chmonitor

The landing page for each ClickHouse host: instant health status, active queries, memory, disk, and CPU at a glance.

Routes/overview
Feature idoverview
Default accesspublic
Requires authNo (set CHM_FEATURE_OVERVIEW_ACCESS=authenticated to gate)
System tablessystem.asynchronous_metrics, system.metrics, system.processes, system.disks, system.tables
ClickHouse grantsSELECT on the system tables above

What it does

Overview is the first page you land on. It pulls live numbers from several system tables to answer:

  • Is the server healthy right now?
  • How many queries are running?
  • How much memory and disk are in use?
  • What is CPU load?

The page is organized into tabs: Connections, Queries, Merges, Replication, and System. Each tab shows a small set of charts that auto-refresh. There are no write operations on this page.

Pages

PageRouteWhat it showsSystem tables
Overview/overviewCPU, memory, disk gauges; active query count; connection count; replication lag indicatorsystem.asynchronous_metrics, system.metrics, system.processes, system.disks, system.tables

Permissions & access

Overview is public by default. To restrict it:

## Env var
CHM_FEATURE_OVERVIEW_ACCESS=authenticated
## CHM_CONFIG_FILE (TOML)
[features.overview]
access = "authenticated"

To disable the feature entirely:

CHM_DISABLED_FEATURES=overview
## or
CHM_FEATURE_OVERVIEW_ENABLED=false

When disabled, the route is removed from the nav and a disabled screen is shown on direct visit.

Configuration

No feature-specific configuration. The refresh interval and caching behaviour follow the global settings:

CLICKHOUSE_MAX_EXECUTION_TIME=60   # query timeout in seconds
NEXT_QUERY_CACHE_TTL=3600          # server-side cache TTL in seconds

Notes & limitations

  • system.asynchronous_metrics values are recalculated by ClickHouse on a background interval (default ~1 s). The values shown are as fresh as that interval allows.
  • system.disks reports the disks ClickHouse knows about. If you use multiple storage policies or remote disks (S3, HDFS), all configured disks appear.
  • system.tables is queried to derive a table count. On clusters with many databases this can be slow; increase CLICKHOUSE_MAX_EXECUTION_TIME if overviews time out.
  • The Replication tab shows replication lag only when replicated tables exist on the host. The tab still appears even if there are none.