feat(admin): add observability dashboard UI with Grafana embeds

.changeset/admin-dashboard-ui.md

@@ -0,0 +1,5 @@
+---
+"nostream": minor
+---
+
+feat: add admin observability dashboard with Grafana embed and provisioned metrics panels

.env.example

@@ -41,6 +41,16 @@ WORKER_COUNT=2 # Defaults to CPU count. Use 1 or 2 for local testing.
 # PROMETHEUS_URL=http://127.0.0.1:9090
 # PROMETHEUS_QUERY_TIMEOUT_MS=5000
 # ADMIN_METRICS_SSE_INTERVAL_MS=5000
+# ADMIN_METRICS_SNAPSHOT_TIMEOUT_MS=10000
+# ADMIN_DEPENDENCY_PING_TIMEOUT_MS=3000
+
+# --- ADMIN CONSOLE (opt-in; disabled by default in default-settings.yaml) ---
+# ADMIN_PASSWORD=dev-admin-password
+# Set admin.enabled: true in .nostr/settings.yaml to enable the console
+# GRAFANA_URL=http://127.0.0.1:3000
+# GRAFANA_DASHBOARD_UID=nostream-overview
+# GRAFANA_ADMIN_USER=admin
+# GRAFANA_ADMIN_PASSWORD=admin
 
 # --- RELAY PRIVATE KEY (Optional) ---
 # RELAY_PRIVATE_KEY=your_hex_private_key

CONFIGURATION.md

@@ -54,6 +54,11 @@ The following environment variables can be set:
 | REDIS_PORT                       | Redis Port                       | 6379                   |
 | REDIS_USER                       | Redis User                       | default                |
 | REDIS_PASSWORD                   | Redis Password                   | nostr_ts_relay         |
+| PROMETHEUS_URL                   | Prometheus base URL for admin metrics queries | http://127.0.0.1:9090 |
+| PROMETHEUS_QUERY_TIMEOUT_MS      | Timeout for each Prometheus admin metrics query (ms) | 5000 |
+| ADMIN_METRICS_SSE_INTERVAL_MS    | Interval between `/admin/metrics` SSE snapshots (ms) | 5000 |
+| ADMIN_METRICS_SNAPSHOT_TIMEOUT_MS| Timeout for collecting one `/admin/metrics` snapshot (ms) | 10000 |
+| ADMIN_DEPENDENCY_PING_TIMEOUT_MS | Timeout for admin DB/Redis dependency pings (ms) | 3000 |
 | NOSTR_CONFIG_DIR                 | Configuration directory          | <project_root>/.nostr/ |
 | DEBUG                            | Debugging filter                 |                        |
 | ZEBEDEE_API_KEY                  | Zebedee Project API Key          |                        |

docker-compose.yml

@@ -179,6 +179,29 @@ services:
     networks:
       default:
 
+  grafana:
+    image: grafana/grafana:11.2.0
+    container_name: grafana
+    environment:
+      GF_SECURITY_ADMIN_USER: ${GRAFANA_ADMIN_USER:-admin}
+      GF_SECURITY_ADMIN_PASSWORD: ${GRAFANA_ADMIN_PASSWORD:-admin}
+      GF_AUTH_ANONYMOUS_ENABLED: 'true'
+      GF_AUTH_ANONYMOUS_ORG_ROLE: Viewer
+      GF_SECURITY_ALLOW_EMBEDDING: 'true'
+      GF_SERVER_ROOT_URL: ${GRAFANA_URL:-http://127.0.0.1:3000}
+    volumes:
+      - ./grafana/provisioning:/etc/grafana/provisioning:ro
+      - ./grafana/dashboards:/var/lib/grafana/dashboards:ro
+      - grafana-data:/var/lib/grafana
+    ports:
+      - 127.0.0.1:3000:3000
+    depends_on:
+      prometheus:
+        condition: service_started
+    restart: always
+    networks:
+      default:
+
 networks:
   default:
     name: nostream
@@ -190,3 +213,4 @@ networks:
 volumes:
   cache:
   prometheus-data:
+  grafana-data:

docs/REDIS.md

@@ -11,6 +11,8 @@ Nostream needs.
 
 Rate limiting is implemented using a sliding window strategy, which uses Redis sorted sets to track request timestamps. This allows Nostream to accurately enforce rate limits over a rolling time window rather than a fixed one, preventing clients from bursting requests at window boundaries.
 
+When Redis is temporarily unavailable, Nostream keeps serving the relay and admin HTTP endpoints and treats rate-limiter Redis failures as fail-open (events are not blocked solely because Redis is down). Admin health degrades accordingly (`redis.ok=false`), so operators can detect and recover Redis without taking the relay offline.
+
 ## Requirements
 
 - Redis 6.0 or higher (for ACL support)