xds interop: Log operation id (#30407)

Add consistent operation id logs for GCP long-running operations - both old-style (compute) and the new APIs.
At the moment it's a bit more verbose than I'd want, f.e. it doubles the number of log messages during the teardown. We should probably only log failed ops. But to do this reliably, we should probably revisit the issue with improving tenacity retry error fail reports.

tools/run_tests/xds_k8s_test_driver/framework/infrastructure/gcp/api.py

@@ -435,19 +435,19 @@ class GcpStandardCloudApiResource(GcpProjectApiResource, metaclass=abc.ABCMeta):
     def _execute(  # pylint: disable=arguments-differ
             self,
             request: HttpRequest,
-            timeout_sec=GcpProjectApiResource._WAIT_FOR_OPERATION_SEC):
+            timeout_sec: int = GcpProjectApiResource._WAIT_FOR_OPERATION_SEC):
         operation = request.execute(num_retries=self._GCP_API_RETRIES)
-        self._wait(operation, timeout_sec)
+        logger.debug('Operation %s', operation)
+        self._wait(operation['name'], timeout_sec)
 
     def _wait(self,
-              operation,
-              timeout_sec=GcpProjectApiResource._WAIT_FOR_OPERATION_SEC):
-        op_name = operation['name']
-        logger.debug('Waiting for %s operation, timeout %s sec: %s',
-                     self.api_name, timeout_sec, op_name)
+              operation_id: str,
+              timeout_sec: int = GcpProjectApiResource._WAIT_FOR_OPERATION_SEC):
+        logger.info('Waiting %s sec for %s operation id: %s', timeout_sec,
+                    self.api_name, operation_id)
 
         op_request = self.api.projects().locations().operations().get(
-            name=op_name)
+            name=operation_id)
         operation = self.wait_for_operation(
             operation_request=op_request,
             test_success_fn=lambda result: result['done'],

tools/run_tests/xds_k8s_test_driver/framework/infrastructure/gcp/compute.py

@@ -430,28 +430,29 @@ class ComputeV1(gcp.api.GcpProjectApiResource):  # pylint: disable=too-many-publ
             self,
             request,
             *,
-            test_success_fn=None,
             timeout_sec=_WAIT_FOR_OPERATION_SEC):
         operation = request.execute(num_retries=self._GCP_API_RETRIES)
-        logger.debug('Response %s', operation)
+        logger.debug('Operation %s', operation)
+        return self._wait(operation['name'], timeout_sec)
+
+    def _wait(self,
+              operation_id: str,
+              timeout_sec: int = _WAIT_FOR_OPERATION_SEC) -> dict:
+        logger.info('Waiting %s sec for compute operation id: %s', timeout_sec,
+                    operation_id)
 
         # TODO(sergiitk) try using wait() here
         # https://googleapis.github.io/google-api-python-client/docs/dyn/compute_v1.globalOperations.html#wait
-        operation_request = self.api.globalOperations().get(
-            project=self.project, operation=operation['name'])
-
-        if test_success_fn is None:
-            test_success_fn = self._operation_status_done
-
-        logger.debug('Waiting for global operation %s, timeout %s sec',
-                     operation['name'], timeout_sec)
-        response = self.wait_for_operation(operation_request=operation_request,
-                                           test_success_fn=test_success_fn,
-                                           timeout_sec=timeout_sec)
-
-        if 'error' in response:
-            logger.debug('Waiting for global operation failed, response: %r',
-                         response)
-            raise Exception(f'Operation {operation["name"]} did not complete '
-                            f'within {timeout_sec}s, error={response["error"]}')
-        return response
+        op_request = self.api.globalOperations().get(project=self.project,
+                                                     operation=operation_id)
+        operation = self.wait_for_operation(
+            operation_request=op_request,
+            test_success_fn=self._operation_status_done,
+            timeout_sec=timeout_sec)
+
+        logger.debug('Completed operation: %s', operation)
+        if 'error' in operation:
+            # This shouldn't normally happen: gcp library raises on errors.
+            raise Exception(f'Compute operation {operation_id} '
+                            f'failed: {operation}')
+        return operation