Your whole cluster on one screen
Query rate, memory, connections, merges and disk — charted the moment you connect. No dashboards to build, no exporters to run.
Point it at a cluster, get charts
chmonitor reads the system tables ClickHouse already writes. Give it a host and the overview is populated on first load — no exporter, no scrape config, no dashboard JSON to import.
- Query rate, errors and duration percentiles
- Memory, connections and thread pools
- Merges, mutations and parts pressure
Every chart drills into the rows behind it
A spike is only useful if you can chase it. Expand the chart strip for detail, change the time range, then click through to the queries, tables or merges that made the shape.
- Adjustable time range per chart
- Click-through from a spike to the underlying queries
- Charts keep showing data during a failed refresh
More of Overview in action
Everything in the box
Live metrics
Query rate, memory, connections, threads.
Merge pressure
Merges, mutations and parts per table.
Disk usage
Storage per disk with headroom left.
Time ranges
Rescope any chart without leaving the page.
Multi-host
Switch clusters from the host switcher.
Graceful errors
A failed refresh keeps the last good data.
Common questions
Do I need Prometheus or Grafana?
No. chmonitor queries ClickHouse system tables directly, so there is nothing to scrape and nothing to store outside your cluster.
How much load does the overview add?
The charts run small system-table queries on a refresh interval, all tagged with a query comment so you can see exactly what chmonitor ran.
Can I watch more than one cluster?
Yes. Configure several hosts and switch between them from the host switcher — each keeps its own overview.
More chmonitor features
Start monitoring — advisor and alerts built in
Open the hosted dashboard and connect a cluster, or self-host with Docker, Kubernetes, or from source. Self-hosting is always free.