What causes LIBRARIES_INSTALL_FAILED?

A PyPI or Maven package version does not exist or was yanked from the registry. A network timeout occurred while fetching the package from an external registry. A DBFS or S3 wheel path is incorrect or the file has been deleted. Version conflicts between a cluster-scoped library and a job-scoped library

How do I fix LIBRARIES_INSTALL_FAILED?

Step 1: Check the cluster's Libraries tab event log for the specific package and error message.. Step 2: Verify the package version exists on PyPI or Maven Central; pin to a known-good version.. Step 3: For private wheel files, confirm the DBFS or cloud storage path is still valid and accessible.. Step 4: If network timeouts are the cause, upload the wheel to DBFS and install from there instead of from PyPI.. Step 5: Consider moving library installation to an init script to catch errors before the driver starts.

High severityconfiguration

Power BI Refresh Error:
LIBRARIES_INSTALL_FAILED

What does this error mean?

A cluster failed to start or a job was aborted because one or more libraries could not be installed during cluster initialisation.

Common causes

1A PyPI or Maven package version does not exist or was yanked from the registry
2A network timeout occurred while fetching the package from an external registry
3A DBFS or S3 wheel path is incorrect or the file has been deleted
4Version conflicts between a cluster-scoped library and a job-scoped library

How to fix it

1Step 1: Check the cluster's Libraries tab event log for the specific package and error message.
2Step 2: Verify the package version exists on PyPI or Maven Central; pin to a known-good version.
3Step 3: For private wheel files, confirm the DBFS or cloud storage path is still valid and accessible.
4Step 4: If network timeouts are the cause, upload the wheel to DBFS and install from there instead of from PyPI.
5Step 5: Consider moving library installation to an init script to catch errors before the driver starts.

Frequently asked questions

Why did a library that worked yesterday suddenly fail?

PyPI packages can be yanked or updated. If you pinned to a major version without a minor pin (e.g., pandas>=1.0 instead of pandas==1.5.3), a new release with breaking changes or a yanked version can break installation overnight.

Can I pre-bake libraries into a custom Docker image to avoid this?

Yes — Databricks supports custom Docker containers for clusters (on supported runtimes). Pre-baking libraries eliminates install-time failures and speeds up cluster start.