Merge pull request #240 from trails-cool/feat/overpass-metrics

Add Prometheus metrics + Grafana panels for Overpass proxy

apps/planner/app/lib/metrics.server.ts

@@ -41,4 +41,35 @@ export const brouterRequestDuration = getOrCreate("brouter_request_duration_seco
   }),
 );
 
+export const overpassCacheEvents = getOrCreate("overpass_cache_events_total", () =>
+  new client.Counter({
+    name: "overpass_cache_events_total",
+    help: "Overpass proxy cache events",
+    labelNames: ["result"] as const, // hit | miss | coalesced
+  }),
+);
+
+export const overpassCacheSize = getOrCreate("overpass_cache_size", () =>
+  new client.Gauge({
+    name: "overpass_cache_size",
+    help: "Current number of entries in the Overpass proxy cache",
+  }),
+);
+
+export const overpassUpstreamDuration = getOrCreate("overpass_upstream_duration_seconds", () =>
+  new client.Histogram({
+    name: "overpass_upstream_duration_seconds",
+    help: "Duration of upstream Overpass API requests in seconds",
+    buckets: [0.05, 0.1, 0.25, 0.5, 1, 2.5, 5, 10],
+  }),
+);
+
+export const overpassUpstreamRequests = getOrCreate("overpass_upstream_requests_total", () =>
+  new client.Counter({
+    name: "overpass_upstream_requests_total",
+    help: "Upstream Overpass API requests by status",
+    labelNames: ["status"] as const,
+  }),
+);
+
 export const registry = client.register;

apps/planner/app/routes/api.overpass.ts

@@ -1,5 +1,11 @@
 import type { Route } from "./+types/api.overpass";
 import { checkRateLimit } from "~/lib/rate-limit";
+import {
+  overpassCacheEvents,
+  overpassCacheSize,
+  overpassUpstreamDuration,
+  overpassUpstreamRequests,
+} from "~/lib/metrics.server";
 
 const UPSTREAM_URL = process.env.OVERPASS_URL ?? "https://overpass.private.coffee/api/interpreter";
 const USER_AGENT = "trails.cool Planner (https://trails.cool; legal@trails.cool)";
@@ -36,6 +42,7 @@ function putInCache(key: string, body: string, contentType: string) {
     cache.delete(oldest);
   }
   cache.set(key, { body, contentType, expiresAt: Date.now() + CACHE_TTL_MS });
+  overpassCacheSize.set(cache.size);
 }
 
 export async function action({ request }: Route.ActionArgs) {
@@ -56,6 +63,7 @@ export async function action({ request }: Route.ActionArgs) {
   // Cache hit: skip rate limit and upstream entirely
   const cached = getFromCache(cacheKey);
   if (cached) {
+    overpassCacheEvents.inc({ result: "hit" });
     return new Response(cached.body, {
       status: 200,
       headers: { "Content-Type": cached.contentType, "X-Cache": "hit" },
@@ -75,7 +83,9 @@ export async function action({ request }: Route.ActionArgs) {
   // same session don't each hit upstream for the identical bbox.
   let pending = inFlight.get(cacheKey);
   if (!pending) {
+    overpassCacheEvents.inc({ result: "miss" });
     pending = (async () => {
+      const start = Date.now();
       try {
         const upstream = await fetch(UPSTREAM_URL, {
           method: "POST",
@@ -85,6 +95,8 @@ export async function action({ request }: Route.ActionArgs) {
           },
           body,
         });
+        overpassUpstreamDuration.observe((Date.now() - start) / 1000);
+        overpassUpstreamRequests.inc({ status: String(upstream.status) });
         const responseBody = await upstream.text();
         const contentType = upstream.headers.get("content-type") ?? "application/json";
         const cacheable = upstream.status === 200 && !responseBody.includes("rate_limited");
@@ -94,12 +106,15 @@ export async function action({ request }: Route.ActionArgs) {
       }
     })();
     inFlight.set(cacheKey, pending);
+  } else {
+    overpassCacheEvents.inc({ result: "coalesced" });
   }
 
   let result;
   try {
     result = await pending;
   } catch {
+    overpassUpstreamRequests.inc({ status: "error" });
     return new Response("Upstream Overpass unavailable", { status: 502 });
   }
 

infrastructure/grafana/dashboards/planner.json

@@ -97,6 +97,168 @@
           "legendFormat": "{{route}}"
         }
       ]
+    },
+    {
+      "title": "Overpass Upstream Health (5m success rate)",
+      "type": "stat",
+      "gridPos": {
+        "h": 5,
+        "w": 6,
+        "x": 0,
+        "y": 14
+      },
+      "targets": [
+        {
+          "expr": "sum(rate(overpass_upstream_requests_total{status=~\"2..\"}[5m])) / clamp_min(sum(rate(overpass_upstream_requests_total[5m])), 1)",
+          "legendFormat": "success"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "percentunit",
+          "min": 0,
+          "max": 1,
+          "thresholds": {
+            "mode": "absolute",
+            "steps": [
+              { "color": "red", "value": null },
+              { "color": "orange", "value": 0.9 },
+              { "color": "green", "value": 0.99 }
+            ]
+          }
+        }
+      }
+    },
+    {
+      "title": "Overpass Cache Hit Ratio (5m)",
+      "type": "stat",
+      "gridPos": {
+        "h": 5,
+        "w": 6,
+        "x": 6,
+        "y": 14
+      },
+      "targets": [
+        {
+          "expr": "sum(rate(overpass_cache_events_total{result=\"hit\"}[5m])) / clamp_min(sum(rate(overpass_cache_events_total[5m])), 1)",
+          "legendFormat": "hit ratio"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "percentunit",
+          "min": 0,
+          "max": 1
+        }
+      }
+    },
+    {
+      "title": "Overpass Cache Size",
+      "type": "stat",
+      "gridPos": {
+        "h": 5,
+        "w": 6,
+        "x": 12,
+        "y": 14
+      },
+      "targets": [
+        {
+          "expr": "overpass_cache_size",
+          "legendFormat": "entries"
+        }
+      ]
+    },
+    {
+      "title": "Overpass Upstream p95",
+      "type": "stat",
+      "gridPos": {
+        "h": 5,
+        "w": 6,
+        "x": 18,
+        "y": 14
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.95, sum(rate(overpass_upstream_duration_seconds_bucket[5m])) by (le))",
+          "legendFormat": "p95"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      }
+    },
+    {
+      "title": "Overpass Cache Events",
+      "type": "timeseries",
+      "gridPos": {
+        "h": 7,
+        "w": 12,
+        "x": 0,
+        "y": 19
+      },
+      "targets": [
+        {
+          "expr": "sum by (result) (rate(overpass_cache_events_total[5m]))",
+          "legendFormat": "{{result}}"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops"
+        }
+      }
+    },
+    {
+      "title": "Overpass Upstream Status",
+      "type": "timeseries",
+      "gridPos": {
+        "h": 7,
+        "w": 12,
+        "x": 12,
+        "y": 19
+      },
+      "targets": [
+        {
+          "expr": "sum by (status) (rate(overpass_upstream_requests_total[5m]))",
+          "legendFormat": "{{status}}"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "ops"
+        }
+      }
+    },
+    {
+      "title": "Overpass Upstream Latency",
+      "type": "timeseries",
+      "gridPos": {
+        "h": 7,
+        "w": 24,
+        "x": 0,
+        "y": 26
+      },
+      "targets": [
+        {
+          "expr": "histogram_quantile(0.50, sum(rate(overpass_upstream_duration_seconds_bucket[5m])) by (le))",
+          "legendFormat": "p50"
+        },
+        {
+          "expr": "histogram_quantile(0.95, sum(rate(overpass_upstream_duration_seconds_bucket[5m])) by (le))",
+          "legendFormat": "p95"
+        },
+        {
+          "expr": "histogram_quantile(0.99, sum(rate(overpass_upstream_duration_seconds_bucket[5m])) by (le))",
+          "legendFormat": "p99"
+        }
+      ],
+      "fieldConfig": {
+        "defaults": {
+          "unit": "s"
+        }
+      }
     }
   ]
 }