fix(agent): handle None response from retry exhaustion to prevent crashes

Fix AttributeError when session.post() returns None after retry exhaustion.
When the server returns repeated 503 errors, the requests retry mechanism
can return None instead of a Response object, causing agents to crash when
accessing status_code attribute.

Fixed methods:
- post_status_update()
- post_result()
- get_result()
- repost_job()
- post_artifacts()

Resolves agents showing EXITED status in supervisorctl.

Fixes #438 (Fixes #CERTTF-474 Internally)

Fix AttributeError when session.post() returns None after retry exhaustion.
When the server returns repeated 503 errors, the requests retry mechanism
can return None instead of a Response object, causing agents to crash when
accessing status_code attribute.

Fixed methods:
- post_status_update()
- post_result()
- get_result()
- save_artifacts()

Resolves agents showing EXITED status in supervisorctl.
Fixes #438 (Fixes #CERTTF-474 Internally)

Author: gajeshbhat

agent/src/testflinger_agent/client.py

@@ -182,12 +182,20 @@ def post_result(self, job_id, data):
             logger.error(exc)
             raise TFServerError("other exception") from exc
         if not job_request:
-            logger.error(
-                "Unable to post results to: %s (error: %d)",
-                result_uri,
-                job_request.status_code,
-            )
-            raise TFServerError(job_request.status_code)
+            if job_request is None:
+                logger.error(
+                    "Unable to post results to: %s (error: %s)",
+                    result_uri,
+                    "No response received",
+                )
+                raise TFServerError("No response received")
+            else:
+                logger.error(
+                    "Unable to post results to: %s (error: %d)",
+                    result_uri,
+                    job_request.status_code,
+                )
+                raise TFServerError(job_request.status_code)
 
     def get_result(self, job_id):
         """Get current results data to the testflinger server for this job.
@@ -206,11 +214,18 @@ def get_result(self, job_id):
             logger.error(exc)
             return {}
         if not job_request:
-            logger.error(
-                "Unable to get results from: %s (error: %d)",
-                result_uri,
-                job_request.status_code,
-            )
+            if job_request is None:
+                logger.error(
+                    "Unable to get results from: %s (error: %s)",
+                    result_uri,
+                    "No response received",
+                )
+            else:
+                logger.error(
+                    "Unable to get results from: %s (error: %d)",
+                    result_uri,
+                    job_request.status_code,
+                )
             return {}
         if job_request.content:
             return job_request.json()
@@ -299,12 +314,20 @@ def save_artifacts(self, rundir, job_id):
                     artifact_uri, files=file_upload, timeout=600
                 )
             if not artifact_request:
-                logger.error(
-                    "Unable to post results to: %s (error: %d)",
-                    artifact_uri,
-                    artifact_request.status_code,
-                )
-                raise TFServerError(artifact_request.status_code)
+                if artifact_request is None:
+                    logger.error(
+                        "Unable to post results to: %s (error: %s)",
+                        artifact_uri,
+                        "No response received",
+                    )
+                    raise TFServerError("No response received")
+                else:
+                    logger.error(
+                        "Unable to post results to: %s (error: %d)",
+                        artifact_uri,
+                        artifact_request.status_code,
+                    )
+                    raise TFServerError(artifact_request.status_code)
             else:
                 shutil.rmtree(artifacts_dir)
 
@@ -490,10 +513,15 @@ def post_status_update(
             )
             # Response code is greater than 399
             if not job_request:
+                error_msg = (
+                    "No response received"
+                    if job_request is None
+                    else f"HTTP {job_request.status_code}"
+                )
                 logger.error(
-                    "Unable to post status updates to: %s (error: %d)",
+                    "Unable to post status updates to: %s (error: %s)",
                     status_update_uri,
-                    job_request.status_code,
+                    error_msg,
                 )
         except requests.exceptions.RequestException as exc:
             logger.error(

agent/tests/test_client.py

@@ -291,3 +291,50 @@ def test_get_agent_data(self, client, requests_mock):
 
         data = client.get_agent_data("test_agent")
         assert data == agent_data
+
+    def test_post_status_update_none_response(self, client, caplog):
+        """
+        Test that post_status_update handles None response gracefully.
+        This simulates the case where retry exhaustion returns None.
+        """
+        job_id = str(uuid.uuid1())
+
+        # Mock the session.post to return None (simulating retry exhaustion)
+        with patch.object(client.session, "post", return_value=None):
+            client.post_status_update("myjobqueue", "http://foo", [], job_id)
+            assert "No response received" in caplog.text
+
+    def test_post_result_none_response(self, client, caplog):
+        """Test that post_result handles None response gracefully."""
+        job_id = str(uuid.uuid1())
+
+        with patch.object(client.session, "post", return_value=None):
+            with pytest.raises(TFServerError, match="No response received"):
+                client.post_result(job_id, {"test": "data"})
+            assert "No response received" in caplog.text
+
+    def test_get_result_none_response(self, client, caplog):
+        """Test that get_result handles None response gracefully."""
+        job_id = str(uuid.uuid1())
+
+        with patch.object(client.session, "get", return_value=None):
+            result = client.get_result(job_id)
+            assert result == {}
+            assert "No response received" in caplog.text
+
+    def test_save_artifacts_error_response(
+        self, client, requests_mock, tmp_path, caplog
+    ):
+        """Test that save_artifacts handles HTTP error response correctly."""
+        job_id = str(uuid.uuid1())
+
+        # Create artifacts directory with a file
+        artifacts_dir = tmp_path / "artifacts"
+        artifacts_dir.mkdir()
+        (artifacts_dir / "test.txt").write_text("test content")
+
+        artifact_uri = f"http://127.0.0.1:8000/v1/result/{job_id}/artifact"
+        requests_mock.post(artifact_uri, status_code=500)
+        with pytest.raises(TFServerError):
+            client.save_artifacts(tmp_path, job_id)
+        assert "Unable to post results" in caplog.text