Open
Description
Describe the proposal
When a SparkConf is invalid for launching our engines, it is better to fail fast instead of waiting for Spark to check while engine bootstrapping.
Motivation
- Save time to create an invalid engine
- Avoid possible hang
- better error message
Here is an example I did for Spark to validate its own configs when talking to Kubernetes. apache/spark#32610
Task list
Generally speaking, we can have 1) a tool in kyuubi-ctl
to check server/engine conf in configuration files are valid, 2) The logic of the tool can also be called when engine bootstrapping for the configurations from the client side