In order to better support JDK11, we need to restructure the deployment and stop building uber (shaded) jars.
To do this, we'll use the dependency tree and include the required jars with the deployment. There will be two different deployments: one that you would deploy to a CDP/Hadoop configured node that has all the hadoop libraries. This is recommended since we'll use libs that are most compatible with the current distro. The second option supports deployment on a node with no hadoop libraries. In this case, we'll package a determined version of the cdp libraries.
In order to better support JDK11, we need to restructure the deployment and stop building uber (shaded) jars.
To do this, we'll use the dependency tree and include the required jars with the deployment. There will be two different deployments: one that you would deploy to a CDP/Hadoop configured node that has all the hadoop libraries. This is recommended since we'll use libs that are most compatible with the current distro. The second option supports deployment on a node with no hadoop libraries. In this case, we'll package a determined version of the cdp libraries.