mlflow/mlflow · error · MlflowException

AsyncArtifactsLoggingQueue is not activated.

Error message

AsyncArtifactsLoggingQueue is not activated.

What it means

AsyncArtifactsLoggingQueue.log_artifacts_async refuses to enqueue work when the queue has not been activated (`_is_activated` is False). Activation happens in `start()`; without it there is no consumer thread, so artifacts would never be uploaded. MLflow raises MlflowException to fail fast instead of silently dropping artifacts.

Source

Thrown at mlflow/utils/async_logging/async_artifacts_logging_queue.py:203

        """Asynchronously logs runs artifacts.

        Args:
            filename: Filename of the artifact to be logged.
            artifact_path: Directory within the run's artifact directory in which to log the
                artifact.
            artifact: The artifact to be logged.

        Returns:
            mlflow.utils.async_utils.RunOperations: An object that encapsulates the
                asynchronous operation of logging the artifact of run data.
                The object contains a list of `concurrent.futures.Future` objects that can be used
                to check the status of the operation and retrieve any exceptions
                that occurred during the operation.
        """
        from mlflow import MlflowException

        if not self._is_activated:
            raise MlflowException("AsyncArtifactsLoggingQueue is not activated.")
        artifact = RunArtifact(
            filename=filename,
            artifact_path=artifact_path,
            artifact=artifact,
            completion_event=threading.Event(),
        )
        self._queue.put(artifact)
        operation_future = self._artifact_status_check_threadpool.submit(
            self._wait_for_artifact, artifact
        )
        return RunOperations(operation_futures=[operation_future])

    def is_active(self) -> bool:
        return self._is_activated

    def _set_up_logging_thread(self) -> None:
        """Sets up the logging thread.

View on GitHub (pinned to 6a27f2decc)

Solutions

1. Call `queue.start()` before the first `log_artifacts_async` call.
2. If using MlflowClient high-level API, let the client manage activation (log via client.log_artifact with async mode) rather than constructing the queue yourself.
3. If the queue was stopped, create a new AsyncArtifactsLoggingQueue and start it rather than restarting the old one.
4. Check `is_active()` before enqueueing and fall back to synchronous artifact logging when inactive.

Example fix

# before
queue = AsyncArtifactsLoggingQueue()
queue.log_artifacts_async(run_id, "model.pkl", "model")

# after
queue = AsyncArtifactsLoggingQueue()
queue.start()
queue.log_artifacts_async(run_id, "model.pkl", "model")

if not queue.is_active() if hasattr(queue, 'is_active') else not queue._is_activated:
    queue.start()
queue.log_artifacts_async(run_id, filename, artifact_path, artifact)

def is_queue_ready(queue) -> bool:
    return bool(getattr(queue, "_is_activated", False))

from mlflow.exceptions import MlflowException
try:
    queue.log_artifacts_async(run_id, filename, artifact_path, artifact)
except MlflowException as e:
    if "not activated" in str(e):
        queue.start()
        queue.log_artifacts_async(run_id, filename, artifact_path, artifact)
    else:
        raise

When it happens

Trigger: Calling `log_artifacts_async` (or `_log_artifact_async` / `_send_artifact` paths) on an AsyncArtifactsLoggingQueue instance whose `start()` was never called, or after `stop()`/termination deactivated it.

Common situations: Instantiating AsyncArtifactsLoggingQueue manually instead of via the client's managed lifecycle; calling log_artifacts_async after the client shut down the queue; reusing a queue object across processes/after pickling without restarting it.

Related errors

• AsyncLoggingQueue is not activated.
• Exception inside the run data logging thread: {e}
• Artifact location not found in trace tags
• Expected mlflow-artifacts:// URI, got ${url.protocol}
• The MLflow Tracing client is not configured. Please call ini

AI-assisted analysis of mlflow/mlflow@6a27f2decc (2026-08-29). Data as JSON: /api/errors/bfca1540ec2d3909. Report an issue: GitHub.