Status

Everything is running

Checked when you loaded this page, not cached from a green tick somebody set by hand.

Event ingestion

42 ms

Health check round trip.

Data freshness

idle

No events in the last 15 minutes, so there is no lag to measure.

Dashboard

45 ms

Control plane query round trip.

Why data freshness is the headline

Uptime is easy to publish and rarely the thing that hurts. An analytics service can be up in every sense a monitoring tool understands while events sit in a queue and your dashboard quietly shows yesterday.

The number above is how long the slowest recent event took to travel from being accepted to being queryable. Under a minute is normal. It measures the pipeline rather than the traffic, so a quiet afternoon reads as idle instead of pretending to be an incident.

What we commit to

Events are accepted or explicitly refused. Ingestion never drops one silently, and a refusal appears in your event debugger with the reason.
A ClickHouse outage costs latency rather than data. Events queue and are replayed, and the writer only acknowledges an event once the database confirms it.
Incidents that affect customers get a public write-up, including the part where we explain what we broke.
This page runs on the same deployment as the product, which is honest but not ideal. A status page hosted on the thing it monitors is a known joke and will move.