Does this work without NGINX Plus?

Basic upstream request and response metrics come from the access log and work on open-source NGINX. Per-backend health state, active connection counts, and retry counts require NGINX Plus's API module.

How do I alert on an upstream backend going down?

Configure an alert on nginx.upstream.peers.state.unhealthy with threshold >= 1 over 1 minute. Alerts route to Slack, Telegram, email, and webhooks.

Can I see per-backend latency?

Yes. Upstream response time is graphed per peer with p50/p95/p99 where supported.

NGINX Upstream Monitoring
See every backend, every second.

Per-backend health, response times, failures and retries. Alerts when a peer is marked down. Works with both open-source NGINX and NGINX Plus.

Start free See demo

Upstream metrics collected

nginx.upstream.request.count

Requests per upstream, per peer.

nginx.upstream.response.time

Average and percentile backend response time.

nginx.upstream.peers.fails

Consecutive peer failures that triggered health-check marks.

nginx.upstream.peers.state

healthy / unhealthy / checking state for each peer (NGINX Plus).

nginx.upstream.next.count

Retries routed to the next upstream after a failed request.

nginx.upstream.status.5xx

Backend-produced 5xx rate per upstream pool.

Catch backend failures before users do.

Start free View pricing

NGINX Upstream MonitoringSee every backend, every second.