Issue with Oracle Linked Service v2.0 Causing Parquet Java Invocation

[Advith]

Hi all,
After upgrading my Oracle linked service to version 2.0 in Azure Synapse Analytics, I’m encountering the following error during a copy activity to Parquet:

ErrorCode=ParquetJavaInvocationException, Message=java.lang.ArrayIndexOutOfBoundsException:126
total entry:1
com.microsoft.datatransfer.bridge.parquet.ParquetWriterBuilderBridge.addDecimalColumn(ParquetWriterBuilderBridge.java:107)

The pipeline fails when writing decimal columns to Parquet format. This didn’t happen with version 1.0 of the Oracle connector. I’ve tried adjusting source mappings and data types, but the issue persists. Has anyone faced this or found a workaround? Switching the sink format isn't ideal for our use case, and I’d prefer not to downgrade connectors unless absolutely necessary.
Any insights or fixes from the community or Microsoft team would be greatly appreciated!

Thanks in advance,

A.R

[Advith]

Hi Ganesh,

Thank you for the follow-up.

To clarify, I’m running the pipeline within Azure Synapse Analytics, specifically using Synapse Studio, not within a Microsoft Fabric workspace. The pipeline is part of an Azure resource group and not associated with any Fabric item like a Lakehouse or Warehouse.

Let me know if you'd like any additional context or details to help route this appropriately.

[v-ssriganesh]

Hello @Advith,
Thank you for the clarification.

Since you're running the pipeline within Azure Synapse Analytics (via Synapse Studio, not Fabric), your scenario falls under the Azure Synapse / Data Factory domain, not Microsoft Fabric.

To ensure your issue with the Oracle linked service v2.0 and Parquet sink gets the right visibility and technical support, I recommend posting your question in the appropriate Microsoft Q&A forums:

Azure Synapse Analytics Q&A: Azure Synapse Analytics - Microsoft Q&A

Azure Data Factory Q&A:  Azure Data Factory - Microsoft Q&A

If this information is helpful, please “Accept as solution” and give a "kudos" to assist other community members in resolving similar issues more efficiently.
Thank you.

[v-ssriganesh]

Hi @Advith,
Thank you for reaching out to the Microsoft Fabric Forum Community.

To help us direct your question to the right support path, could you please confirm the environment where you're running this pipeline?

Specifically:

1. Are you working inside a Microsoft Fabric workspace (e.g: using Fabric Data Pipelines, Lakehouse, or Warehouse)?
2. Or are you using Azure Data Factory or Azure Synapse Analytics in their standalone portals (like ADF Studio or Synapse Studio)?
3. If possible, can you confirm whether the pipeline is part of a Fabric item (like a Lakehouse pipeline), or running from an Azure resource group?

This will help us identify whether your issue is related to Fabric, or if it should be handled in the Azure Data Factory/Synapse forum, where the Oracle connector v2.0 and Parquet error are more commonly supported.