Skip to content

Prometheus metrics

Northwatch exposes Prometheus metrics at /metrics (always on). Alongside the standard Go runtime and process collectors, it exports OVN-specific gauges and its own HTTP server metrics. Metrics cover the default cluster only.

See Scrape Prometheus metrics for scrape configuration.

OVN & connection metrics

MetricTypeMeaning
northwatch_ovsdb_connectedgaugeWhether an OVSDB endpoint is connected.
northwatch_ovsdb_connectionsgaugeOVSDB connection / cluster-member count.
northwatch_ovsdb_table_rowsgaugeRow count per monitored table.
northwatch_logical_flows_totalgaugeTotal Southbound logical flows.
northwatch_port_bindings_totalgaugeTotal port bindings.
northwatch_bfd_sessionsgaugeBFD session counts by state.

Config-propagation metrics

These track how far each hypervisor's realized config lags the desired config (nb_cfg / sb_cfg / hv_cfg):

MetricTypeMeaning
northwatch_nb_cfggaugeGlobal Northbound nb_cfg sequence.
northwatch_sb_cfggaugeSouthbound sb_cfg sequence.
northwatch_sb_nb_cfggaugeSouthbound view of nb_cfg.
northwatch_hv_cfggaugeHypervisor hv_cfg sequence.
northwatch_chassis_nb_cfggaugePer-chassis realized nb_cfg.
northwatch_chassis_nb_cfg_laggaugePer-chassis config-realization lag.
northwatch_chassis_port_countgaugePorts bound per chassis.

HTTP server metrics

MetricTypeMeaning
northwatch_http_requests_totalcounterHTTP requests served.
northwatch_http_request_duration_secondshistogramRequest latency.

Standard collectors

The registry also includes Go's GoCollector (go_*) and ProcessCollector (process_*) for runtime and process metrics.

INFO

Metric label sets and exact help text are best read from a live scrape: curl -s http://localhost:8080/metrics | grep northwatch_.