Workload and resource statements

This page describes statements accepted by the Query API in RawTree Cloud and RawTree BYOC. Its restrictions apply to the API, not to all RawTree SQL. See ClickHouse compatibility and API scope.

Syntax

CREATE [IF NOT EXISTS | OR REPLACE] WORKLOAD <name>
  [ON CLUSTER 'default'] [IN <parent>]
  [SETTINGS <name> = <value> [FOR <resource>]]
CREATE [IF NOT EXISTS | OR REPLACE] RESOURCE <name> (<operations>)
  [ON CLUSTER 'default']
DROP [IF EXISTS] WORKLOAD|RESOURCE <name> [ON CLUSTER 'default']

Workloads optionally use IN <parent> and numeric SETTINGS such as weight, priority, and max_concurrent_threads. Add FOR <resource> to apply a setting to a resource. Resources contain comma-separated MASTER THREAD, WORKER THREAD, READ DISK <name>, WRITE DISK <name>, READ ANY DISK, or WRITE ANY DISK entries.

Clauses

Query API behavior

• Supported: Workload trees, numeric scheduling settings, and thread or disk resources.
• Not exposed by the Query API: Memory or query-slot resources, unknown settings, and clusters other than default.
• Database storage behavior: Keeper-backed workload storage omits ON CLUSTER; local storage uses the cluster clause to distribute DDL.

Errors

Error: Unsupported scheduling statement.

Hint:

Use CREATE [OR REPLACE] WORKLOAD/RESOURCE [IF NOT EXISTS], or DROP WORKLOAD/RESOURCE [IF EXISTS]. Only ON CLUSTER 'default' is supported. Submit one statement.

Examples

Create a CPU resource.

CREATE RESOURCE analytics_cpu ON CLUSTER 'default' (MASTER THREAD, WORKER THREAD)

Create a workload with a thread limit.

CREATE WORKLOAD analytics_workload ON CLUSTER 'default' IN all SETTINGS max_concurrent_threads = 2 FOR analytics_cpu

Drop the workload before the resource.

DROP WORKLOAD IF EXISTS analytics_workload ON CLUSTER 'default'

DROP RESOURCE IF EXISTS analytics_cpu ON CLUSTER 'default'

The parser rejects memory resources.

CREATE RESOURCE memory_pool (READ ANY MEMORY)

The parser rejects query-slot resources.

CREATE RESOURCE query_slots (QUERY SLOTS)

The parser rejects other clusters.

CREATE WORKLOAD batch_workload ON CLUSTER 'other'

Related

• Workload scheduling
• Isolate workloads