feat(dashboard): serve real recommendations from /api/health

mesh_reporter was never on app.state — add it alongside health_engine etc.
in server.py's existing pattern. mesh_routes.py's health endpoint now
returns mesh_reporter.recommendations_list("mesh") instead of a hardcoded
[] TODO stub.

Add recommendations_available: bool alongside recommendations: string[] so
"the engine ran and found nothing" (healthy mesh) is never indistinguishable
from "the engine couldn't run" (unwired reporter, or an exception — logged
via logger.exception and swallowed so the rest of the health response still
serves). A crashed recommendations engine must not read as "mesh is
healthy" to an operator.

Also delete main.py's phantom `getattr(mh, "recommendations", [])` on the
websocket health_update push: nothing ever set .recommendations on the
mesh_health object (always []), and no frontend consumer reads it — the
REST endpoint above is the supported path.
This commit is contained in:
Matt Johnson 2026-07-16 22:57:39 +00:00
commit 31d8a361d2
4 changed files with 157 additions and 2 deletions

View file

@ -1,11 +1,13 @@
"""Mesh health and node API routes."""
import logging
from datetime import datetime
from typing import Optional
from fastapi import APIRouter, HTTPException, Request
router = APIRouter(tags=["mesh"])
logger = logging.getLogger(__name__)
def _serialize_health_score(score) -> dict:
@ -68,6 +70,22 @@ async def get_health(request: Request):
health = health_engine.mesh_health
score = health.score
# `recommendations_available` distinguishes "the engine ran and found
# nothing" (empty list, mesh is genuinely healthy) from "the engine
# couldn't run" (unwired reporter or an exception) — the two must not
# look identical to the operator. See mesh_reporter.recommendations_list().
mesh_reporter = getattr(request.app.state, "mesh_reporter", None)
recommendations: list[str] = []
recommendations_available = True
if mesh_reporter:
try:
recommendations = mesh_reporter.recommendations_list("mesh")
except Exception:
logger.exception("mesh_reporter.recommendations_list failed")
recommendations_available = False
else:
recommendations_available = False
return {
"score": round(score.composite, 1),
"tier": score.tier,
@ -90,7 +108,8 @@ async def get_health(request: Request):
"total_regions": health.total_regions,
"unlocated_count": len(health.unlocated_nodes),
"last_computed": _format_timestamp(health.last_computed),
"recommendations": [], # TODO: Add recommendations
"recommendations": recommendations,
"recommendations_available": recommendations_available,
}

View file

@ -123,6 +123,7 @@ async def start_dashboard(meshai_instance: "MeshAI") -> DashboardBroadcaster:
app.state.config_path = meshai_instance.config._config_path
app.state.data_store = meshai_instance.data_store
app.state.health_engine = meshai_instance.health_engine
app.state.mesh_reporter = getattr(meshai_instance, "mesh_reporter", None)
app.state.alert_engine = getattr(meshai_instance, "alert_engine", None)
app.state.env_store = getattr(meshai_instance, "env_store", None)
app.state.notification_router = getattr(meshai_instance, "notification_router", None)

View file

@ -190,7 +190,6 @@ class MeshAI:
"total_regions": mh.total_regions,
"unlocated_count": getattr(mh, "unlocated_count", 0),
"last_computed": mh.last_computed,
"recommendations": getattr(mh, "recommendations", []),
}
await self.broadcaster.broadcast("health_update", health_dict)
except Exception as e:

View file

@ -0,0 +1,136 @@
"""API tests for GET /api/health's `recommendations` field.
Covers the dashboard-recommendations wiring: mesh_reporter is exposed on
app.state (mirroring the existing health_engine/data_store/etc. pattern in
dashboard/server.py) and mesh_routes.py's health endpoint now returns real
recommendations from MeshReporter.recommendations_list("mesh") instead of
the old hardcoded `[]`.
"""
from __future__ import annotations
from unittest.mock import MagicMock
import pytest
from fastapi import FastAPI
from fastapi.testclient import TestClient
from meshai.dashboard.api.mesh_routes import router
from meshai.mesh_health import HealthScore, MeshHealth
def _client(health_engine, mesh_reporter=None):
app = FastAPI()
app.include_router(router, prefix="/api")
app.state.health_engine = health_engine
app.state.mesh_reporter = mesh_reporter
return TestClient(app)
def _health_engine(mesh_health):
engine = MagicMock()
engine.mesh_health = mesh_health
return engine
def test_health_endpoint_returns_recommendations():
"""recommendations_list("mesh") output reaches the REST /api/health body,
flagged as `recommendations_available: True` (engine ran successfully)."""
mesh_health = MeshHealth(score=HealthScore())
engine = _health_engine(mesh_health)
reporter = MagicMock()
reporter.recommendations_list.return_value = [
"Coverage gap in TestRegion: 3 nodes only reach 1 gateway.",
"No MQTT uplinks in TestRegion. Enable on at least one infrastructure node.",
]
client = _client(engine, mesh_reporter=reporter)
r = client.get("/api/health")
assert r.status_code == 200
body = r.json()
assert body["recommendations"] == [
"Coverage gap in TestRegion: 3 nodes only reach 1 gateway.",
"No MQTT uplinks in TestRegion. Enable on at least one infrastructure node.",
]
assert body["recommendations_available"] is True
reporter.recommendations_list.assert_called_once_with("mesh")
def test_health_endpoint_empty_recommendations_is_marked_available():
"""A genuinely healthy mesh: empty list AND recommendations_available=True.
This is the "healthy" state it must be distinguishable from the
error/unwired states below, which also produce an empty list but with
recommendations_available=False.
"""
mesh_health = MeshHealth(score=HealthScore())
engine = _health_engine(mesh_health)
reporter = MagicMock()
reporter.recommendations_list.return_value = []
client = _client(engine, mesh_reporter=reporter)
r = client.get("/api/health")
assert r.status_code == 200
body = r.json()
assert body["recommendations"] == []
assert body["recommendations_available"] is True
def test_health_endpoint_no_mesh_reporter_configured():
"""mesh_reporter can be None (e.g. Meshtastic not configured) — no crash,
but this must NOT be indistinguishable from "healthy": empty list with
recommendations_available=False, not True.
"""
mesh_health = MeshHealth(score=HealthScore())
engine = _health_engine(mesh_health)
client = _client(engine, mesh_reporter=None)
r = client.get("/api/health")
assert r.status_code == 200
body = r.json()
assert body["recommendations"] == []
assert body["recommendations_available"] is False
def test_health_endpoint_recommendations_error_is_swallowed_but_flagged(caplog):
"""A raising mesh_reporter must not break the health endpoint (the other
fields are still useful), but the failure must be (a) logged, so it's
traceable, and (b) surfaced via recommendations_available=False, so the
UI never renders a crashed engine as "mesh is healthy"."""
mesh_health = MeshHealth(score=HealthScore())
engine = _health_engine(mesh_health)
reporter = MagicMock()
reporter.recommendations_list.side_effect = RuntimeError("boom")
client = _client(engine, mesh_reporter=reporter)
with caplog.at_level("ERROR"):
r = client.get("/api/health")
assert r.status_code == 200
body = r.json()
assert body["recommendations"] == []
assert body["recommendations_available"] is False
# The other fields on the response are unaffected by the recommendations
# failure — a 500 must not take down the whole health endpoint.
assert body["score"] == round(HealthScore().composite, 1)
assert body["tier"] == HealthScore().tier
assert any("recommendations_list failed" in rec.message for rec in caplog.records)
def test_health_endpoint_no_health_data_yet():
"""health_engine.mesh_health is None (not computed yet) — unaffected by recommendations wiring."""
engine = _health_engine(None)
reporter = MagicMock()
client = _client(engine, mesh_reporter=reporter)
r = client.get("/api/health")
assert r.status_code == 200
body = r.json()
assert body["message"] == "Health engine not ready"
reporter.recommendations_list.assert_not_called()