Is there a way to get the number of active and idle workers, I know the information is available in the web UI, but I couldn’t find any Python API calls.
Hmm this is a nice feature request. What are you trying to do?
Right now there’s a call for
ray.cluster_resources() that might be semi-sufficient?
I want to dynamically switch between parallel processing and single processing depending on the workload, so I can sidestep any unnecessary overhead. I think I can make do with ray.available_resources(), thank you.