I have a map_batches call in ray.data that starts an actor. The __init__ of this actor is very slow and I get ray.exceptions.GetTimeoutError: Timed out while starting actors.
How can I avoid id? I do not find a parameter to avoid it
thx
Could you try configuring DataContext.wait_for_min_actors_s? We added the setting in May 2024: [Data] Allow user to configure timeout for actor pool by bveeramani · Pull Request #45508 · ray-project/ray · GitHub