How severe does this issue affect your experience of using Ray?
High: It blocks me to complete my task.
I am thinking some personal project.
I have a ray cluster on aws. it is possible for my local computer (mac) join the ray cluster as a worker?
Is there any document about this?
UPDATE
i have a ray cluster running on AWS K8S:
in one terminal I run
aws --profile stage eks update-kubeconfig --name stag
kubectl --namespace demo port-forward svc/kuberay-head-svc 6379:6379
(also forwarded dashboard ports)
(then i can access ray dashboard and saw two nodes there, which make sense)
And in another terminal I run
docker run --platform linux/amd64 --network="host" -it rayproject/ray:latest
ray start --address=127.0.0.1:6379
and I got
% docker run --shm-size=5.07gb --platform linux/amd64 --network=host -it rayproject/ray:2.9.0
(base) ray@docker-desktop:~$ RAY_REDIS_ADDRESS=host.docker.internal:6379 ray start --address=host.docker.internal:6379
<jemalloc>: MADV_DONTNEED does not work (memset will be used instead)
<jemalloc>: (This is the expected behaviour if you are running under QEMU)
Local node IP: xxx
[2024-07-12 20:49:48,664 W 38 38] global_state_accessor.cc:465: Some processes that the driver needs to connect to have not registered with GCS, so retrying. Have you run 'ray start' on this node?
[2024-07-12 20:49:49,748 W 38 38] global_state_accessor.cc:465: Some processes that the driver needs to connect to have not registered with GCS, so retrying. Have you run 'ray start' on this node?
--------------------
Ray runtime started.
--------------------
To terminate the Ray runtime, run
ray stop
BUt from the cluster dashboard, i saw the worker become DEAD after several seconds Active. i guess the cluster on AWS EKS cannot access my local docker container on my mac. any idea? thanks