Scrape Prometheus metrics

Northwatch exposes Prometheus metrics at /metrics, covering OVSDB connection health, table row counts, config-propagation state, and its own HTTP server. The endpoint is always on.

Scrape it

curl -s http://localhost:8080/metrics | grep northwatch_

A minimal Prometheus scrape config:

scrape_configs:
  - job_name: northwatch
    static_configs:
      - targets: ['northwatch.example.com:8080']

Liveness and readiness

For orchestrators, Northwatch also serves plain health probes:

curl -s http://localhost:8080/healthz   # liveness
curl -s http://localhost:8080/readyz    # readiness (ready after initial sync)

What is exported

The custom collector exports OVN-specific gauges alongside the standard Go and process collectors. Highlights:

Metric	Meaning
northwatch_ovsdb_connected	Whether each OVSDB endpoint is connected
northwatch_ovsdb_table_rows	Row count per monitored table
northwatch_logical_flows_total	Total Southbound logical flows
northwatch_port_bindings_total	Total port bindings
northwatch_chassis_nb_cfg_lag	Per-chassis config-realization lag
northwatch_bfd_sessions	BFD session counts by state
northwatch_http_requests_total	Northwatch HTTP request counter

For the complete list, see Prometheus metrics.

Single-cluster scope

Metrics are exported for the default cluster only. Per-cluster routes do not register their own Prometheus collectors.