Merge branch 'master' into fix/cache-hit-rate

The flat 'deploy' concurrency group caused ALL PRs to share one queue,
so pushing to any PR would cancel CI runs on other PRs.

Changed to deploy-${{ github.event.pull_request.number || github.ref }}
so each PR gets its own concurrency group while re-pushes to the same
PR still cancel the previous run.

.github/workflows/deploy.yml

@@ -17,7 +17,7 @@ on:
       - 'docs/**'
 
 concurrency:
-  group: deploy
+  group: deploy-${{ github.event.pull_request.number || github.ref }}
   cancel-in-progress: true
 
 env:

cmd/server/routes.go

@@ -33,6 +33,11 @@ type Server struct {
 	memStatsMu   sync.Mutex
 	memStatsCache runtime.MemStats
 	memStatsCachedAt time.Time
+
+	// Cached /api/stats response — recomputed at most once every 10s
+	statsMu      sync.Mutex
+	statsCache   *StatsResponse
+	statsCachedAt time.Time
 }
 
 // PerfStats tracks request performance.
@@ -380,6 +385,17 @@ func (s *Server) handleHealth(w http.ResponseWriter, r *http.Request) {
 }
 
 func (s *Server) handleStats(w http.ResponseWriter, r *http.Request) {
+	const statsTTL = 10 * time.Second
+
+	s.statsMu.Lock()
+	if s.statsCache != nil && time.Since(s.statsCachedAt) < statsTTL {
+		cached := s.statsCache
+		s.statsMu.Unlock()
+		writeJSON(w, cached)
+		return
+	}
+	s.statsMu.Unlock()
+
 	var stats *Stats
 	var err error
 	if s.store != nil {
@@ -392,7 +408,7 @@ func (s *Server) handleStats(w http.ResponseWriter, r *http.Request) {
 		return
 	}
 	counts := s.db.GetRoleCounts()
-	writeJSON(w, StatsResponse{
+	resp := &StatsResponse{
 		TotalPackets:       stats.TotalPackets,
 		TotalTransmissions: &stats.TotalTransmissions,
 		TotalObservations:  stats.TotalObservations,
@@ -411,7 +427,14 @@ func (s *Server) handleStats(w http.ResponseWriter, r *http.Request) {
 			Companions: counts["companions"],
 			Sensors:    counts["sensors"],
 		},
-	})
+	}
+
+	s.statsMu.Lock()
+	s.statsCache = resp
+	s.statsCachedAt = time.Now()
+	s.statsMu.Unlock()
+
+	writeJSON(w, resp)
 }
 
 func (s *Server) handlePerf(w http.ResponseWriter, r *http.Request) {

cmd/server/store.go

@@ -98,6 +98,11 @@ type PacketStore struct {
 	// computed during Load() and incrementally updated on ingest.
 	distHops  []distHopRecord
 	distPaths []distPathRecord
+
+	// Cached GetNodeHashSizeInfo result — recomputed at most once every 15s
+	hashSizeInfoMu    sync.Mutex
+	hashSizeInfoCache map[string]*hashSizeNodeInfo
+	hashSizeInfoAt    time.Time
 }
 
 // Precomputed distance records for fast analytics aggregation.
@@ -3722,8 +3727,26 @@ type hashSizeNodeInfo struct {
 	Inconsistent bool
 }
 
-// GetNodeHashSizeInfo scans advert packets to compute per-node hash size data.
+// GetNodeHashSizeInfo returns cached per-node hash size data, recomputing at most every 15s.
 func (s *PacketStore) GetNodeHashSizeInfo() map[string]*hashSizeNodeInfo {
+	const ttl = 15 * time.Second
+	s.hashSizeInfoMu.Lock()
+	if s.hashSizeInfoCache != nil && time.Since(s.hashSizeInfoAt) < ttl {
+		cached := s.hashSizeInfoCache
+		s.hashSizeInfoMu.Unlock()
+		return cached
+	}
+	s.hashSizeInfoMu.Unlock()
+	result := s.computeNodeHashSizeInfo()
+	s.hashSizeInfoMu.Lock()
+	s.hashSizeInfoCache = result
+	s.hashSizeInfoAt = time.Now()
+	s.hashSizeInfoMu.Unlock()
+	return result
+}
+
+// computeNodeHashSizeInfo scans advert packets to compute per-node hash size data.
+func (s *PacketStore) computeNodeHashSizeInfo() map[string]*hashSizeNodeInfo {
 	s.mu.RLock()
 	defer s.mu.RUnlock()
 

docker-compose.yml

@@ -5,9 +5,12 @@
 
 services:
   prod:
+    build: .
     image: corescope:latest
     container_name: corescope-prod
     restart: unless-stopped
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
     ports:
       - "${PROD_HTTP_PORT:-80}:${PROD_HTTP_PORT:-80}"
       - "${PROD_HTTPS_PORT:-443}:${PROD_HTTPS_PORT:-443}"
@@ -26,9 +29,12 @@ services:
       retries: 3
 
   staging:
+    build: .
     image: corescope:latest
     container_name: corescope-staging
     restart: unless-stopped
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
     ports:
       - "${STAGING_HTTP_PORT:-81}:${STAGING_HTTP_PORT:-81}"
       - "${STAGING_MQTT_PORT:-1884}:1883"
@@ -57,6 +63,8 @@ services:
     image: corescope-go:latest
     container_name: corescope-staging-go
     restart: unless-stopped
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
     ports:
       - "${STAGING_GO_HTTP_PORT:-82}:80"
       - "${STAGING_GO_MQTT_PORT:-1885}:1883"

server.js

@@ -207,6 +207,13 @@ class TTLCache {
       if (key.startsWith(prefix)) this.store.delete(key);
     }
   }
+  debouncedInvalidateBulkHealth() {
+    if (this._bulkHealthTimer) return;
+    this._bulkHealthTimer = setTimeout(() => {
+      this._bulkHealthTimer = null;
+      this.invalidate('bulk-health');
+    }, 30000);
+  }
   debouncedInvalidateAll() {
     if (this._debounceTimer) return;
     this._debounceTimer = setTimeout(() => {
@@ -410,7 +417,7 @@ app.get('/api/perf', (req, res) => {
     avgMs: perfStats.requests ? Math.round(perfStats.totalMs / perfStats.requests * 10) / 10 : 0,
     endpoints: Object.fromEntries(sorted),
     slowQueries: perfStats.slowQueries.slice(-20),
-    cache: { size: cache.size, hits: cache.hits, misses: cache.misses, staleHits: cache.staleHits, recomputes: cache.recomputes, hitRate: cache.hits + cache.misses > 0 ? Math.round(cache.hits / (cache.hits + cache.misses) * 1000) / 10 : 0 },
+    cache: { size: cache.size, hits: cache.hits, misses: cache.misses, staleHits: cache.staleHits, recomputes: cache.recomputes, hitRate: cache.hits + cache.staleHits + cache.misses > 0 ? Math.round((cache.hits + cache.staleHits) / (cache.hits + cache.staleHits + cache.misses) * 1000) / 10 : 0 },
     packetStore: pktStore.getStats(),
     sqlite: (() => {
       try {
@@ -519,7 +526,7 @@ app.get('/api/health', (req, res) => {
       misses: cache.misses,
       staleHits: cache.staleHits,
       recomputes: cache.recomputes,
-      hitRate: cache.hits + cache.misses > 0 ? Math.round(cache.hits / (cache.hits + cache.misses) * 1000) / 10 : 0,
+      hitRate: cache.hits + cache.staleHits + cache.misses > 0 ? Math.round((cache.hits + cache.staleHits) / (cache.hits + cache.staleHits + cache.misses) * 1000) / 10 : 0,
     },
     websocket: {
       clients: wsClients,
@@ -723,7 +730,7 @@ for (const source of mqttSources) {
             // Invalidate this node's caches on advert
             cache.invalidate('node:' + p.pubKey);
             cache.invalidate('health:' + p.pubKey);
-            cache.invalidate('bulk-health');
+            cache.debouncedInvalidateBulkHealth();
 
             // Cross-reference: if this node's pubkey matches an existing observer, backfill observer name
             if (p.name && p.pubKey) {

test-server-routes.js

@@ -1254,6 +1254,24 @@ seedTestData();
     lastPathSeenMap.delete(liveNode);
   });
 
+  // ── Cache hit rate includes stale hits ──
+  await t('Cache hitRate includes staleHits in formula', async () => {
+    cache.clear();
+    cache.hits = 0;
+    cache.misses = 0;
+    cache.staleHits = 0;
+    // Simulate: 3 hits, 2 stale hits, 5 misses => rate = (3+2)/(3+2+5) = 50%
+    cache.hits = 3;
+    cache.staleHits = 2;
+    cache.misses = 5;
+    const r = await request(app).get('/api/health').expect(200);
+    assert(r.body.cache.hitRate === 50, 'hitRate should be (hits+staleHits)/(hits+staleHits+misses) = 50%, got ' + r.body.cache.hitRate);
+    // Reset
+    cache.hits = 0;
+    cache.misses = 0;
+    cache.staleHits = 0;
+  });
+
   // ── Summary ──
   console.log(`\n═══ Server Route Tests: ${passed} passed, ${failed} failed ═══`);
   if (failed > 0) process.exit(1);