[SPARK-4908][SQL] Prevent multiple concurrent hive native commands

This is just a quick fix that locks when calling `runHive`.  If we can find a way to avoid the error without a global lock that would be better.

Author: Michael Armbrust <[email protected]>

Closes #3834 from marmbrus/hiveConcurrency and squashes the following commits:

bf25300 [Michael Armbrust] prevent multiple concurrent hive native commands

Author: marmbrus

sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveContext.scala

@@ -284,7 +284,7 @@ class HiveContext(sc: SparkContext) extends SQLContext(sc) {
    * Execute the command using Hive and return the results as a sequence. Each element
    * in the sequence is one row.
    */
-  protected def runHive(cmd: String, maxRows: Int = 1000): Seq[String] = {
+  protected def runHive(cmd: String, maxRows: Int = 1000): Seq[String] = synchronized {
     try {
       val cmd_trimmed: String = cmd.trim()
       val tokens: Array[String] = cmd_trimmed.split("\\s+")

sql/hive/src/test/scala/org/apache/spark/sql/hive/execution/HiveQuerySuite.scala

@@ -56,6 +56,13 @@ class HiveQuerySuite extends HiveComparisonTest with BeforeAndAfter {
     Locale.setDefault(originalLocale)
   }
 
+  test("SPARK-4908: concurent hive native commands") {
+    (1 to 100).par.map { _ =>
+      sql("USE default")
+      sql("SHOW TABLES")
+    }
+  }
+
   createQueryTest("constant object inspector for generic udf",
     """SELECT named_struct(
       lower("AA"), "10",