Problem

You may want to extend the number of subnets available in your workspace for compute creation. When you try to add subnets, you get an error that says your requested instance type is not supported in your requested availability zone.

Example error

In zone us-east-le, insufficient capacity for the following node type: rid.12xlarge. AWS Error: Failure happened when talking to AWS, AWS API error code: Unsupported AWS error message: Your requested instance type (rid.12xlarge) is not supported in your requested Availability Zone (us-east-1e).

Cause

During workspace creation, you likely defaulted to creating two private subnets. This is sufficient for most use cases, however depending on your specific requirements, you may want to add additional availability zones.

Solution

1. Login to your AWS account and identify the VPC associated with your Databricks workspace.
2. Create a new subnet that complies with the requirements in the Additional Subnets requirements documentation.

Make sure you create the new subnet in an availability zone that is different from your existing subnets. For example, if your first two subnets were created in us-west-2a and  us-west-2b, make sure your new subnet is created in us-west-2c or similar.

3. Once the new subnet is created, collect the VPC id, the subnet ids (including the original ones), and the security group ids.
4. Login to the Databricks Account Console.
5. Click Cloud Resources.
6. Click Network. 
7. Click Network configuration.
8. Create a new network configuration object with your new subnet.
9. After the new network config is set, open the Databricks Account Console.
10. Click Workspaces.
11. Select the desired workspace.
12. Click Update Workspace.
13. Under Advanced Settings go to Network configuration and replace your current Network configuration with the new one.
14. Click Update.