HAProxy monitoring with Topologix

Infrastructure & Networking

Load-balancer health from HAProxy’s built-in Prometheus exporter — backend and server up/down state, request and error rates, queueing, DNS resolver activity, and backend SSL certificate expiry, aggregated from one /metrics scrape per poll.

Requirements

  • The Prometheus exporter enabled in haproxy.cfg (HAProxy 2.0+): a stats frontend with http-request use-service prometheus-exporter if { path /metrics }, typically on port 8404.
  • HTTP(S) reach from the Topologix server or a proxy agent to that port; optional Basic auth if the stats frontend requires it.

1 · Set up the integration

  1. Open the Integrations pageClick Integrations in the toolbar and pick HAProxy from the catalogue.
  2. Enter the metrics URLThe full exporter path — http://haproxy.internal:8404/metrics, not just the host. Optional Username/Password for a protected stats frontend, and an Allow self-signed TLS certificates toggle.
  3. Create, then verifyClick Create, then reopen the integration and click Test — it verifies the endpoint actually serves HAProxy metrics (not just any web page) and reports the version plus frontend/backend/server counts, so you know you hit the right instance.
The integration catalogue — HAProxy found via search.
The integration catalogue — HAProxy found via search.
The HAProxy configuration drawer with example values. Every screen in this guide is the real product.
The HAProxy configuration drawer with example values. Every screen in this guide is the real product.
Target on a private network? Route it through a proxy agent. Proxy agents run inside your network — behind NAT, on a management VLAN, in an air-gapped segment — and reach targets the Topologix server can't. The agent polls the server every 15 seconds for its assignments, runs the checks locally, and pushes results back every 5 seconds, so no inbound firewall rules are ever needed. Set a default proxy on the integration; every metric source can override it. If a proxy goes offline, its sources are marked stale within ~90 seconds and the last known values stay visible.

2 · Add it to a flow

  1. Select the load-balancer nodeClick the node, open Metrics, then + Add and choose the HAProxy integration.
  2. Add the sourceNo per-type options — one source covers the whole HAProxy process: every frontend, backend, and server it knows.
  3. SaveSet the interval and click Add Source. The card shows backends up/down, sessions, per-second request and 5xx/error rates, and worst-backend rollups (max 5xx/s, deepest queue, slowest response) for picker-free alerting.
The add-source form with the integration selected and its per-type options loaded.
The add-source form with the integration selected and its per-type options loaded.
The live card on the canvas — latest value, label, and sparkline, updating on the source’s interval.
The live card on the canvas — latest value, label, and sparkline, updating on the source’s interval.

Good to know

  • Request and error rates are counter deltas between polls — the first sample shows no rate, every later one is a true per-second figure that survives HAProxy reloads.
  • Backend SSL certificate expiry is tracked (with an expiring-soon window); frontend bind certificates are not exposed by the exporter — add a separate HTTP monitor for those.
  • The scrape is capped at 10 seconds so a wedged HAProxy under load can’t pile up the polling queue.

All integrations →

See it on your own infrastructure

Book a demo and we will walk you through Topologix with your systems in mind: on our cloud, or self-hosted in your own network.

Book a demoContact us