What causes "Runtime is not started" in microsoft/autogen?

Call stop() exactly once per start(); track lifecycle in your orchestrator with a flag or use try/except RuntimeError in cleanup as a deliberate idempotency guard

microsoft/autogen · error · RuntimeError

Runtime is not started

Error message

Runtime is not started

What it means

stop() raises RuntimeError when self._run_context is None, meaning the runtime's processing loop is not currently running. Stopping requires an active run context; stopping an idle/never-started (or already-stopped) runtime is an error, not a no-op.

Source

Thrown at python/packages/autogen-core/src/autogen_core/_single_threaded_agent_runtime.py:836

        """
        if self._run_context is not None:
            raise RuntimeError("Runtime is already started")
        self._run_context = RunContext(self)

    async def close(self) -> None:
        """Calls :meth:`stop` if applicable and the :meth:`Agent.close` method on all instantiated agents"""
        # stop the runtime if it hasn't been stopped yet
        if self._run_context is not None:
            await self.stop()
        # close all the agents that have been instantiated
        for agent_id in self._instantiated_agents:
            agent = await self._get_agent(agent_id)
            await agent.close()

    async def stop(self) -> None:
        """Immediately stop the runtime message processing loop. The currently processing message will be completed, but all others following it will be discarded."""
        if self._run_context is None:
            raise RuntimeError("Runtime is not started")

        try:
            await self._run_context.stop()
        finally:
            self._run_context = None
            self._message_queue = Queue()

    async def stop_when_idle(self) -> None:
        """Stop the runtime message processing loop when there is
        no outstanding message being processed or queued. This is the most common way to stop the runtime."""
        if self._run_context is None:
            raise RuntimeError("Runtime is not started")

        try:
            await self._run_context.stop_when_idle()
        finally:
            self._run_context = None
            self._message_queue = Queue()

View on GitHub (pinned to 027ecf0a37)

Solutions

Call stop() exactly once per start(); track lifecycle in your orchestrator with a flag or use try/except RuntimeError in cleanup as a deliberate idempotency guard
Prefer await runtime.close() for teardown: it stops only if not already stopped and also closes agents
Ensure close()/stop() are not both invoked in the same teardown sequence for the stop portion

Example fix

# before
await runtime.stop_when_idle()
await runtime.stop()  # RuntimeError

# after
await runtime.stop_when_idle()
await runtime.close()  # safe: no-ops the stop portion, closes agents

Defensive patterns

Strategy: try-catch

Validate before calling

def can_stop(runtime) -> bool:
    return runtime._run_context is not None

Try / catch

try:
    await runtime.stop()
except RuntimeError as e:
    if "not started" in str(e):
        pass  # already stopped; idempotent teardown
    else:
        raise

Prevention

Prefer close() for teardown (it stops only if needed and closes agents)
Run stop in exactly one place (success path or finally, not both unguarded)
Track runtime lifecycle state in your app if you stop at multiple points

When it happens

Trigger: Calling await runtime.stop() before any start(), after stop()/stop_when_idle() already completed, or after close() (which itself calls stop when applicable). Double-stop in cleanup paths is the most common form.

Common situations: Cleanup/teardown code running on both the success and exception paths so stop() executes twice; orchestrators that stop at end-of-task and again at shutdown; tests tearing down a runtime that a previous line already stopped.

Related errors

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/ec7ad5e2b7c034ed. Report an issue: GitHub.