The error is coming from AWS, not Databricks: your AWS account is restricted to Free Tier–eligible instance types, but the node type you picked in Databricks maps to an EC2 instance that is not Free Tier–eligible, so AWS rejects the launch request with InvalidParameterCombination: The specified instance type is not eligible for Free Tier.
1. “Compatible” nodes between AWS and Databricks
On Databricks on AWS, each node type is just a curated EC2 instance type (driver + workers) that Databricks is allowed to launch in your account (for example, m5.xlarge, m5d.xlarge, r5d.xlarge, m6i.large, etc.).
Databricks only supports specific instance families for nodes (general purpose M, compute-optimized C, memory-optimized R, storage-optimized I/D, GPU G/P, and some fleet types).
AWS Free Tier–eligible instances (like the various `t_micro` types) are not in this supported set, so there is effectively no Databricks node type that is both Free Tier–eligible and supported for clusters.
2. Ways to “connect both” / get this working
You can’t attach an existing Free Tier EC2 instance to Databricks; Databricks must create and manage the EC2 nodes itself. The practical options are:
-
Use a normal (billable) AWS instance family that Databricks supports
- In the Databricks Compute UI, pick a standard node type such as m5d.xlarge, m6i.large, etc.
- Your AWS account must allow non–Free Tier instances (no policy blocking them). You’ll then pay normal EC2 + Databricks DBU costs.
-
Use Databricks Serverless compute (if available in your region/plan)
- With Serverless, Databricks manages the underlying instances and you only choose the size / SKU; infra is included in the serverless price, but it’s not Free Tier—you’re still billed through Databricks rather than your EC2 Free Tier.
-
For learning only: use Databricks Community Edition (separate environment)
- Community Edition runs on Databricks’ own AWS account with small, managed clusters and does not use your AWS Free Tier at all. It’s good for demos and notebooks, not for production.
If your goal is “no AWS charges at all” on your own account, then you’ll need to use Community Edition; running a full Databricks workspace on your AWS account inevitably requires paid instance types, not Free Tier–only ones.
