Datbricks JDK 17 upgrade error

[prith]

We tried upgrading to JDK 17

Using Spark version 3.0.5 and runtime 14.3 LTS

Getting this exception using parallelstream()

With Java 17 I am not able to parallel process different partitions at the same time.  This means when there is more than 1 partition to process, Java11 (which allows parallel processing) takes ~75 minutes while Java 17 takes  ~150 minutes.The exception I face in Java 17 when I use list.parallelStream().foreach() is (workflow that faced this exception:

SecurityException: java.lang.SecurityException: setContextClassLoader Caused by: SecurityException: setContextClassLoader at java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method) at java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:77) at java.base/jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45) at java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:499) at java.base/java.lang.reflect.Constructor.newInstance(Constructor.java:480) at java.base/java.util.concurrent.ForkJoinTask.getThrowableException(ForkJoinTask.java:562) at java.base/java.util.concurrent.ForkJoinTask.reportException(ForkJoinTask.java:591) at java.base/java.util.concurrent.ForkJoinTask.invoke(ForkJoinTask.java:689) at java.base/java.util.stream.ForEachOps$ForEachOp.evaluateParallel(ForEachOps.java:159) at java.base/java.util.stream.ForEachOps$ForEachOp$OfRef.evaluateParallel(ForEachOps.java:173) at java.base/java.util.stream.AbstractPipeline.evaluate(AbstractPipeline.java:233) at java.base/java.util.stream.ReferencePipeline.forEach(ReferencePipeline.java:596) at java.base/java.util.stream.ReferencePipeline$Head.forEach(ReferencePipeline.java:765)

[raphaelblg · Accepted solution]

Hello @prith,

The Databricks DBR is bundled with Java already. For DBR 14.3 the system environment is:

• Operating System: Ubuntu 22.04.3 LTS

• Java: Zulu 8.74.0.17-CA-linux64

• Scala: 2.12.15

• Python: 3.10.12

• R: 4.3.1

• Delta Lake: 3.1.0

Changing the DBR Java version is not supported.

[prith]

Well we use spark_version: "14.3.x-scala2.12" along with

spark_env_vars:
JNAME: "zulu17-ca-arm64"

This is a pure java jar based workflow - the entire code and jar is in Java - nothing to do with Scala / Python

[raphaelblg]

@prith, Databricks DBR includes a specific Java version, and altering it is not a supported scenario by Databricks. I hope this information is helpful.

[prith]

I'm sorry - I dont understand - We are not trying to alter the JDK version.. Isn't JDK 17 supported or not?

[raphaelblg]

Hello @prith ,

Currently, Java 17 is not available on Databricks runtimes (DBR). If JDK 17 becomes available on Databricks in the future, it will be included in the DBR. You can refer to the Databricks Runtime release notes versions and compatibility .

[prith]

Anyways - thanks for your response - We found a workaround for this error and JDK 17 is actually working - it appears faster than JDK 8

[duliu]

Hi @prith,

I'm also trying to use Java 17 or 11 in Databricks clusters. Are you using the environment variable `JNAME=zulu17-ca-amd64` as mentioned in https://docs.databricks.com/en/dev-tools/sdk-java.html#create-a-cluster-that-uses-jdk-17 ? Could you share your experience and workaround? Much appreciated!