Skip to content

Conversation

dulinriley
Copy link
Contributor

@dulinriley dulinriley commented Sep 18, 2025

Summary:
Part of: #1209

Use casting to implement the supervision_events API instead of iterating over all
ProcMeshAgents. This will scale better as the size of the ProcMesh increases.
This requires making some trait bound changes to resource::GetState so that it can be casted with.

Unfortunately, the Actor name for the mesh agent is not compatible with the v1::Name struct
due to the missing uuid. Make v1::Name an enum to allow reserved names to be used for
things like ActorMeshes.

Also, minor improvement: make ActorMeshRef::supervision_events not take a Name, we can assume
it is for the current mesh's name.

Differential Revision: D82687236

@meta-cla meta-cla bot added the CLA Signed This label is managed by the Meta Open Source bot. label Sep 18, 2025
@facebook-github-bot
Copy link
Contributor

@dulinriley has exported this pull request. If you are a Meta employee, you can view the originating diff in D82687236.

Summary:

Part of: meta-pytorch#1209

Use casting to implement the `supervision_events` API instead of iterating over all
ProcMeshAgents. This will scale better as the size of the ProcMesh increases.
This requires making some trait bound changes to `resource::GetState` so that it can be casted with.

Unfortunately, the Actor name for the mesh agent is not compatible with the v1::Name struct
due to the missing uuid. Make `v1::Name` an enum to allow reserved names to be used for
things like ActorMeshes.

Also, minor improvement: make ActorMeshRef::supervision_events not take a Name, we can assume
it is for the current mesh's name.

Differential Revision: D82687236
@facebook-github-bot
Copy link
Contributor

@dulinriley has exported this pull request. If you are a Meta employee, you can view the originating diff in D82687236.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
CLA Signed This label is managed by the Meta Open Source bot. fb-exported meta-exported
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants