MetricSign
EN|NLRequest Access
High severitycluster

Power BI Refresh Error:
AWS_INSUFFICIENT_INSTANCE_CAPACITY_FAILURE

What does this error mean?

AWS does not have enough capacity for the requested EC2 instance type in the selected availability zone. This is a cloud provider capacity issue — Databricks cannot launch the cluster until AWS has available instances.

Common causes

  • 1High demand for the selected instance type in the target AWS region or AZ
  • 2GPU or memory-optimized instances are especially prone to capacity shortages
  • 3Spot instance capacity is depleted in the AZ at the requested price
  • 4A regional event (e.g., other large customers scaling up) exhausted AZ capacity

How to fix it

  1. 1Configure the cluster to use multiple availability zones so Databricks can fall back to another AZ
  2. 2Switch to an alternative instance type with similar specs (e.g., r5.4xlarge instead of r5.2xlarge)
  3. 3Use an instance pool with multiple instance types to improve capacity availability
  4. 4For spot clusters, configure on-demand fallback
  5. 5Check the AWS capacity dashboard or service health dashboard for ongoing capacity events

Frequently asked questions

Is this a Databricks problem or an AWS problem?

It is an AWS capacity problem. Databricks has made the correct API call, but AWS cannot fulfill the instance launch request.

Does Databricks retry automatically?

Databricks will retry launching the cluster a few times. If capacity is still unavailable, the cluster terminates with this error code.

Other cluster errors