[SPARK-4412][SQL] Fix Spark's control of Parquet logging.

sql/core/src/main/scala/org/apache/spark/sql/parquet/ParquetRelation.scala

@@ -84,6 +84,21 @@ private[sql] case class ParquetRelation(
 private[sql] object ParquetRelation {
 
   def enableLogForwarding() {
+    // Note: the parquet.Log class has a static initializer that
+    // sets the java.util.logging Logger for "parquet". This
+    // checks first to see if there's any handlers already set
+    // and if not it creates them. If this method executes prior
+    // to that class being loaded then:
+    //  1) there's no handlers installed so there's none to 
+    // remove. But when it IS finally loaded the desired affect
+    // of removing them is circumvented.
+    //  2) The parquet.Log static initializer calls setUseParentHanders(false)
+    // undoing the attempt to override the logging here.
+    //
+    // Therefore we need to force the class to be loaded.
+    // This should really be resolved by Parquet.
+    Class.forName(classOf[parquet.Log].getName())
+
     // Note: Logger.getLogger("parquet") has a default logger
     // that appends to Console which needs to be cleared.
     val parquetLogger = java.util.logging.Logger.getLogger("parquet")