Monitor with Prometheus & Grafana
Scrape each NF's /metrics endpoint with Prometheus and visualize them with Grafana.
Every network function exposes Prometheus metrics over a /metrics endpoint. Point Prometheus at them, then import the NRF Grafana dashboard to get a view of the core.
Metrics endpoints
| NF | Endpoint | Notes |
|---|---|---|
| NRF | :8080/metrics | Served on the NRF's SBI port |
| DMF | :8081/metrics | Served on the DMF's SBI port |
| DPF | :8082/metrics | Served on the DPF's SBI port |
| SMSF | :8083/metrics | Served on the SMSF's SBI port (optional NF) |
| NSSF | :8084/metrics | Served on the NSSF's SBI port (optional NF) |
| UPF | :9095/metrics | Dedicated metrics listener (-metrics-addr) |
/health, /ready, and /metrics stay open even when
OAuth2 is enabled on the SBI.Scraping with Prometheus
Add a scrape job per NF. The ports above are the defaults; the UPF's is configurable with -metrics-addr. The targets below use Compose DNS names — native scrapes would use 127.0.0.1 instead:
The NRF Grafana dashboard
A ready-made Grafana dashboard, nrf-overview.json, lives under nrf/dashboards/. Import it into Grafana (Dashboards → Import → upload the JSON) with your Prometheus as the data source to get an "NRF Overview" out of the box.
A few metrics to watch
NRF (registry and discovery health):
nrf_nf_instances_registered— registered NF instances by type.nrf_nf_registrations_total/nrf_nf_deregistrations_total— registration churn.nrf_heartbeat_expirations_total— NFs the NRF stopped hearing from.nrf_discovery_requests_total— how often NFs ask the NRF to find peers.
DMF (registration and sessions):
dmf_registrations_total— UE registration procedures by type and result.dmf_authentications_total— primary authentication outcomes.dmf_pdu_sessions_total— PDU session events by type and result.dmf_ngap_messages_total— NGAP messages by procedure and direction.
UPF (data plane health):
upf_sessions_active— PFCP sessions installed in the datapath.upf_associations_active— active PFCP node associations.upf_datapath_packets_total/upf_datapath_bytes_total— forwarded traffic by direction and IP family (userspace datapath).upf_pfcp_errors_total— PFCP processing errors.
DPF and NRF both expose HTTP request counters and latency histograms (the dpf_http_* and nrf_http_* series) for SBI traffic. Optional NSSF and SMSF export nssf_selections_total and smsf_messages_total respectively.
Where to go next
See the full metric list and labels in the metrics reference.