Medium severityfabric
Power BI Refresh Error:
UserErrorJniException
What does this error mean?
A Fabric Data Factory copy activity writing Parquet files failed because the Java Virtual Machine (JVM) could not start on the self-hosted integration runtime. This is caused by invalid global JVM arguments (typically an incorrect _JAVA_OPTIONS system variable) on the SHIR host machine.
Common causes
- 1The _JAVA_OPTIONS system environment variable on the SHIR host contains invalid or conflicting arguments
- 2JVM memory settings (Xms/Xmx) set below the required threshold (minimum 8 GB for Parquet operations with Fabric Data Factory)
- 3A previous installation left conflicting JVM configuration in system environment variables
- 4Multiple Java installations on the SHIR host with conflicting PATH entries
How to fix it
- 1Sign in to the machine hosting each SHIR node and check the _JAVA_OPTIONS system environment variable
- 2Set _JAVA_OPTIONS to '-Xms256m -Xmx16g' (ensure Xmx is at least 8 GB)
- 3After updating the environment variable, restart all SHIR nodes
- 4Rerun the pipeline after restarting the integration runtime
Beyond the docs
Common practitioner solutions not covered in the official documentation.
- 1If multiple Java versions are installed, ensure the SHIR is using the correct JRE (check JAVA_HOME environment variable)
Official documentation: https://learn.microsoft.com/en-us/fabric/data-factory/connector-troubleshoot-parquet-format