Java 21 support with Databricks JDBC driver

[NathanE]

Hello,

I was wondering if there was any timeline for Java 21 support with the Databricks JDBC driver (current version is 2.34).

One of the required change is to update the dependency to arrow to version 13.0 (current version is 9.0.0).

The current workaround is to disable arrow in the connection string with `EnableArrow=0` but I would like to still be able to use arrow serialization.

Regards

[ziv_cohen]

Hi,

Is there a committed date when this issue will be fixed? 

Toggling off Arrow creates 70% performance degradation in our benchmark and it's unacceptable.  

Waiting for your update.

Thanks in advance. 

[yunbodeng]

Try to apply the following JVM parameters:

--add-opens java.base/java.nio=ALL-UNNAMED

[yunbodeng]

--add-opens java.base/java.nio=ALL-UNNAMED

[Fabich]

Hello @Retired_mod 

Any update on this topic of Java 21 ? Any timeline ?

Our clients really want to upgrade to Java 21 and we don't want to disable Arrow for performance reasons

[loicmathieu]

Hi,
Any news on this topic?
We recently upgraded to Java 21 (which is out since 1 months) and we had to disable Arrow which is not good for performance.

[Oleksandr_K]

 Hi, this issue blocks us from updating Java 21, which is critical for addressing security issues. Please, update on the timelines, please. 

[151640]

Regarding EnableArrow, it is not discussed in the PDF SIMBA distributes with the driver, it is buried in the release notes.txt file. 

It may be required to avoid corrupted character values being returned from the driver. This can occur with JRE 8. 

Hopefully the SIMBA/Databricks PM for the driver will review docs etc.

[yunbodeng]

You can download the latest Databricks JDBC here (https://www.databricks.com/spark/jdbc-drivers-download) in which the latest Arrow version is 17 (https://databricks-bi-artifacts.s3.us-east-2.amazonaws.com/simbaspark-drivers/jdbc/2.7.1/docs/releas...).

We encourage you not to turn off Arrow, which has severe performance degradation. Instead, you can apply the following JVM parameters, as it is also called out in the driver configuration and installation guide (https://docs.databricks.com/aws/en/assets/files/Databricks-JDBC-Driver-Install-and-Configuration-Gui... or you can find it embedded in the driver package).

--add-opens java.base/java.nio=ALL-UNNAMED