From 88188ed693bf630e752fc74af3e22c05b0979694 Mon Sep 17 00:00:00 2001
From: Apple <apple@MacBook-Pro.local>
Date: Tue, 2 Dec 2025 07:02:08 -0800
Subject: [PATCH] fix(node2): Use node_cache router_healthy for DAGI Router
 agents status

- Fix get_dagi_router_agents to use router_healthy from node_cache first
- Fallback to direct API call only if cache is unavailable
- This fixes NODE2 agents showing as 'stale' when router is actually healthy
- Fix CITY_SERVICE_URL in scripts (remove /api/city, use /api)
---
 scripts/start-node2-guardian.sh      | 13 ++++++++-
 services/city-service/routes_city.py | 41 ++++++++++++++++++----------
 2 files changed, 38 insertions(+), 16 deletions(-)

diff --git a/scripts/start-node2-guardian.sh b/scripts/start-node2-guardian.sh
index 7e516590..f61c29c8 100755
--- a/scripts/start-node2-guardian.sh
+++ b/scripts/start-node2-guardian.sh
@@ -26,7 +26,7 @@ export NODE_NAME="НОДА2"
 export NODE_ENVIRONMENT="development"
 export NODE_ROLES="gpu,ai_runtime"
 export NODE_HOSTNAME="$(hostname)"
-export CITY_SERVICE_URL="https://daarion.space/api/city"
+export CITY_SERVICE_URL="https://daarion.space/api"
 export NODE_SWAPPER_URL="http://localhost:8890"
 export NODE_ROUTER_URL="http://localhost:9102"
 export GUARDIAN_INTERVAL="60"
@@ -38,6 +38,17 @@ echo "   Swapper URL: $NODE_SWAPPER_URL"
 echo "   Router URL: $NODE_ROUTER_URL"
 echo ""
 
+# Export environment variables for node-guardian-loop
+export NODE_ID
+export NODE_NAME
+export NODE_ENVIRONMENT
+export NODE_ROLES
+export NODE_HOSTNAME
+export CITY_SERVICE_URL
+export NODE_SWAPPER_URL
+export NODE_ROUTER_URL
+export GUARDIAN_INTERVAL
+
 # Run node-guardian-loop
 python3 "$PROJECT_ROOT/scripts/node-guardian-loop.py" \
     --node-id "$NODE_ID" \
diff --git a/services/city-service/routes_city.py b/services/city-service/routes_city.py
index 31260e3e..beab72d1 100644
--- a/services/city-service/routes_city.py
+++ b/services/city-service/routes_city.py
@@ -4571,23 +4571,34 @@ async def get_dagi_router_agents(node_id: str):
     """
     import httpx
     
-    # Get router URL from database (node-specific)
-    endpoints = await repo_city.get_node_endpoints(node_id)
-    base_url = endpoints.get("router_url")
+    # First, try to get router health from node_cache (populated by node-guardian)
+    # This is preferred for remote nodes (like NODE2) where direct connection may not work
     router_healthy = False
+    try:
+        metrics = await repo_city.get_node_metrics(node_id)
+        if metrics and metrics.get("router_healthy") is not None:
+            router_healthy = metrics.get("router_healthy", False)
+            logger.debug(f"Using router_healthy from node_cache for {node_id}: {router_healthy}")
+    except Exception as e:
+        logger.debug(f"Failed to get cached router health for {node_id}: {e}")
     
-    # Check if router is healthy
-    if base_url:
-        try:
-            async with httpx.AsyncClient(timeout=3.0) as client:
-                resp = await client.get(f"{base_url}/health")
-            if resp.status_code == 200:
-                data = resp.json()
-                # Router can return "healthy" or "ok"
-                status = data.get("status", "").lower()
-                router_healthy = status in ("healthy", "ok")
-        except Exception as e:
-            logger.warning(f"Failed to check router health for {node_id} at {base_url}: {e}")
+    # Fallback: try direct health check (only works for local nodes like NODE1)
+    if router_healthy is False:
+        endpoints = await repo_city.get_node_endpoints(node_id)
+        base_url = endpoints.get("router_url")
+        
+        if base_url:
+            try:
+                async with httpx.AsyncClient(timeout=3.0) as client:
+                    resp = await client.get(f"{base_url}/health")
+                if resp.status_code == 200:
+                    data = resp.json()
+                    # Router can return "healthy" or "ok"
+                    status = data.get("status", "").lower()
+                    router_healthy = status in ("healthy", "ok")
+                    logger.debug(f"Direct router health check for {node_id}: {router_healthy}")
+            except Exception as e:
+                logger.debug(f"Failed to check router health for {node_id} at {base_url}: {e}")
     
     # Get agents from DB for this node
     try: