microsoft/autogen · error · ValueError

Runtime must be running when publishing message.

Error message

Runtime must be running when publishing message.

What it means

GrpcWorkerAgentRuntime.publish_message requires the runtime to be running: if _running is False it raises ValueError('Runtime must be running when publishing message.') since publishing needs the host connection established by start().

Source

Thrown at python/packages/autogen-ext/src/autogen_ext/runtimes/grpc/_worker_runtime.py:423

            # TODO: Find a way to handle timeouts/errors
            task = asyncio.create_task(self._send_message(runtime_message, "send", recipient, telemetry_metadata))
            self._background_tasks.add(task)
            task.add_done_callback(self._raise_on_exception)
            task.add_done_callback(self._background_tasks.discard)
            return await future

    async def publish_message(
        self,
        message: Any,
        topic_id: TopicId,
        *,
        sender: AgentId | None = None,
        cancellation_token: CancellationToken | None = None,
        message_id: str | None = None,
    ) -> None:
        if not self._running:
            raise ValueError("Runtime must be running when publishing message.")
        if self._host_connection is None:
            raise RuntimeError("Host connection is not set.")
        if message_id is None:
            message_id = str(uuid.uuid4())

        message_type = self._serialization_registry.type_name(message)
        with self._trace_helper.trace_block(
            "create", topic_id, parent=None, extraAttributes={"message_type": message_type}
        ):
            serialized_message = self._serialization_registry.serialize(
                message, type_name=message_type, data_content_type=self._payload_serialization_format
            )

            sender_id = sender or AgentId("unknown", "unknown")
            attributes = {
                _constants.DATA_CONTENT_TYPE_ATTR: cloudevent_pb2.CloudEvent.CloudEventAttributeValue(
                    ce_string=self._payload_serialization_format
                ),

View on GitHub (pinned to 027ecf0a37)

Solutions

  1. Await runtime.start() before any publish_message call
  2. Move eager publishes from constructors into the agent's first message handler or an explicit on_start hook
  3. Gate publishers on runtime state during shutdown to avoid racing stop()

Example fix

# before
await runtime.publish_message(event, topic)  # before start()
await runtime.start()

# after
await runtime.start()
await runtime.publish_message(event, topic)
Defensive patterns

Strategy: validation

Validate before calling

if not started:
    await runtime.start()
    started = True
await runtime.publish_message(event, topic)

Try / catch

try:
    await runtime.publish_message(event, topic)
except ValueError as e:
    if "must be running" in str(e):
        await runtime.start()
        await runtime.publish_message(event, topic)
    else:
        raise

Prevention

When it happens

Trigger: Calling `await runtime.publish_message(msg, topic_id)` before await runtime.start() completes or after stop(); registering a callback/type that publishes during registration before start.

Common situations: Publishing from module-level or __init__ code of an agent or service before the runtime started; shutdown ordering where background publishers outlive stop(); notebook cell re-runs.

Related errors

AI-assisted analysis of microsoft/autogen@027ecf0a37 (2026-08-15). Data as JSON: /api/errors/f322e7b558914712. Report an issue: GitHub.