Github user yhuai commented on a diff in the pull request:
https://github.com/apache/spark/pull/9453#discussion_r44047128
--- Diff:
sql/core/src/test/scala/org/apache/spark/sql/CachedTableSuite.scala ---
@@ -375,53 +375,98 @@ class CachedTableSuite extends QueryTest with
SharedSQLContext {
sql("SELECT key, count(*) FROM orderedTable GROUP BY key ORDER BY
key"),
sql("SELECT key, count(*) FROM testData3x GROUP BY key ORDER BY
key").collect())
sqlContext.uncacheTable("orderedTable")
+ sqlContext.dropTempTable("orderedTable")
// Set up two tables distributed in the same way. Try this with the
data distributed into
// different number of partitions.
for (numPartitions <- 1 until 10 by 4) {
- testData.repartition(numPartitions, $"key").registerTempTable("t1")
- testData2.repartition(numPartitions, $"a").registerTempTable("t2")
+ withTempTable("t1", "t2") {
+ testData.repartition(numPartitions, $"key").registerTempTable("t1")
+ testData2.repartition(numPartitions, $"a").registerTempTable("t2")
+ sqlContext.cacheTable("t1")
+ sqlContext.cacheTable("t2")
+
+ // Joining them should result in no exchanges.
+ verifyNumExchanges(sql("SELECT * FROM t1 t1 JOIN t2 t2 ON t1.key =
t2.a"), 0)
+ checkAnswer(sql("SELECT * FROM t1 t1 JOIN t2 t2 ON t1.key = t2.a"),
+ sql("SELECT * FROM testData t1 JOIN testData2 t2 ON t1.key =
t2.a"))
+
+ // Grouping on the partition key should result in no exchanges
+ verifyNumExchanges(sql("SELECT count(*) FROM t1 GROUP BY key"), 0)
+ checkAnswer(sql("SELECT count(*) FROM t1 GROUP BY key"),
+ sql("SELECT count(*) FROM testData GROUP BY key"))
+
+ sqlContext.uncacheTable("t1")
+ sqlContext.uncacheTable("t2")
+ }
+ }
+
+ // Distribute the tables into non-matching number of partitions. Need
to shuffle one side.
+ withTempTable("t1", "t2") {
+ testData.repartition(6, $"key").registerTempTable("t1")
+ testData2.repartition(3, $"a").registerTempTable("t2")
sqlContext.cacheTable("t1")
sqlContext.cacheTable("t2")
- // Joining them should result in no exchanges.
- verifyNumExchanges(sql("SELECT * FROM t1 t1 JOIN t2 t2 ON t1.key =
t2.a"), 0)
- checkAnswer(sql("SELECT * FROM t1 t1 JOIN t2 t2 ON t1.key = t2.a"),
- sql("SELECT * FROM testData t1 JOIN testData2 t2 ON t1.key =
t2.a"))
+ val query = sql("SELECT key, value, a, b FROM t1 t1 JOIN t2 t2 ON
t1.key = t2.a")
+ verifyNumExchanges(query, 1)
+
assert(query.queryExecution.executedPlan.outputPartitioning.numPartitions === 6)
+ checkAnswer(
+ query,
+ testData.join(testData2, $"key" === $"a").select($"key", $"value",
$"a", $"b"))
+ sqlContext.uncacheTable("t1")
+ sqlContext.uncacheTable("t2")
+ }
- // Grouping on the partition key should result in no exchanges
- verifyNumExchanges(sql("SELECT count(*) FROM t1 GROUP BY key"), 0)
- checkAnswer(sql("SELECT count(*) FROM t1 GROUP BY key"),
- sql("SELECT count(*) FROM testData GROUP BY key"))
+ // One side of join is not partitioned in the desired way. Need to
shuffle one side.
+ withTempTable("t1", "t2") {
+ testData.repartition(6, $"value").registerTempTable("t1")
+ testData2.repartition(6, $"a").registerTempTable("t2")
+ sqlContext.cacheTable("t1")
+ sqlContext.cacheTable("t2")
+ val query = sql("SELECT key, value, a, b FROM t1 t1 JOIN t2 t2 ON
t1.key = t2.a")
+ verifyNumExchanges(query, 1)
+
assert(query.queryExecution.executedPlan.outputPartitioning.numPartitions === 6)
+ checkAnswer(
+ query,
+ testData.join(testData2, $"key" === $"a").select($"key", $"value",
$"a", $"b"))
sqlContext.uncacheTable("t1")
sqlContext.uncacheTable("t2")
- sqlContext.dropTempTable("t1")
- sqlContext.dropTempTable("t2")
}
- // Distribute the tables into non-matching number of partitions. Need
to shuffle.
- testData.repartition(6, $"key").registerTempTable("t1")
- testData2.repartition(3, $"a").registerTempTable("t2")
- sqlContext.cacheTable("t1")
- sqlContext.cacheTable("t2")
+ withTempTable("t1", "t2") {
--- End diff --
In this case, we will not Exchange. But, let me double check it and add a
test case.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
