Error in microsoft fabric account related to capacity
I am frequently getting below error message on my MS Fabric account. kindly help me on this.
[TooManyRequestsForCapacity] HTTP Response code 430: This Spark job can’t be run because you’ve hit a Spark compute or API rate limit. To proceed, cancel an active Spark job through the Monitoring hub, choose a larger capacity SKU, or try again later. For more visibility and control, go to Workspace settings → Job management (Job Concurrency & Queue Monitoring) to review running and queued Spark jobs, understand capacity contention, and take action as needed. [Learn more at 'https://go.microsoft.com/fwlink/?linkid=2356970&clcid=0x409'].
1 answer
-
Amira Bedhiafi 42,941 Reputation points • MVP • Volunteer Moderator
Hi Deepak,
I think your Fabric workspace has reached the available Spark compute or concurrency limit for the capacity. Spark VCores are shared by notebooks, Spark job definitions, lakehouse jobs and pipeline-triggered Spark workloads in the same capacity. When the capacity is fully used, Fabric may queue eligible jobs or reject new submissions with HTTP 430.
https://learn.microsoft.com/en-us/fabric/data-engineering/spark-job-concurrency-and-queueing
Go to Monitoring hub and cancel any active or stuck Spark jobs that are no longer needed then go to Workspace settings under Data Engineering/Science then Spark settings then Jobs / Job management and review running or queued jobs.
If this happens during pipeline runs, you should reduce parallel execution for example lower the ForEach batch count, stagger schedules or avoid starting many notebooks at the same time.
You can ask your Fabric capacity admin to check the Fabric Capacity Metrics app for throttling, overload time and which workspace/item is consuming the capacity.
https://learn.microsoft.com/en-us/fabric/enterprise/capacity-planning-troubleshoot-errors
If this happens frequently, consider moving to a larger Fabric SKU or enabling Autoscale Billing for Spark workloads, if available in your tenant. You also have autoscale, monitoring, stopping stalled sessions, job optimization and scaling up capacity as mitigation options.
https://learn.microsoft.com/en-us/fabric/data-engineering/troubleshoot-permissions-capacity
