Compute start issue

[Aparajita_Nag]

Hi Databricks team,

I have unity catalog enabled in azure databricks. My region is West India, I have created a compute to run sql notebook with policy unrestricted ,checked photon acceleration and single node in Trial edition of azure. Databricks runtime selected as 19 Scala 2.13, Spark 4.2.0. I am getting error while starting the compute:

cloud provider resource stockout. the VM size you are specifying is not available

JSON:

{

  "data_security_mode": "DATA_SECURITY_MODE_AUTO",

  "cluster_name": "myunitycatalogcompute",

  "kind": "CLASSIC_PREVIEW",

  "runtime_engine": "PHOTON",

  "spark_version": "19.x-scala2.13",

  "node_type_id": "Standard_D4ds_v5",

  "autotermination_minutes": 30,

  "is_single_node": true,

  "cluster_id": "0825-074459-gdo58ihn"

}

Please let me know the workaround to work on notebooks. I have 2 serverless  compute. How can these be connected to notebook. Drop down is not giving these options. 

Default Interactive Compute

Default Automated Compute

And I cannot attach starter warehouse too to the sql notebook.

[balajij8]

@Aparajita_Nag 

Notebooks rely on either classic all purpose clusters or serverless compute in databricks. You can use serverless compute because your workspace already has serverless enabled. The error occurred as Azure is experiencing temporary capacity constraints for that specific VM SKU in the West India region during classic cluster provisioning. You can bypass the regional capacity bottleneck by selecting alternative VM families such as Standard_D4ds_v4, Standard_D8ds_v5 or Standard_E4ds_v5 if classic cluster is definitely required.

You can stay in the current notebook and run any code you need. The serverless compute will handle it. Instead of classic cluster select serverless. Everything you need to work with Unity Catalog tables via SQL is already available in the current environment. You can also use SQL editor and create sql queries if you plan to use SQL warehouses.

[nick_martinek]

Hi @Aparajita_Nag 

First, try selecting a different VM size for the classic compute, as SkuNotAvailable / Capacity Restrictions usually means Azure currently has no capacity for Standard_D4ds_v5 in West India.

It is also worth checking your Azure subscription quotas under Subscriptions → Usage + quotas, filtered for West India. Check both Total Regional vCPUs and the quota for the VM family you are using (D-series). If either quota is exhausted, the cluster will not be able to provision.

If the quotas are fine, then this is most likely a regional SKU capacity issue and switching to another VM size/family is the simplest workaround.