feat(notifications): production-ready WebSocket gateway (TEC-2766)

- Add RedisIoAdapter (shared/infra) for multi-instance Socket.IO fan-out
  with graceful fallback to the in-memory IoAdapter when Redis is
  unreachable.
- Pin Socket.IO heartbeat (pingInterval/pingTimeout/connectTimeout)
  via env-tunable gateway options for reconnect stability.
- Expose Prometheus metrics on /notifications: goodgo_ws_connected_clients
  (Gauge) and goodgo_ws_messages_total (Counter) with namespace/event/
  direction labels. Wired through MetricsService and tracked across
  connect/disconnect + emits.
- Unit tests: RedisIoAdapter connect/fallback/close, new MetricsService
  WS helpers, and gateway metric increments/decrements on auth paths.

Co-Authored-By: Paperclip <noreply@paperclip.ing>
This commit is contained in:
Ho Ngoc Hai
2026-04-18 15:06:25 +07:00
parent 5d4ecdeb2f
commit 329a821b4a
13 changed files with 410 additions and 5 deletions

View File

@@ -11,6 +11,10 @@ export const DB_QUERY_DURATION = 'db_query_duration_seconds';
export const DB_POOL_ACTIVE_CONNECTIONS = 'db_pool_active_connections';
export const SEARCH_QUERY_DURATION = 'search_query_duration_seconds';
// ── WebSocket Metrics ──
export const GOODGO_WS_CONNECTED_CLIENTS = 'goodgo_ws_connected_clients';
export const GOODGO_WS_MESSAGES_TOTAL = 'goodgo_ws_messages_total';
// ── Web Vitals / RUM Metrics ──
export const WEB_VITALS_LCP = 'goodgo_web_vitals_lcp_seconds';
export const WEB_VITALS_FCP = 'goodgo_web_vitals_fcp_seconds';