Spark Sql executions do not use thread local jobgroup

val sparkThreadLocal: SparkSession = DataCurator.spark.newSession()

sparkThreadLocal.sparkContext.setJobGroup("<id>", "<description>")

OR

sparkThreadLocal.sparkContext.setLocalProperty("spark.job.description", "<id>")
sparkThreadLocal.sparkContext.setLocalProperty("spark.jobGroup.id", "<description>")

The jobgroup property works fine for spark jobs/stages created by spark dataframe operations but in case of sparksql the jobgroup is randomly assigned to stages or is null sometimes.