[Appservice] az functionapp deployment source config-zip: Fix intermittent failure on flex consumption apps due to Kudu restart

src/azure-cli/azure/cli/command_modules/appservice/custom.py

@@ -9652,21 +9652,40 @@ def _check_zip_deployment_status_flex(cmd, rg_name, name, deployment_status_url,
     # Indicates whether the status has been non empty in previous calls
     has_response = False
     has_partial_success = False
+    # Indicates Kudu is restarting after package deployment (transient state for flex consumption apps)
+    kudu_restart_in_progress = False
+    # Tracks whether the Kudu-restart 404 warning has already been emitted once
+    kudu_restart_404_warned = False
+    # Default return value; overwritten on each successful JSON response
+    res_dict = {}
     while num_trials < total_trials:
         time.sleep(1)
         response = requests.get(deployment_status_url, headers=headers,
                                 verify=not should_disable_connection_verify())
         try:
             if response.status_code == 202 and not has_partial_success:
                 has_partial_success = True
-            if response.status_code == 404 and has_partial_success:
-                break
-            if (response.status_code == 404 or response.json().get('status') is None) and has_response:
-                raise CLIError("Failed to retrieve deployment status. Please try again in a few minutes.")
-            if (response.status_code != 404 and response.json().get('status') is not None) and not has_response:
-                has_response = True
-
-            res_dict = response.json()
+            if response.status_code == 404:
+                if has_partial_success and not kudu_restart_in_progress:
+                    break
+                elif kudu_restart_in_progress:
+                    # Kudu is restarting after package deployment — continue polling until it comes back
+                    if not kudu_restart_404_warned:
+                        kudu_restart_404_warned = True
+                        logger.warning("Deployment status endpoint returned 404. "
+                                       "Kudu may be restarting after package deployment. Retrying...")
+                    else:
+                        logger.info("Deployment status endpoint still returning 404 during Kudu restart. Retrying...")
+                elif has_response:
+                    raise CLIError("Failed to retrieve deployment status. Please try again in a few minutes.")
+            else:
+                res_json = response.json()
+                res_status = res_json.get('status')
+                if res_status is None and has_response:
+                    raise CLIError("Failed to retrieve deployment status. Please try again in a few minutes.")
+                if res_status is not None and not has_response:
+                    has_response = True
+                res_dict = res_json
         except json.decoder.JSONDecodeError:
             logger.warning("Deployment status endpoint %s returns malformed data. Retrying...", deployment_status_url)
             res_dict = {}
@@ -9685,12 +9704,14 @@ def _check_zip_deployment_status_flex(cmd, rg_name, name, deployment_status_url,
         if status == 5:
             raise CLIError("Deployment was cancelled and another deployment is in progress.")
         if status == 6:
-            raise CLIError("Deployment was partially successful. These are the deployment logs:\n{}".format(
-                           json.dumps(show_deployment_log(cmd, rg_name, name))))
+            if not kudu_restart_in_progress:
+                kudu_restart_in_progress = True
+                logger.warning("Deployment is partially complete. Kudu may be restarting after package "
+                               "deployment. Continuing to poll for completion...")
         if 'progress' in res_dict:
             logger.info(res_dict['progress'])  # show only in debug mode, customers seem to find this confusing
     # if the deployment is taking longer than expected
-    if res_dict.get('status', 0) != 4 and not has_partial_success:
+    if res_dict.get('status', 0) != 4 and not has_partial_success and not kudu_restart_in_progress:
         raise CLIError("""Timeout reached by the command, however, the deployment operation
                        is still on-going. Navigate to your scm site to check the deployment status""")
     return res_dict

src/azure-cli/azure/cli/command_modules/appservice/tests/latest/test_functionapp_commands_thru_mock.py

@@ -12,6 +12,7 @@
     enable_zip_deploy_functionapp,
     enable_zip_deploy,
     enable_zip_deploy_flex,
+    _check_zip_deployment_status_flex,
     add_remote_build_app_settings,
     remove_remote_build_app_settings,
     config_source_control,
@@ -681,3 +682,99 @@ def test_config_source_control(self,
 
         # assert
         self.assertEqual(response.git_hub_action_configuration.container_configuration.password, None)
+
+
+class TestCheckZipDeploymentStatusFlex(unittest.TestCase):
+    """Tests for _check_zip_deployment_status_flex handling of Kudu restart (status 6) scenario."""
+
+    def _make_response(self, status_code, json_body=None):
+        """Create a mock HTTP response."""
+        resp = mock.MagicMock()
+        resp.status_code = status_code
+        if json_body is not None:
+            resp.json.return_value = json_body
+        else:
+            resp.json.side_effect = Exception("No JSON body")
+        return resp
+
+    @mock.patch('azure.cli.command_modules.appservice.custom.get_scm_site_headers_flex', return_value={})
+    @mock.patch('azure.cli.core.util.should_disable_connection_verify', return_value=False)
+    @mock.patch('time.sleep', return_value=None)
+    @mock.patch('requests.get')
+    def test_status_6_then_status_4_succeeds(self, requests_get_mock, sleep_mock,
+                                              should_disable_mock, get_headers_mock):
+        """Status 6 (Kudu restart) followed by status 4 (success) should not raise an error."""
+        cmd_mock = _get_test_cmd()
+
+        # Simulate: first poll returns status 6 (partial success / Kudu restarting),
+        # second poll returns status 4 (success)
+        requests_get_mock.side_effect = [
+            self._make_response(200, {'status': 6, 'progress': ''}),
+            self._make_response(200, {'status': 4, 'complete': True}),
+        ]
+
+        result = _check_zip_deployment_status_flex(cmd_mock, 'rg', 'name',
+                                                   'https://mock-scm/api/deployments/latest',
+                                                   timeout=None)
+        self.assertEqual(result.get('status'), 4)
+
+    @mock.patch('azure.cli.command_modules.appservice.custom.get_scm_site_headers_flex', return_value={})
+    @mock.patch('azure.cli.core.util.should_disable_connection_verify', return_value=False)
+    @mock.patch('time.sleep', return_value=None)
+    @mock.patch('requests.get')
+    def test_status_6_then_404_then_status_4_succeeds(self, requests_get_mock, sleep_mock,
+                                                       should_disable_mock, get_headers_mock):
+        """Status 6 followed by 404 (Kudu restarting) followed by status 4 should succeed."""
+        cmd_mock = _get_test_cmd()
+
+        requests_get_mock.side_effect = [
+            self._make_response(200, {'status': 6, 'progress': ''}),
+            self._make_response(404),
+            self._make_response(200, {'status': 4, 'complete': True}),
+        ]
+
+        result = _check_zip_deployment_status_flex(cmd_mock, 'rg', 'name',
+                                                   'https://mock-scm/api/deployments/latest',
+                                                   timeout=None)
+        self.assertEqual(result.get('status'), 4)
+
+    @mock.patch('azure.cli.command_modules.appservice.custom.get_scm_site_headers_flex', return_value={})
+    @mock.patch('azure.cli.core.util.should_disable_connection_verify', return_value=False)
+    @mock.patch('time.sleep', return_value=None)
+    @mock.patch('requests.get')
+    def test_status_6_multiple_404_then_status_4_succeeds(self, requests_get_mock, sleep_mock,
+                                                           should_disable_mock, get_headers_mock):
+        """Multiple 404 responses during Kudu restart should continue polling until success."""
+        cmd_mock = _get_test_cmd()
+
+        requests_get_mock.side_effect = [
+            self._make_response(200, {'status': 6, 'progress': ''}),
+            self._make_response(404),
+            self._make_response(404),
+            self._make_response(200, {'status': 4, 'complete': True}),
+        ]
+
+        result = _check_zip_deployment_status_flex(cmd_mock, 'rg', 'name',
+                                                   'https://mock-scm/api/deployments/latest',
+                                                   timeout=None)
+        self.assertEqual(result.get('status'), 4)
+
+    @mock.patch('azure.cli.command_modules.appservice.custom.get_scm_site_headers_flex', return_value={})
+    @mock.patch('azure.cli.core.util.should_disable_connection_verify', return_value=False)
+    @mock.patch('time.sleep', return_value=None)
+    @mock.patch('requests.get')
+    def test_status_3_still_raises_error(self, requests_get_mock, sleep_mock,
+                                          should_disable_mock, get_headers_mock):
+        """Status 3 (deployment failed) should still raise an error."""
+        cmd_mock = _get_test_cmd()
+
+        with mock.patch('azure.cli.command_modules.appservice.custom.show_deployment_log', return_value=[]):
+            requests_get_mock.side_effect = [
+                self._make_response(200, {'status': 3}),
+            ]
+
+            with self.assertRaises(CLIError) as cm:
+                _check_zip_deployment_status_flex(cmd_mock, 'rg', 'name',
+                                                  'https://mock-scm/api/deployments/latest',
+                                                  timeout=None)
+            self.assertIn("Zip deployment failed", str(cm.exception))