Skip to content

Latest commit

 

History

History
68 lines (52 loc) · 4.85 KB

spark_version.md

File metadata and controls

68 lines (52 loc) · 4.85 KB
subcategory
Compute

databricks_spark_version Data Source

-> Note If you have a fully automated setup with workspaces created by databricks_mws_workspaces or azurerm_databricks_workspace, please make sure to add depends_on attribute in order to prevent default auth: cannot configure default credentials errors.

Gets Databricks Runtime (DBR) version that could be used for spark_version parameter in databricks_cluster and other resources that fits search criteria, like specific Spark or Scala version, ML or Genomics runtime, etc., similar to executing databricks clusters spark-versions, and filters it to return the latest version that matches criteria. Often used along databricks_node_type data source.

-> Note This is experimental functionality, which aims to simplify things. In case of wrong parameters given (e.g. together ml = true and genomics = true, or something like), data source will throw an error. Similarly, if search returns multiple results, and latest = false, data source will throw an error.

Example Usage

data "databricks_node_type" "with_gpu" {
  local_disk  = true
  min_cores   = 16
  gb_per_core = 1
  min_gpus    = 1
}

data "databricks_spark_version" "gpu_ml" {
  gpu = true
  ml  = true
}

resource "databricks_cluster" "research" {
  cluster_name            = "Research Cluster"
  spark_version           = data.databricks_spark_version.gpu_ml.id
  node_type_id            = data.databricks_node_type.with_gpu.id
  autotermination_minutes = 20
  autoscale {
    min_workers = 1
    max_workers = 50
  }
}

Argument Reference

Data source allows you to pick groups by the following attributes:

  • latest - (boolean, optional) if we should return only the latest version if there is more than one result. Default to true. If set to false and multiple versions are matching, throws an error.
  • long_term_support - (boolean, optional) if we should limit the search only to LTS (long term support) & ESR (extended support) versions. Default to false.
  • ml - (boolean, optional) if we should limit the search only to ML runtimes. Default to false.
  • genomics - (boolean, optional) if we should limit the search only to Genomics (HLS) runtimes. Default to false.
  • gpu - (boolean, optional) if we should limit the search only to runtimes that support GPUs. Default to false.
  • beta - (boolean, optional) if we should limit the search only to runtimes that are in Beta stage. Default to false.
  • scala - (string, optional) if we should limit the search only to runtimes that are based on specific Scala version. Default to 2.12.
  • spark_version - (string, optional) if we should limit the search only to runtimes that are based on specific Spark version. Default to empty string. It could be specified as 3, or 3.0, or full version, like, 3.0.1.
  • photon - (boolean, optional) if we should limit the search only to Photon runtimes. Default to false. Deprecated with DBR 14.0 release. Specify runtime_engine=\"PHOTON\" in the cluster configuration instead!
  • graviton - (boolean, optional) if we should limit the search only to runtimes supporting AWS Graviton CPUs. Default to false. Deprecated with DBR 14.0 release. DBR version compiled for Graviton will be automatically installed when nodes with Graviton CPUs are specified in the cluster configuration.

Attribute Reference

Data source exposes the following attributes:

Related Resources

The following resources are used in the same context: