oh cluster
Manage compute clusters.
Cluster types: Managed SQL Spark Open_Engines Notebook LakeBase
oh cluster show
Lists all clusters.
oh cluster show
oh cluster describe NAME
Describes a single cluster.
oh cluster describe prod
oh cluster create NAME
Creates a new cluster.
Flags
| Flag | Required | Description |
|---|---|---|
--type | Yes | Cluster type: Managed, SQL, Spark, Open_Engines, Notebook, LakeBase |
--max-ocu INT | No | Maximum OCU limit. Required for all types except LakeBase. |
--min-ocu INT | No | Minimum OCU limit. Required for all types except LakeBase. |
--options KEY=VALUE | No | WITH clause parameters; repeatable (see Special parameters) |
Special parameters
All WITH clause parameters from the CREATE CLUSTER API are passed via --options KEY=VALUE.
| Parameter | Description |
|---|---|
worker.type | Worker instance type (e.g., oh-general-4) |
worker.spot | True or False — enable spot instances for workers |
driver.type | Driver instance type. Default is Auto (same as workers) |
quanton.enabled | true or false — enable Quanton engine. Valid for SQL, Spark, Notebook only |
open_engines.engine | Trino, Flink, or Ray. Required for Open_Engines type |
open_engines.catalog | Catalog name for Trino/Flink clusters. Required for Trino |
open_engines.ray.max_cpu_units | Max CPU units for Ray clusters |
open_engines.ray.min_cpu_units | Min CPU units for Ray clusters |
open_engines.ray.max_gpu_units | Max GPU units for Ray clusters |
worker.gpu_type | GPU worker instance type (e.g., oh-gpu-4). Ray clusters only |
attached_storage.driver_storage_size | Driver attached storage in GB (AWS only) |
attached_storage.worker_storage_size | Worker attached storage in GB (AWS only) |
ocu.notifications.enabled | True or False — OCU usage notifications. Default True |
ocu.notifications.threshold | Percentage of max OCU that triggers a notification. Default 80 |
lakebase.engine_size | Engine size for LakeBase clusters: 2x-small, x-small, small, medium, large, x-large, 2x-large, 3x-large, 4x-large |
lakebase.min_engines | Minimum number of Spark clusters always running (LakeBase only) |
lakebase.max_engines | Maximum number of Spark clusters (LakeBase only) |
catalog.name | Catalog name (LakeBase only). Defaults to the Onehouse catalog |
catalog.permissions_enforcement | none (default) or lake_formation (LakeBase only) |
Examples
Managed cluster
oh cluster create prod \
--type Managed \
--max-ocu 20 \
--min-ocu 2 \
--options worker.type=oh-general-4 \
--options driver.type=oh-general-8 \
--options ocu.notifications.enabled=True \
--options ocu.notifications.threshold=75
Spark cluster with spot workers
oh cluster create jobs-prod \
--type Spark \
--max-ocu 10 \
--options worker.type=oh-general-8 \
--options worker.spot=True
Trino Open Engines cluster
oh cluster create trino-prod \
--type Open_Engines \
--max-ocu 10 \
--min-ocu 1 \
--options worker.type=oh-general-4 \
--options open_engines.engine=Trino \
--options open_engines.catalog=my-glue-catalog
Ray Open Engines cluster (with GPU)
oh cluster create ray-prod \
--type Open_Engines \
--max-ocu 10 \
--min-ocu 1 \
--options worker.type=oh-general-4 \
--options worker.gpu_type=oh-gpu-4 \
--options open_engines.engine=Ray \
--options open_engines.ray.max_cpu_units=7 \
--options open_engines.ray.min_cpu_units=1 \
--options open_engines.ray.max_gpu_units=3
LakeBase cluster
oh cluster create lakebase-prod \
--type LakeBase \
--options lakebase.engine_size=medium \
--options lakebase.min_engines=1 \
--options lakebase.max_engines=5 \
--options catalog.name=my-glue-catalog \
--options catalog.permissions_enforcement=lake_formation
oh cluster alter NAME
Alters an existing cluster. Supply any combination of flags.
Flags
| Flag | Description |
|---|---|
--state | START or STOP |
--max-ocu INT | New maximum OCU |
--min-ocu INT | New minimum OCU |
--new-name | Rename the cluster |
--options KEY=VALUE | WITH clause parameters; repeatable (see Special parameters) |
Special parameters
Uses the same parameter keys as create, with the following available for alter:
| Parameter | Description |
|---|---|
worker.type | New worker instance type |
worker.spot | True or False |
driver.type | New driver instance type |
quanton.enabled | true or false |
open_engines.ray.max_cpu_units | New max CPU units (Ray clusters) |
open_engines.ray.min_cpu_units | New min CPU units (Ray clusters) |
open_engines.ray.max_gpu_units | New max GPU units (Ray clusters) |
attached_storage.driver_storage_size | New driver storage in GB (AWS only) |
attached_storage.worker_storage_size | New worker storage in GB (AWS only) |
lakebase.engine_size | New engine size (LakeBase only) |
lakebase.min_engines | New minimum engines (LakeBase only) |
lakebase.max_engines | New maximum engines (LakeBase only) |
Examples
# Scale up a Managed cluster
oh cluster alter prod --max-ocu 40 --min-ocu 4
# Stop a cluster
oh cluster alter prod --state STOP
# Rename a cluster
oh cluster alter prod --new-name prod-v2
# Update Ray cluster compute limits
oh cluster alter ray-prod \
--max-ocu 8 \
--min-ocu 2 \
--options open_engines.ray.max_cpu_units=5 \
--options open_engines.ray.min_cpu_units=2 \
--options open_engines.ray.max_gpu_units=3
# Resize a LakeBase cluster
oh cluster alter lakebase-prod \
--options lakebase.engine_size=large \
--options lakebase.min_engines=2 \
--options lakebase.max_engines=10
oh cluster delete NAME
Deletes a cluster.
oh cluster delete prod