mesh-llm/contrib/windows
Ivan Golovach 07b9dea178
Add split readiness doctor (#723)
Validation
* Validation tier: Tier 4 - split-readiness verification tooling, read-only management diagnostics, Windows capture helper, and post-review diagnostic correctness repair; no routing, wire protocol, or model-load behavior change.
* git fetch --no-tags origin main:refs/remotes/origin/main: PASS, origin/main at 6f6d606c61.
* git rebase origin/main: PASS, no conflicts.
* git diff --check: PASS, no output.
* git diff --cached --check: PASS, no output.
* cargo fmt --all -- --check: PASS, no output.
* LLAMA_STAGE_BUILD_DIR=/Users/Funtland/Downloads/mesh-llm/.deps/llama-build/build-stage-abi-metal cargo test -p mesh-llm-host-runtime split_readiness --lib -- --test-threads=1: PASS, 9 passed.
* LLAMA_STAGE_BUILD_DIR=/Users/Funtland/Downloads/mesh-llm/.deps/llama-build/build-stage-abi-metal cargo check -p mesh-llm: PASS.
* LLAMA_STAGE_BUILD_DIR=/Users/Funtland/Downloads/mesh-llm/.deps/llama-build/build-stage-abi-metal /opt/homebrew/bin/cargo-clippy clippy -p mesh-llm-host-runtime --all-targets -- -D warnings: PASS.
* Ledger: not applicable - not required for selected validation tier/change family.
* Version: not applicable - diagnostics/tooling/docs only; no release/version sync required.
* Not run: pwsh syntax/help validation - pwsh is not installed in this environment.
* Not run: live two-node split smoke - no local split-capable multi-node runtime/model endpoint was available; deterministic split-readiness tests cover the diagnostic classification and capture path.

Rollback
* git revert HEAD
2026-05-29 10:44:16 +10:00
..
CollectSplitDiagnostics.ps1 Add split readiness doctor (#723) 2026-05-29 10:44:16 +10:00
README.md Add split readiness doctor (#723) 2026-05-29 10:44:16 +10:00
StartChat.ps1 Fix Windows mesh startup compatibility (#448) 2026-05-06 21:43:52 +10:00
StartMeshServer.ps1 Fix Windows mesh startup compatibility (#448) 2026-05-06 21:43:52 +10:00

Windows helpers

These optional PowerShell helpers wrap a local Windows build of mesh-llm.

Build first from the repository root:

just build backend=vulkan

Start a local server:

.\contrib\windows\StartMeshServer.ps1 -Model Qwen2.5-3B-Instruct-Q4_K_M -Device Vulkan1

Chat with that server:

.\contrib\windows\StartChat.ps1 -Model Qwen2.5-3B-Instruct-Q4_K_M

Collect split diagnostics from one or more already-running nodes:

.\contrib\windows\CollectSplitDiagnostics.ps1 `
  -Model meshllm/Qwen3-8B-Q4_K_M-layers `
  -ConsoleUrls http://127.0.0.1:3131 `
  -ApiUrls http://127.0.0.1:9337/v1

The collector writes a timestamped folder and zip containing redacted management API payloads, /v1/models, GPU/process facts, optional probe results, and recent skippy-native.log tails. It attaches to running nodes; it does not start or stop mesh processes.

The scripts default to target\release\mesh-llm.exe when it exists, otherwise they fall back to mesh-llm on PATH.