Skip to main content

metrics

Metrics & Observability​

Prometheus metrics (low‑cardinality labels: connector, side where applicable):

  • nowconnect_tcp_connections_total
  • nowconnect_tcp_active_connections
  • nowconnect_tcp_bytes_total{direction="c2p|p2c"}
  • nowconnect_tunnel_reconnects_total{agent,reason}
  • nowconnect_tunnel_uptime_seconds{agent}
  • nowconnect_listener_accept_errors_total{connector}
  • nowconnect_queue_overflow_total{side,connector}
  • nowconnect_fin_sent/fin_recv/rst_sent/rst_recv
  • nowconnect_pdp_allow_total/nowconnect_pdp_deny_total
  • nowconnect_pdp_errors_total
  • nowconnect_connection_duration_seconds (histogram)

Health endpoints:

  • /healthz — process alive
  • /readyz — WS server up (cloud) / tunnel established (premise)
  • /metrics — Prometheus scrape

Logging: structured JSON without payloads; include event, cid, agent_id, connector, seq, sizes, timings; redact sensitive values.

Health/readiness semantics​

  • Agent: single TCP health server; returns OK prior to HELLO/ACK, READY after ACK. Use tcpSocket probes in k8s.
  • Cloud: /healthz (alive), /readyz (WS server up and accepting). Use Prometheus /metrics for scraping.