Skip to content

fix(amd): OpenClaw schema + dashboard-api health for Lemonade#591

Merged
Lightheartdevs merged 1 commit intomainfrom
fix/amd-openclaw-dashboard-health
Mar 24, 2026
Merged

fix(amd): OpenClaw schema + dashboard-api health for Lemonade#591
Lightheartdevs merged 1 commit intomainfrom
fix/amd-openclaw-dashboard-health

Conversation

@Lightheartdevs
Copy link
Collaborator

Summary

  • Issue 12: Update openclaw-strix-halo.json to OpenClaw 2026.3.8 schema. Old schema keys (agent, providers, subagent, version, gateway.host) replaced with new schema (agents.defaults, models.providers, agents.defaults.subagents). Routes through LiteLLM (litellm:4000/v1) instead of direct llama-server:8080/api/v1. Adds __LITELLM_KEY__ placeholder resolved by sed at install time.

  • Issue 2b: Dashboard-api overrides llama-server health path from /health to /api/v1/health when LLM_BACKEND=lemonade. Also fixes get_loaded_model() to query /api/v1/models instead of /v1/models. Metrics and props endpoints stay at root (llama.cpp-specific, gracefully handled if unavailable).

Files changed (4)

  • config/openclaw/openclaw-strix-halo.json — new OpenClaw 2026.3.8 schema
  • extensions/services/dashboard-api/config.py — Lemonade health path override
  • extensions/services/dashboard-api/helpers.py_LLM_API_PREFIX for model queries
  • installers/phases/06-directories.sh — sed replacement for __LITELLM_KEY__

Non-AMD paths: zero change

  • LLM_BACKEND defaults to empty string — health override doesn't fire
  • _LLM_API_PREFIX defaults to /v1 — same as before
  • Other OpenClaw configs (openclaw.json, pro.json) untouched

Test plan

  • Fresh AMD install: dashboard shows llama-server as healthy
  • Fresh AMD install: OpenClaw starts without crash-loop
  • Fresh AMD install: dashboard inference metrics populate
  • Fresh NVIDIA install: zero behavioral change

🤖 Generated with Claude Code

Issue 12: Update openclaw-strix-halo.json to OpenClaw 2026.3.8 schema.
Replaces deprecated keys (agent, root-level providers, subagent, version,
gateway.host) with new schema (agents.defaults, models.providers, etc.).
Routes through LiteLLM instead of direct llama-server. Adds __LITELLM_KEY__
placeholder resolved at install time.

Issue 2b: Dashboard-api health check overrides for Lemonade backend.
When LLM_BACKEND=lemonade, overrides llama-server health path from /health
to /api/v1/health in config.py. Adds _LLM_API_PREFIX in helpers.py so
get_loaded_model() queries /api/v1/models instead of /v1/models.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
@Lightheartdevs Lightheartdevs merged commit dcdf9b9 into main Mar 24, 2026
15 of 21 checks passed
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.

1 participant