def test_handle_killed_health_task(self): """Tests handling killed health task""" when = now() node = Node(self.node_agent, self.node, self.scheduler) node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # Get pull task task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) task_1_id = task.id # Kill task after running and get different task next time self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.KILLED, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) self.assertNotEqual(task.id, task_1_id) self.assertTrue(node._conditions.is_health_check_normal)
def test_handle_successful_health_task(self): """Tests handling the health task successfully""" when = now() node = Node(self.node_agent, self.node, self.scheduler) node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # Get health task task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) self.assertEqual(task.agent_id, self.node_agent) # Schedule health task and make sure no new task is ready self.task_mgr.launch_tasks([task], now()) self.assertListEqual([], node.get_next_tasks(when)) self.assertTrue(node._conditions.is_health_check_normal) # Complete pull task, verify no new task update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.FINISHED, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) self.assertListEqual([], node.get_next_tasks(when)) self.assertTrue(node._conditions.is_health_check_normal)
def test_handle_failed_health_task_low_docker_space(self): """Tests handling a failed health task where Docker has low disk space""" when = now() node = Node(self.node_agent, self.node, self.scheduler) node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # Get health task task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) # Fail task with low Docker space exit code self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.FAILED, now(), exit_code=HealthTask.LOW_DOCKER_SPACE_CODE) self.task_mgr.handle_task_update(update) node.handle_task_update(update) # Check node state self.assertEqual(node._state, Node.DEGRADED) self.assertTrue(NodeConditions.LOW_DOCKER_SPACE_ERR.name in node._conditions._active_errors)
def test_handle_regular_cleanup_task(self): """Tests handling a regular cleanup task""" when = now() node = Node(self.node_agent, self.node, self.scheduler) node._last_health_task = when node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # No task since there are no job executions to clean self.assertListEqual([], node.get_next_tasks(when)) # Add job execution and complete task to clean it up node.add_job_execution(self.job_exe) task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(CLEANUP_TASK_ID_PREFIX)) self.assertFalse(task.is_initial_cleanup) self.assertListEqual(task.job_exes, [self.job_exe]) self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update(task.id, task.agent_id, TaskStatusUpdate.FINISHED, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) # No task since all job executions have been cleaned self.assertListEqual([], node.get_next_tasks(when))
def test_handle_lost_health_task(self): """Tests handling lost health task""" when = now() node = Node(self.node_agent, self.node) node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # Get pull task task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) task_1_id = task.id self.assertIsNotNone(task) # Lose task without scheduling and get same task again update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.LOST, now()) node.handle_task_update(update) task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) self.assertEqual(task.id, task_1_id) self.assertTrue(node._conditions.is_health_check_normal) # Lose task with scheduling and get same task again self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.LOST, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) self.assertEqual(task.id, task_1_id) self.assertTrue(node._conditions.is_health_check_normal) # Lose task after running and get same task again self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.LOST, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) task = node.get_next_tasks(when)[0] self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) self.assertEqual(task.id, task_1_id) self.assertTrue(node._conditions.is_health_check_normal)
def test_handle_failed_health_task(self): """Tests handling failed health task""" when = now() node = Node(self.node_agent, self.node) node._initial_cleanup_completed() node._image_pull_completed() node._update_state() # Get health task task = node.get_next_tasks(when)[0] task_1_id = task.id self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX)) # Fail task after running self.task_mgr.launch_tasks([task], now()) update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.RUNNING, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) update = job_test_utils.create_task_status_update( task.id, task.agent_id, TaskStatusUpdate.FAILED, now()) self.task_mgr.handle_task_update(update) node.handle_task_update(update) # Check node state self.assertEqual(node._state, Node.DEGRADED) self.assertTrue(NodeConditions.HEALTH_FAIL_ERR.name in node._conditions._active_errors) # No new health task right away tasks = node.get_next_tasks(when + datetime.timedelta(seconds=5)) self.assertListEqual([], tasks) self.assertFalse(node._conditions.is_health_check_normal) # After error threshold, we should get new health task new_time = when + Node.HEALTH_ERR_THRESHOLD + datetime.timedelta( seconds=5) task = node.get_next_tasks(new_time)[0] self.assertNotEqual(task.id, task_1_id) self.assertTrue(task.id.startswith(HEALTH_TASK_ID_PREFIX))