Ray.init() hangs on macOS M4

Ray Clusters
1

I simply am doing:

ray.init()

on my MacOS M4 with ray version 2.42.1, and it hangs indefinitely. Is there any workaround? I have a need to do lots of simple, local testing with different ray functions and it’s quite blocking to not have it work locally

2

ack, unfortunately we don’t have access on m4 right now but are looking into it!

3

I tried to reproduce on MacOS with M4. Here’s the output:

sysctl -a  | grep cpu
kern.sched_rt_avoid_cpu0: 0
...
machdep.cpu.brand_string: Apple M4 Pro

>>> import ray
>>> ray.init()
2025-03-10 14:20:58,668	INFO worker.py:1841 -- Started a local Ray instance.
RayContext(dashboard_url='', python_version='3.9.21', ray_version='2.42.1', ray_commit='c2e38f7b75be223c0c033986472daada8622d64f')
>>> ray.__version__
'2.42.1'

The OS version is: 15.3.1 (24D70)

Can you provide any additional information to reproduce the issue?

4 
❯ sysctl -a  | grep cpu
kern.sched_rt_avoid_cpu0: 0
...
machdep.cpu.brand_string: Apple M4 Pro

Actual examples of hanging:

❯ uv run ipython
Python 3.10.15 (main, Oct 16 2024, 08:33:15) [Clang 18.1.8 ]
Type 'copyright', 'credits' or 'license' for more information
IPython 8.32.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: import ray

In [2]: ray.init()
2025-03-10 21:39:25,501	INFO worker.py:1832 -- Started a local Ray instance. View the dashboard at 127.0.0.1:8265
Out[2]: RayContext(dashboard_url='127.0.0.1:8265', python_version='3.10.15', ray_version='2.42.0', ray_commit='637116a090c052d061af5ba3bef8a467c8c3fc25')


❯ uv run ipython
Python 3.10.15 (main, Oct 16 2024, 08:33:15) [Clang 18.1.8 ]
Type 'copyright', 'credits' or 'license' for more information
IPython 8.32.0 -- An enhanced Interactive Python. Type '?' for help.

In [1]: from ray.data import read_parquet

In [2]: ds = read_parquet("s3://bucket/prefix/file.parquet").materialize()
2025-03-10 21:47:35,970	INFO worker.py:1832 -- Started a local Ray instance. View the dashboard at 127.0.0.1:8265
Parquet Files Sample 0: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 1.00/1.00 [00:01<00:00, 1.65s/ file]
2025-03-10 21:47:38,288	INFO streaming_executor.py:108 -- Starting execution of Dataset. Full logs are in /tmp/ray/session_2025-03-10_21-47-35_102794_79906/logs/ray-data
2025-03-10 21:47:38,288	INFO streaming_executor.py:109 -- Execution plan of Dataset: InputDataBuffer[Input] -> TaskPoolMapOperator[ReadParquet]
✔️  Dataset execution finished in 17.23 seconds: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████| 9.92k/9.92k [00:17<00:00, 576 row/s]
- ReadParquet->SplitBlocks(67): Tasks: 0; Queued blocks: 0; Resources: 0.0 CPU, 17.5MB object store: : 9.92k row [00:17, 576 row/s]

They all seem to hang consistently on any ray ops I try. Here is my OS build:

❯ sw_vers
ProductName:		macOS
ProductVersion:		15.2
BuildVersion:		24C101

I will mention I was not having this issue previously, it only showed up recently so I’m not sure what I could have done to my Mac to create the problem