[SPARK-12275][SQL] No plan for BroadcastHint in some condition

When SparkStrategies.BasicOperators's "case BroadcastHint(child) => apply(child)" is hit, it only recursively invokes BasicOperators.apply with this "child". It makes many strategies have no change to process this plan, which probably leads to "No plan" issue, so we use planLater to go through all strategies. https://issues.apache.org/jira/browse/SPARK-12275 Author: yucai <[email protected]> Closes #10265 from yucai/broadcast_hint.
apache · Dec 14, 2015 · ed87f6d · ed87f6d
1 parent 834e714
commit ed87f6d
Show file tree

Hide file tree

Showing 2 changed files with 8 additions and 1 deletion.
diff --git a/sql/core/src/main/scala/org/apache/spark/sql/execution/SparkStrategies.scala b/sql/core/src/main/scala/org/apache/spark/sql/execution/SparkStrategies.scala
@@ -364,7 +364,7 @@ private[sql] abstract class SparkStrategies extends QueryPlanner[SparkPlan] {
       case e @ EvaluatePython(udf, child, _) =>
         BatchPythonEvaluation(udf, e.output, planLater(child)) :: Nil
       case LogicalRDD(output, rdd) => PhysicalRDD(output, rdd, "ExistingRDD") :: Nil
-      case BroadcastHint(child) => apply(child)
+      case BroadcastHint(child) => planLater(child) :: Nil
       case _ => Nil
     }
   }

diff --git a/sql/core/src/test/scala/org/apache/spark/sql/DataFrameJoinSuite.scala b/sql/core/src/test/scala/org/apache/spark/sql/DataFrameJoinSuite.scala
@@ -120,5 +120,12 @@ class DataFrameJoinSuite extends QueryTest with SharedSQLContext {
 
     // planner should not crash without a join
     broadcast(df1).queryExecution.executedPlan
+
+    // SPARK-12275: no physical plan for BroadcastHint in some condition
+    withTempPath { path =>
+      df1.write.parquet(path.getCanonicalPath)
+      val pf1 = sqlContext.read.parquet(path.getCanonicalPath)
+      assert(df1.join(broadcast(pf1)).count() === 4)
+    }
   }
 }