def on_worker_event_running(machine_id: UUID,
event: WorkerRunningEvent) -> Result[None]:
task = Task.get_by_task_id(event.task_id)
if isinstance(task, Error):
return task
node = get_node(machine_id)
if isinstance(node, Error):
return node
if node.state not in NodeState.ready_for_reset():
node.state = NodeState.busy
node.save()
node_task = NodeTasks(machine_id=machine_id,
task_id=event.task_id,
state=NodeTaskState.running)
node_task.save()
if task.state in TaskState.shutting_down():
logging.info(
"ignoring task start from node. machine_id:%s %s:%s (state: %s)",
machine_id,
task.job_id,
task.task_id,
task.state,
)
return None
logging.info(
"task started on node. machine_id:%s %s:%s",
machine_id,
task.job_id,
task.task_id,
)
task.state = TaskState.running
task.save()
# Start the clock for the task if it wasn't started already
# (as happens in 1.0.0 agents)
task.on_start()
task_event = TaskEvent(
task_id=task.task_id,
machine_id=machine_id,
event_data=WorkerEvent(running=event),
)
task_event.save()
return None
def on_worker_event_done(machine_id: UUID,
event: WorkerDoneEvent) -> Result[None]:
task = Task.get_by_task_id(event.task_id)
if isinstance(task, Error):
return task
node = get_node(machine_id)
if isinstance(node, Error):
return node
if event.exit_status.success:
logging.info("task done. %s:%s status:%s", task.job_id, task.task_id,
event.exit_status)
task.mark_stopping()
if (task.config.debug and TaskDebugFlag.keep_node_on_completion
in task.config.debug):
node.debug_keep_node = True
node.save()
else:
task.mark_failed(
Error(
code=ErrorCode.TASK_FAILED,
errors=[
"task failed. exit_status:%s" % event.exit_status,
event.stdout[-MAX_OUTPUT_SIZE:],
event.stderr[-MAX_OUTPUT_SIZE:],
],
))
if task.config.debug and (
TaskDebugFlag.keep_node_on_failure in task.config.debug
or TaskDebugFlag.keep_node_on_completion in task.config.debug):
node.debug_keep_node = True
node.save()
if not node.debug_keep_node:
node_task = NodeTasks.get(machine_id, event.task_id)
if node_task:
node_task.delete()
event.stdout = event.stdout[-MAX_OUTPUT_SIZE:]
event.stderr = event.stderr[-MAX_OUTPUT_SIZE:]
task_event = TaskEvent(task_id=task.task_id,
machine_id=machine_id,
event_data=WorkerEvent(done=event))
task_event.save()
return None
def on_worker_event_running(machine_id: UUID,
event: WorkerRunningEvent) -> Result[None]:
task = Task.get_by_task_id(event.task_id)
if isinstance(task, Error):
return task
node = get_node(machine_id)
if isinstance(node, Error):
return node
if node.state not in NodeState.ready_for_reset():
node.set_state(NodeState.busy)
node_task = NodeTasks(machine_id=machine_id,
task_id=event.task_id,
state=NodeTaskState.running)
node_task.save()
if task.state in TaskState.shutting_down():
logging.info(
"ignoring task start from node. "
"machine_id:%s job_id:%s task_id:%s (state: %s)",
machine_id,
task.job_id,
task.task_id,
task.state,
)
return None
logging.info(
"task started on node. machine_id:%s job_id%s task_id:%s",
machine_id,
task.job_id,
task.task_id,
)
task.set_state(TaskState.running)
task_event = TaskEvent(
task_id=task.task_id,
machine_id=machine_id,
event_data=WorkerEvent(running=event),
)
task_event.save()
return None