Skip to main content
The one endpoint that needs no authentication. Intended for a monitor.
Real response from a running 0.7.3:
No parameters. No authentication. It counts against no rate limit.

​
What a 200 here actually proves

That the HTTP ingress is up and serving. That is all, and the gap matters when you are choosing what to alert on.
A gateway with every vendor down still answers /ping with 200. Binds, queue depth and throughput live on the management port, which is not this one — see the management port. Alerting on /ping alone gives you a monitor that stays green through the outages you most want to know about.

​
Using it as a health check

Reasonable as a load-balancer liveness probe, where “is this process serving?” is exactly the question. For readiness — “should this instance receive traffic?” — use /ops/health on the management port instead, and read bound_transmittable rather than the status field.

The management port

Binds, sessions, queue depth and measured TPS.

Live health

The handful of signals worth an alert.