Telemetry

If you ship OpenTelemetry data into ClickHouse (via the OTel ClickHouse exporter or your own pipeline), the Telemetry page gives you a purpose-built log explorer on top of it — no Grafana or Kibana needed for the everyday "what just happened" check.

Setup

There is almost none. CH-UI auto-detects OTel tables by scanning system.tables for otel_* names (otel_logs, otel_traces, otel_metrics) in your connected ClickHouse. If your tables live in a different database or use different names, point Telemetry at them in the page's configuration.

Setup Wizard

If no tables are auto-detected, a Setup Wizard appears: enter your logs database and table names (or hit Retry to scan again) and save. The selection is persisted per connection via the config API and loaded on every visit — the Log Explorer queries the table you configured, not a hardcoded default.

GET /api/telemetry/config   # saved telemetry config for the active connection
PUT /api/telemetry/config   # persist logs database/table selection

Logs explorer

• Severity filtering — narrow to errors/warnings instantly
• Service breakdown — the service list is queried live from your logs table, so you can see which services are producing logs and filter by service
• Volume histogram — log volume over time, so spikes stand out before you scroll
• Time-range controls with live querying — everything is computed by ClickHouse on demand; CH-UI stores nothing beyond the table selection

Time bounds are validated as RFC3339 and rendered via parseDateTime64BestEffort, so filters work against both DateTime and DateTime64 timestamp columns (a raw string literal would fail on DateTime64).

v2.6.2 fixes: the saved wizard config is now actually respected (previously the Log Explorer could fall back to default.otel_logs regardless of your selection), and time filters no longer error with TYPE_MISMATCH on DateTime64 columns thanks to the parseDateTime64BestEffort handling above.

Brain integration

The Brain AI assistant has telemetry tools built in (query_logs, query_traces, list_services, query_metrics) and so you can ask things like "show me error logs for the checkout service in the last hour" in plain language and get the underlying query back.

Roadmap

Traces and metrics exploration views are in progress — the data layer (and the Brain tools above) already speak traces and metrics; dedicated UI tabs are next.

Requirements

• OTel data in ClickHouse — any schema the ClickHouse exporter produces works out of the box
• Telemetry is part of the open-source core — no license required