and removed from legacy PyPI
in favor of using the APIs available in the new PyPI.org implementation of PyPI
(located at https://pypi.org/).
This is a 24h brownout of the API and access will be restored on June 30th. Please switch to PyPI.ORG
for uploads prior to July 3rd when this API will be removed for good. For more information about
migrating your use of this API to PyPI.org, please see
https://packaging.python.org/guides/migrating-to-pypi-org/#uploading.
For more information about the sunsetting of this API, please see
https://mail.python.org/pipermail/distutils-sig/2017-June/030766.html)
for url: https://pypi.python.org/pypi

It would be nice to have randomized VALID characters instead.

For dis upload scenario, the size should be 1024*x + 512 bytes, which will require size with "x.5" kb

Running azdev test is failing on our private preview extension with the following error, please help resolve this

self = <azext_k8s_extension.tests.latest.test_k8s_extension_scenario.K8sExtensionScenarioTest testMethod=test_k8s_extension>

    def tearDown(self):
        os.environ = self.original_env
        # Autorest.Python 2.x
        assert not [t for t in threading.enumerate() if t.name.startswith("AzureOperationPoller")], \
            "You need to call 'result' or 'wait' on all AzureOperationPoller you have created"
        # Autorest.Python 3.x
>       assert not [t for t in threading.enumerate() if t.name.startswith("LROPoller")], \
            "You need to call 'result' or 'wait' on all LROPoller you have created"
E       AssertionError: You need to call 'result' or 'wait' on all LROPoller you have created

env\lib\site-packages\azure_devtools\scenario_tests\base.py:154: AssertionError
------------------------------------------------------------------------------------------------------------------ Captured stderr call ------------------------------------------------------------------------------------------------------------------- 
WARNING: Command group 'k8s-extension' is in preview and under development. Reference and support levels: https://aka.ms/CLI_refstatus
Ignoring name, release-namespace and scope parameters since microsoft.azuremonitor.containers only supports cluster scope and single instance of this extension
--------------------------------------------------------------------- generated xml file: C:\Users\joinnis\.azdev\env_config\Users\joinnis\github\narayan-cli-pr\env\test_results.xml --------------------------------------------------------------------- 
================================================================================================================= short test summary info ================================================================================================================= 
FAILED src\k8s-extension\azext_k8s_extension\tests\latest\test_k8s_extension_scenario.py::K8sExtensionScenarioTest::test_k8s_extension - AssertionError: You need to call 'result' or 'wait' on all LROPoller you have created
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! stopping after 1 failures !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! xdist.dsession.Interrupted: stopping after 1 failures !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
=================================================================================================================== 1 failed in 11.24s ====================================================================================================================

Issues:

1. Too few public methods, which makes me wonder does this class has to be a class? (https://www.youtube.com/watch?v=o9pEzgHorH0)
2. The usage of the ArgumentParser in it is weird. It essentially parses an empty array. Why?

As suggested in #69 (comment)

/home/travis/virtualenv/python3.6.7/lib/python3.6/site-packages/azure_devtools/scenario_tests/utilities.py:73: DeprecationWarning: inspect.getargspec() is deprecated, use inspect.signature() or inspect.getfullargspec()

verbose on Travis

I created 3 preparers: P1, P2 and P3. These depend on each other as such: P3 depends on P2 and P2 depends on P1. I've decorated the preparers on my test method in the order: P1, P2, P3. During resource creation, everything goes well. But during resource deletion, the order seems to be: P1, P2, P3 - whereas the order should be: P3, P2, P1.

Please look into this!

https://github.com/Azure/azure-python-devtools/blob/master/doc/recording_vcr_tests.md

The version number handling is a little sloppy. There's a __version__ in azure_devtools.scenario_tests.__init__ but not in azure_devtools.__init__, and the version number in setup.py is duplicated rather than imported as it probably should be. This makes releases a little messier than they should be.

Currently ReplayableTest.__init__ passes self.config.record_mode through unchanged as record_mode to vcr.VCR(). But it's a boolean value, which is not an appropriate value for that argument. It should be a string. Somehow it seems to work ok regardless? But I think it only does so incidentally.

To match what I believe we expect (and what I claim is done in #19), we should pass record_mode like this:

record_mode='all' if self.config.record_mode else 'once'

https://github.com/Azure/azure-python-devtools/blob/master/src/azure_devtools/scenario_tests/base.py#L66

It is not useful at all.

kevin1024/vcrpy#318

• If the issue is to do with Azure CLI 2.0 in-particular, create an issue here at Azure/azure-cli

Extension name (the extension in question)

k8sconfiguration

Description of issue (in as much detail as possible)

Parameter names must be under 23 characters according to the linter. When specifying a name in the options_list for an argument, my shortest parameter name --<parameter_name> is throwing an error in the linter because the linter includes the dashes in front of the parameter when taking the number of characters into account.

When no options_list is specified and the default argument name is used, the linter will not include these dashes in counting.

For instance, the parameter name --config-protected-file raises an error when it shouldn't