[Bug Report] Missing Ray initialization while running`tuner.py`

[ozhanozen]

Ray is not initialized while running tuner.py, throwing an exception.

Steps to reproduce

Try to run the Docker-based Local Quickstart part of the docs. You should get the following exception:
"Ray is not initialized. Please initialize Ray before getting node resources."

Additional context

PR #2847, commit 286eb39, modified get_gpu_node_resources() to be:

IsaacLab/scripts/reinforcement_learning/ray/util.py

Lines 345 to 346 in 5dae83e

	if not ray.is_initialized():
	raise Exception("Ray is not initialized. Please initialize Ray before getting node resources.")

which was originally:

if not ray.is_initialized():
        ray.init(address=ray_address)

Now, ray is not initialized when tuner.py is directly used.

Is there a reason to change the previous implementation? If not, we could simply revert this line or use the new ray_init().

System Info

• Commit: 5dae83e
• Isaac Sim Version: 5.0
• OS: Ubuntu 22.04
• GPU: RTX 4090
• CUDA: 12.2
• GPU Driver: 535.129.03

Checklist

• I have checked that there is no similar issue in the repo (required)
• I have checked that the issue is not in running Isaac Sim itself and is related to the repo

Acceptance Criteria

• tuner.py runs without errors

[ozhanozen]

Hi @garylvov and @binw666, I'm tagging you because I thought you might know best if reverting the aforementioned line is safe for the new functionalities, or alternatively, if doing ray_init() somewhere else is better.

[garylvov]

I think reverting to what we had before is safe I can test with both

[binw666]

Hi @garylvov and @binw666, I'm tagging you because I thought you might know best if reverting the aforementioned line is safe for the new functionalities, or alternatively, if doing ray_init() somewhere else is better.

I'm very sorry for introducing the bug. I think reverting this line or using the new ray_init() is safe for the new feature.

[ozhanozen]

No problem, I will create a PR for the fix then.