Monitor/Overview
Monitor
Everything that watches something and reports what it saw.
The Monitor section answers is it up, how fast, and where. Each page below covers one source; they share the same alerting model, described in How alerting works.
Services
HTTP and TCP checks, health documents, status and response time.
Servers
Machines: vitals, listeners, processes, disk prediction.
Dependency map
What calls what, and what breaks if this goes.
Synthetic checks
Scripted journeys run on a schedule from outside.
APM and traces
Where the time actually goes inside a request.
Network
Reachability, latency and packet loss between points.
Databases
MySQL nodes: replication, GTIDs, config drift.
Database pool
ProxySQL: routing, backends, failover.
Cloudflare
Traffic, errors, latency, bots, DNS, Workers, RUM, load balancing.
Web analytics
GA4: sessions, real-time, breakdowns.
Which one should I use?
| Question | Page |
|---|---|
| Is my API answering? | Services |
| Is the box healthy? | Servers |
| Can users actually complete checkout? | Synthetic |
| Why is this endpoint slow? | APM |
| Which country is getting errors? | Cloudflare |
| Is the replica behind? | Databases |
| Are connections queueing? | Database pool |
| What would break if this service went? | Dependency map |
Two things worth knowing
A check that cannot run is not a passing check. Every source here distinguishes "measured and healthy" from "could not measure". A service whose check has been failing to run for an hour is not green.
Never measured is not healthy. A service or rule that has produced no reading at all is marked as such, rather than blending into the healthy majority.
Part of Monitor — Everything that watches something and reports what it saw.