Using CHouse UIMonitoring
Monitoring overview
The ten Monitoring tabs, the permission each one needs, and the system tables behind them.
- In the app
- Monitoring
/monitoring
Monitoring (/monitoring/<tab>) shows what your ClickHouse servers are doing, read straight from their system.* tables through the CHouse UI server — no exporter or agent needed. Each tab needs its own permission, so a role can open some and not others.
The tabs
| Tab | Route | Needs | Answers |
|---|---|---|---|
| Live queries | /monitoring/live-queries | live_queries:view | What is running right now, and can I stop it? |
| Query logs | /monitoring/logs | logs:view | What ran, how long it took, what it cost |
| Metrics | /monitoring/metrics | metrics:view (metrics:view:advanced for the deeper views) | How the server is doing over time |
| Parts | /monitoring/parts | parts:view | Merges, mutations and part movements |
| Schema advisor | /monitoring/schema | schema_advisor:view | Which columns waste disk |
| Cluster | /monitoring/cluster | cluster:view | Replication, mutations, topology, insert backlog and DDL |
| Errors | /monitoring/errors | errors:view | Recurring server errors and crashes |
| Performance | /monitoring/performance | performance:view | Which query shapes got slower, and what changed |
| Capacity | /monitoring/capacity | capacity:view | When disks fill up, what to reclaim, what reads cost |
| Upgrades | /monitoring/upgrades | upgrades:view | Whether an upgrade is safe for your workload |
The first seven read system tables live. Performance, Capacity and Upgrades read the evidence the collector stored, so they show history even for things ClickHouse has already rotated out.
Tabs the user lacks permission for simply don't appear — the tab strip is RBAC-driven.
Shared controls
- Time range — 15 m / 1 h / 6 h / 24 h presets plus a Grafana-style drill-down calendar (day → month → year) in one popover
- Connection context — every tab reads the active connection
- Chouse AI hooks — with
ai:optimize, rows in logs/errors/parts expose Optimize / Fix / Diagnose actions (see Chouse AI in-tab)
System tables behind the suite
| Table | Used by |
|---|---|
system.query_log | Query logs (all sub-views), timeline chart |
system.query_views_log | Views-triggered drill-down |
system.part_log | Parts |
system.parts_columns | Schema advisor |
system.mutations, system.replication_queue, system.replicas | Cluster activity |
system.metrics, system.events, system.asynchronous_metrics | Metrics tabs |
system.errors, crash log | Errors |
Stored evidence (query_log, system.disks, part_log, versions and settings over time) | Performance, Capacity, Upgrades |
Deployment requirements
- The connection user must be able to read
system.* query_logmust be enabled (default on most installs)- Version caveats on the compatibility page
Not here, but related
- Multi-cluster at a glance → Fleet view
- Automated diagnosis → Doctor
- Page someone when it breaks → Alerting