Skip to content

Comments

Fixes in mlx.distributed_config#2947

Merged
angeloskath merged 2 commits intomainfrom
dist-config-fix
Dec 23, 2025
Merged

Fixes in mlx.distributed_config#2947
angeloskath merged 2 commits intomainfrom
dist-config-fix

Conversation

@angeloskath
Copy link
Member

@angeloskath angeloskath commented Dec 22, 2025

Fixes #2944.

In particular:

  • Imports log_warning that was forgotten
  • Fetches the ip from en1 if en0 is not connected.

@angeloskath angeloskath changed the title Import log_warning Fixes in mlx.distributed_config Dec 22, 2025
Copy link
Member

@awni awni left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🙏

continue

ip = run(
["ssh", h.ssh_hostname, "ipconfig", "getifaddr", "en1"],

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@angeloskath

in my case im using ethernet->usbc adapter en1 will return nothing. I wonder if possible to iterate over enX for first match or maybe better provide hint as flag like --enX 8

sumel@mini-1 ~ % ipconfig getifaddr en4
sumel@mini-1 ~ % ipconfig getifaddr en5
sumel@mini-1 ~ % ipconfig getifaddr en6
sumel@mini-1 ~ % ipconfig getifaddr en7
sumel@mini-1 ~ % ipconfig getifaddr en8
192.168.1.101

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Interesting I will think about adding a flag. The config is due for a refactor to simplify it further. However, at that point it is trivial to also edit the config manually since you only need the ip of rank 0.

@angeloskath angeloskath merged commit 9cfda1a into main Dec 23, 2025
14 checks passed
@angeloskath angeloskath deleted the dist-config-fix branch December 23, 2025 01:38
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[BUG] issues with RDMA and JACCL

3 participants