microsoft/autogen · error · RuntimeError

Runtime is not running.

Error message

Runtime is not running.

What it means

GrpcWorkerAgentRuntime.stop() refuses to stop a runtime that is not running: if _running is False it raises RuntimeError('Runtime is not running.'). Stop is only valid after a successful start(), making the lifecycle strict and errors loud.

Source

Thrown at python/packages/autogen-ext/src/autogen_ext/runtimes/grpc/_worker_runtime.py:311

                    case "response":
                        task = asyncio.create_task(self._process_response(message.response))
                        self._background_tasks.add(task)
                        task.add_done_callback(self._raise_on_exception)
                        task.add_done_callback(self._background_tasks.discard)
                    case "cloudEvent":
                        task = asyncio.create_task(self._process_event(message.cloudEvent))
                        self._background_tasks.add(task)
                        task.add_done_callback(self._raise_on_exception)
                        task.add_done_callback(self._background_tasks.discard)
                    case None:
                        logger.warning("No message")
            except Exception as e:
                logger.error("Error in read loop", exc_info=e)

    async def stop(self) -> None:
        """Stop the runtime immediately."""
        if not self._running:
            raise RuntimeError("Runtime is not running.")
        self._running = False
        # Wait for all background tasks to finish.
        final_tasks_results = await asyncio.gather(*self._background_tasks, return_exceptions=True)
        for task_result in final_tasks_results:
            if isinstance(task_result, Exception):
                logger.error("Error in background task", exc_info=task_result)
        # Close the host connection.
        if self._host_connection is not None:
            try:
                await self._host_connection.close()
            except asyncio.CancelledError:
                pass
        # Cancel the read task.
        if self._read_task is not None:
            self._read_task.cancel()
            try:
                await self._read_task
            except asyncio.CancelledError:

View on GitHub (pinned to 027ecf0a37)

Solutions

1. Mirror the runtime lifecycle: only stop what you started, tracking a started flag
2. In cleanup code, tolerate the error: try/except RuntimeError around stop()
3. Prefer stopping inside the same scope that called start() (async context manager pattern)

Example fix

# before
finally:
    await runtime.stop()  # RuntimeError if never started

# after
finally:
    if runtime._running:
        await runtime.stop()
    # or: wrap in try/except RuntimeError and ignore

if runtime._running:  # private but reliable mirror of the check
    await runtime.stop()

try:
    await runtime.stop()
except RuntimeError as e:
    if "not running" in str(e):
        pass  # nothing to stop
    else:
        raise

When it happens

Trigger: Calling `await runtime.stop()` before start(), after stop() already completed, or in a shutdown/cleanup path where start() was skipped because setup failed early.

Common situations: finally blocks and signal handlers that unconditionally stop the runtime; error paths where start() raised before setting _running; double-stop in atexit/asyncio shutdown hooks.

Related errors

• Connection is not open.
• Runtime is already running.
• Host connection is not set.
• Runtime must be running when sending message.
• Runtime must be running when publishing message.

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/d730b95f37bfdf6d. Report an issue: GitHub.