Power BI Refresh Error:
HTTP 430

What does this error mean?

A Spark notebook, job, or pipeline in Fabric was rejected with HTTP 430 because the workspace has exceeded its allocated Compute Units (CUs) for Spark workloads. This is a Fabric-specific throttling code for Spark capacity limits, distinct from the generic HTTP 429 API rate limit.

Common causes

• 1Multiple Spark sessions running simultaneously in the workspace, exhausting the concurrent job limit for the capacity SKU
• 2A large-scale ETL operation consuming Compute Units beyond the SKU allocation
• 3Orphaned or stalled Spark sessions left running in the background, consuming CUs without doing work
• 4Capacity SKU is too small for the current Spark workload demands
• 5All capacity CUs consumed by other workloads in the same workspace

How to fix it

1. 1Stop orphaned or stalled Spark sessions in the Fabric workspace (Workspace settings > Spark > Active sessions)
2. 2Enable Autoscale Billing for Spark workloads — this offloads Spark jobs to dedicated serverless resources and eliminates CU consumption from capacity
3. 3Stagger Spark job schedules to avoid peak-time overlap where multiple heavy jobs run simultaneously
4. 4Upgrade the Fabric capacity SKU if your sustained Spark workload regularly exceeds the current CU allocation
5. 5Use Fabric capacity metrics (Capacity Metrics app) to identify which workloads are consuming the most CUs and optimize them

Other capacity errors