Troubleshoot

Troubleshoot is the observability entry point in the left rail. It opens on a view built for one moment: something is wrong and you do not yet know where.

What is on it

• A service dependency graph — which services call which, and where latency and errors concentrate.
• Slowest operations, ranked by p99.
• Slowest traces, individually.

Any node in the graph, any slow operation, and any slow trace has an Explain action that opens root-cause analysis for it. Clicking a slow-trace row instead opens the full trace waterfall.

Below the overview sit five tabs: Traces, Metrics, Logs, Sampling and Cardinality.

Ask Runtimez

The same page hosts Ask Runtimez — describe the symptom in your own words (“checkout is slow since this morning”) and it routes to the right evidence rather than making you pick the signal first.

Answers cite their evidence. When the evidence is not there, the answer says so instead of filling the gap.