desktop: show where relay-mesh inference runs in the activity panel - #2205
Closed
michaelneale wants to merge 1 commit into
Closed
desktop: show where relay-mesh inference runs in the activity panel#2205michaelneale wants to merge 1 commit into
michaelneale wants to merge 1 commit into
Conversation
Buzz shared compute agents gave no indication of where their inference actually executes. Add a small indicator to the Activity panel header, e.g. 'Shared compute · Qwen3 8B · 3 nodes', with a full-sentence tooltip. - New read-only mesh_availability Tauri command exposing the existing member-verified, freshness-filtered MeshAvailability (serve targets + models) that discovery already computes for routing. Stubbed for non-mesh builds. - meshInferenceLocation.ts derives the label from availability + the agent's model ref: distinct node counting (deviceId/endpointId, so one node serving two models counts once) and Rust-parity @main-revision model matching. Explicit / auto-routed / no-live-nodes states. - useMeshInferenceLocation polls availability (60s vs the 45s relay heartbeat) only when the agent's provider is relay-mesh; returns null for non-mesh agents and while availability is unknown, so the header renders nothing rather than a guess. - E2E mock bridge answers mesh_availability with the matching shape.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Buzz shared compute (relay-mesh) agents gave no indication of where their inference actually executes. This adds a small indicator to the Activity panel header, next to the scope label:
Hovering shows the full sentence: "This agent runs inference with Qwen3 8B on 3 nodes serving on this relay via Buzz shared compute."
States:
Shared compute · Qwen3 8B · 3 nodesShared compute · auto-routed · 2 nodesShared compute · no live serving nodesNon-mesh agents render nothing; while availability is loading it also renders nothing rather than guessing.
How
mesh_availabilityTauri command — exposes the existing member-verified, freshness-filteredMeshAvailability(serve targets + models) that discovery already computes for routing. No new trust surface. Stubbed for non-mesh-llmbuilds.meshInferenceLocation.ts— pure derivation from availability + the agent's model ref. Counts distinct nodes (bydeviceId/endpointId, so one node serving two models counts once) and mirrors the Rust@main-revision canonical model matching frompick_serve_target_for_model.useMeshInferenceLocation— resolves the managed agent record; polls availability (60s, vs the 45s relay status heartbeat) only when the provider isrelay-mesh.mesh_availabilitywith the matching shape.Testing
meshInferenceLocation.test.mjs); full desktop suite 3259 passtsc --noEmit, biome,cargo clippy --features mesh-llmcleanmesh-llm); all 60 mesh Rust tests pass