How to tell cluster launcher to not reserve so much memory for a ray worker node?

How severe does this issue affect your experience of using Ray?

  • Low: It annoys or frustrates me for a moment.

How do I tell the cluster launcher to not give so much memory to the Ray object store? Ideally, I’d have Ray only use of 5GB or so of memory on the worker nodes.

You can hardcode the value you want via the resources field

available_node_types:
  my_node_type:
    resources:
      memory: XXXXXX # in bytes
      object_store_memory: YYYYYY # in bytes