What causes STORAGE_DOWNLOAD_FAILURE_SLOW?

Throttled or degraded performance from the cloud storage service (S3, ADLS) in the region. An init script or library file is stored in a bucket with insufficient bandwidth for concurrent cluster launches. Network bottleneck between the cluster subnet and the cloud storage endpoint. A very large custom library or Docker image causing download timeouts on cluster startup. A cloud provider storage service degradation event affecting the region

How do I fix STORAGE_DOWNLOAD_FAILURE_SLOW?

Check the cloud provider status page for active storage performance incidents in the region. Move large libraries or init scripts to a storage bucket in the same region as the Databricks workspace. Use Databricks-managed libraries (installed via the cluster library settings) instead of downloading from external URLs at startup. Reduce the size of init scripts and custom libraries. If the error is persistent, try a different AZ or check VPC endpoint configuration for storage traffic

Medium severitycluster

Power BI Refresh Error:
STORAGE_DOWNLOAD_FAILURE_SLOW

What does this error mean?

A Databricks cluster terminated because downloading a required artifact from cloud storage (such as a library, init script, or Spark binary) was too slow and exceeded the timeout threshold.

Common causes

1Throttled or degraded performance from the cloud storage service (S3, ADLS) in the region
2An init script or library file is stored in a bucket with insufficient bandwidth for concurrent cluster launches
3Network bottleneck between the cluster subnet and the cloud storage endpoint
4A very large custom library or Docker image causing download timeouts on cluster startup
5A cloud provider storage service degradation event affecting the region

How to fix it

1Check the cloud provider status page for active storage performance incidents in the region
2Move large libraries or init scripts to a storage bucket in the same region as the Databricks workspace
3Use Databricks-managed libraries (installed via the cluster library settings) instead of downloading from external URLs at startup
4Reduce the size of init scripts and custom libraries
5If the error is persistent, try a different AZ or check VPC endpoint configuration for storage traffic

Frequently asked questions

Does this error affect running clusters?

No — STORAGE_DOWNLOAD_FAILURE_SLOW occurs during cluster startup when required artifacts are downloaded. Already-running clusters are not affected.

Can I speed up cluster startup to avoid this error?

Yes — use instance pools with pre-warmed nodes (cluster bootstrap already done), minimize init scripts, and store all artifacts in a storage bucket in the same region as the workspace.