diff options
author | Cheng Lian <lian@databricks.com> | 2015-06-02 13:32:13 -0700 |
---|---|---|
committer | Yin Huai <yhuai@databricks.com> | 2015-06-02 13:32:13 -0700 |
commit | 686a45f0b9c50ede2a80854ed6a155ee8a9a4f5c (patch) | |
tree | 5bf5776140d9906a2cc4c677df3de9313b6effe2 /sql/hive | |
parent | ad06727fe985ca243ebdaaba55cd7d35a4749d0a (diff) | |
download | spark-686a45f0b9c50ede2a80854ed6a155ee8a9a4f5c.tar.gz spark-686a45f0b9c50ede2a80854ed6a155ee8a9a4f5c.tar.bz2 spark-686a45f0b9c50ede2a80854ed6a155ee8a9a4f5c.zip |
[SPARK-8014] [SQL] Avoid premature metadata discovery when writing a HadoopFsRelation with a save mode other than Append
The current code references the schema of the DataFrame to be written before checking save mode. This triggers expensive metadata discovery prematurely. For save mode other than `Append`, this metadata discovery is useless since we either ignore the result (for `Ignore` and `ErrorIfExists`) or delete existing files (for `Overwrite`) later.
This PR fixes this issue by deferring metadata discovery after save mode checking.
Author: Cheng Lian <lian@databricks.com>
Closes #6583 from liancheng/spark-8014 and squashes the following commits:
1aafabd [Cheng Lian] Updates comments
088abaa [Cheng Lian] Avoids schema merging and partition discovery when data schema and partition schema are defined
8fbd93f [Cheng Lian] Fixes SPARK-8014
Diffstat (limited to 'sql/hive')
-rw-r--r-- | sql/hive/src/test/scala/org/apache/spark/sql/sources/hadoopFsRelationSuites.scala | 59 |
1 files changed, 45 insertions, 14 deletions
diff --git a/sql/hive/src/test/scala/org/apache/spark/sql/sources/hadoopFsRelationSuites.scala b/sql/hive/src/test/scala/org/apache/spark/sql/sources/hadoopFsRelationSuites.scala index af36fa6f1f..7409542674 100644 --- a/sql/hive/src/test/scala/org/apache/spark/sql/sources/hadoopFsRelationSuites.scala +++ b/sql/hive/src/test/scala/org/apache/spark/sql/sources/hadoopFsRelationSuites.scala @@ -17,6 +17,9 @@ package org.apache.spark.sql.sources +import java.io.File + +import com.google.common.io.Files import org.apache.hadoop.fs.Path import org.apache.spark.{SparkException, SparkFunSuite} @@ -453,6 +456,20 @@ abstract class HadoopFsRelationTest extends QueryTest with SQLTestUtils { } } } + + test("SPARK-7616: adjust column name order accordingly when saving partitioned table") { + val df = (1 to 3).map(i => (i, s"val_$i", i * 2)).toDF("a", "b", "c") + + df.write + .format(dataSourceName) + .mode(SaveMode.Overwrite) + .partitionBy("c", "a") + .saveAsTable("t") + + withTable("t") { + checkAnswer(table("t"), df.select('b, 'c, 'a).collect()) + } + } } class SimpleTextHadoopFsRelationSuite extends HadoopFsRelationTest { @@ -534,20 +551,6 @@ class ParquetHadoopFsRelationSuite extends HadoopFsRelationTest { } } - test("SPARK-7616: adjust column name order accordingly when saving partitioned table") { - val df = (1 to 3).map(i => (i, s"val_$i", i * 2)).toDF("a", "b", "c") - - df.write - .format("parquet") - .mode(SaveMode.Overwrite) - .partitionBy("c", "a") - .saveAsTable("t") - - withTable("t") { - checkAnswer(table("t"), df.select('b, 'c, 'a).collect()) - } - } - test("SPARK-7868: _temporary directories should be ignored") { withTempPath { dir => val df = Seq("a", "b", "c").zipWithIndex.toDF() @@ -563,4 +566,32 @@ class ParquetHadoopFsRelationSuite extends HadoopFsRelationTest { checkAnswer(read.format("parquet").load(dir.getCanonicalPath), df.collect()) } } + + test("SPARK-8014: Avoid scanning output directory when SaveMode isn't SaveMode.Append") { + withTempDir { dir => + val path = dir.getCanonicalPath + val df = Seq(1 -> "a").toDF() + + // Creates an arbitrary file. If this directory gets scanned, ParquetRelation2 will throw + // since it's not a valid Parquet file. + val emptyFile = new File(path, "empty") + Files.createParentDirs(emptyFile) + Files.touch(emptyFile) + + // This shouldn't throw anything. + df.write.format("parquet").mode(SaveMode.Ignore).save(path) + + // This should only complain that the destination directory already exists, rather than file + // "empty" is not a Parquet file. + assert { + intercept[RuntimeException] { + df.write.format("parquet").mode(SaveMode.ErrorIfExists).save(path) + }.getMessage.contains("already exists") + } + + // This shouldn't throw anything. + df.write.format("parquet").mode(SaveMode.Overwrite).save(path) + checkAnswer(read.format("parquet").load(path), df) + } + } } |