Keeper HTTP API and Dashboard

Documentation for the ClickHouse Keeper HTTP API and embedded dashboard

ClickHouse Keeper provides an HTTP API and embedded web dashboard for monitoring, health checks, and storage management. This interface allows operators to inspect cluster status, execute commands, and manage Keeper storage through a web browser or HTTP clients.

Configuration

To enable the HTTP API, add the http_control section to your keeper_server configuration:

<keeper_server>
    <!-- Other keeper_server configuration -->

    <http_control>
        <port>9182</port>
        <!-- <secure_port>9443</secure_port> -->
    </http_control>
</keeper_server>

Configuration Options

Setting	Default	Description
http_control.port	-	HTTP port for dashboard and API
http_control.secure_port	-	HTTPS port (requires SSL configuration)
http_control.readiness.endpoint	/ready	Custom path for the readiness probe
http_control.storage.session_timeout_ms	30000	Session timeout for storage API operations

Endpoints

Dashboard

• Path: /dashboard
• Method: GET
• Description: Serves an embedded web dashboard for monitoring and managing Keeper

The dashboard provides:

• Real-time cluster status visualization
• Node monitoring (role, latency, connections)
• Storage browser
• Command execution interface

Cluster tab

The Cluster tab renders the Raft membership as a topology graph and a table. Each member is shown with a health color:

• Green — alive and in sync with the leader
• Yellow — alive but lagging behind the leader by more than stale_log_gap log entries
• Red — unreachable (no successful Raft response within the heartbeat expiry window)
• Gray — unknown (peer health is only visible from the leader; followers see their peers as unknown)

The table also shows each member’s role (leader, follower, or observer), Raft priority, last log index, replication lag relative to the leader, and the time since the last successful Raft response. When the current node is not the leader, the tab offers a deep-link that opens the leader’s dashboard, where full peer health is available. The tab can be opened directly with /dashboard?tab=cluster.

Readiness Probe

• Path: /ready (configurable)
• Method: GET
• Description: Health check endpoint

Success response (HTTP 200):

{
  "status": "ok",
  "details": {
    "role": "leader",
    "hasLeader": true
  }
}

Commands API

• Path: /api/v1/commands/{command}
• Methods: GET, POST
• Description: Executes Four-Letter Word commands or ClickHouse Keeper Client CLI commands

Query parameters:

• command - The command to execute
• cwd - Current working directory for path-based commands (default: /)

Examples:

# Four-Letter Word command
curl http://localhost:9182/api/v1/commands/stat

# ZooKeeper CLI command
curl "http://localhost:9182/api/v1/commands/ls?command=ls%20'/'&cwd=/"

Storage API

• Base Path: /api/v1/storage
• Description: REST API for Keeper storage operations

The Storage API follows REST conventions where HTTP methods indicate the operation type:

Operation	Path	Method	Status Code	Description
Get	/api/v1/storage/{path}	GET	200	Get node data
List	/api/v1/storage/{path}?children=true	GET	200	List child nodes
Exists	/api/v1/storage/{path}	HEAD	200	Check if node exists
Create	/api/v1/storage/{path}	POST	201	Create new node
Update	/api/v1/storage/{path}?version={v}	PUT	200	Update node data
Delete	/api/v1/storage/{path}?version={v}	DELETE	204	Delete node