ODIn/spark-instrumented-optimizer
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[SPARK-19359][SQL] Revert Clear useless path after rename a partition with upper-case by HiveExternalCatalog

Browse source 
### What changes were proposed in this pull request?

This PR is to revert the changes made in https://github.com/apache/spark/pull/16700. It could cause the data loss after partition rename, because we have a bug in the file renaming.

Not all the OSs have the same behaviors. For example, on mac OS, if we renaming a path from `.../tbl/a=5/b=6` to `.../tbl/A=5/B=6`. The result is `.../tbl/a=5/B=6`. The expected result is `.../tbl/A=5/B=6`. Thus, renaming on mac OS is not recursive. However, the systems used in Jenkin does not have such an issue. Although this PR is not the root cause, it exposes an existing issue on the code `tablePath.getFileSystem(hadoopConf).rename(wrongPath, rightPath)`

---

Hive metastore is not case preserving and keep partition columns with lower case names.

If SparkSQL create a table with upper-case partion name use HiveExternalCatalog, when we rename partition, it first call the HiveClient to renamePartition, which will create a new lower case partition path, then SparkSql rename the lower case path to the upper-case.

while if the renamed partition contains more than one depth partition ,e.g. A=1/B=2, hive renamePartition change to a=1/b=2, then SparkSql rename it to A=1/B=2, but the a=1 still exists in the filesystem, we should also delete it.

### How was this patch tested?
N/A

Author: gatorsmile <gatorsmile@gmail.com>

Closes #16728 from gatorsmile/revert-pr-16700.
This commit is contained in:
gatorsmile 2017-01-28 13:32:30 -08:00
parent 42ad93b2c9
commit cfcfc92f7b
2 changed files with 0 additions and 71 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

35
sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveExternalCatalog.scala
View file

@ -839,26 +839,6 @@ private[spark] class HiveExternalCatalog(conf: SparkConf, hadoopConf: Configurat
spec.map { case (k, v) => partCols.find(_.equalsIgnoreCase(k)).get -> v }
}
/**
* The partition path created by Hive is in lowercase, while Spark SQL will
* rename it with the partition name in partitionColumnNames, and this function
* returns the extra lowercase path created by Hive, and then we can delete it.
* e.g. /path/A=1/B=2/C=3 is changed to /path/A=4/B=5/C=6, this function returns
* /path/a=4
*/
def getExtraPartPathCreatedByHive(
spec: TablePartitionSpec,
partitionColumnNames: Seq[String],
tablePath: Path): Path = {
val partColumnNames = partitionColumnNames
.take(partitionColumnNames.indexWhere(col => col.toLowerCase != col) + 1)
.map(_.toLowerCase)
ExternalCatalogUtils.generatePartitionPath(lowerCasePartitionSpec(spec),
partColumnNames, tablePath)
}
override def createPartitions(
db: String,
table: String,
@ -919,21 +899,6 @@ private[spark] class HiveExternalCatalog(conf: SparkConf, hadoopConf: Configurat
spec, partitionColumnNames, tablePath)
try {
tablePath.getFileSystem(hadoopConf).rename(wrongPath, rightPath)
// If the newSpec contains more than one depth partition, FileSystem.rename just deletes
// the leaf(i.e. wrongPath), we should check if wrongPath's parents need to be deleted.
// For example, give a newSpec 'A=1/B=2', after calling Hive's client.renamePartitions,
// the location path in FileSystem is changed to 'a=1/b=2', which is wrongPath, then
// although we renamed it to 'A=1/B=2', 'a=1/b=2' in FileSystem is deleted, but 'a=1'
// is still exists, which we also need to delete
val delHivePartPathAfterRename = getExtraPartPathCreatedByHive(
spec,
partitionColumnNames,
tablePath)
if (delHivePartPathAfterRename != wrongPath) {
tablePath.getFileSystem(hadoopConf).delete(delHivePartPathAfterRename, true)
}
} catch {
case e: IOException => throw new SparkException(
s"Unable to rename partition path from $wrongPath to $rightPath", e)

36
sql/hive/src/test/scala/org/apache/spark/sql/hive/PartitionProviderCompatibilitySuite.scala
View file

@ -19,11 +19,8 @@ package org.apache.spark.sql.hive
import java.io.File
import org.apache.hadoop.fs.Path
import org.apache.spark.metrics.source.HiveCatalogMetrics
import org.apache.spark.sql.{AnalysisException, QueryTest}
import org.apache.spark.sql.catalyst.TableIdentifier
import org.apache.spark.sql.hive.test.TestHiveSingleton
import org.apache.spark.sql.internal.SQLConf
import org.apache.spark.sql.test.SQLTestUtils
@ -484,37 +481,4 @@ class PartitionProviderCompatibilitySuite
assert(spark.sql("show partitions test").count() == 5)
}
}
test("partition path created by Hive should be deleted after renamePartitions with upper-case") {
withTable("t", "t1", "t2") {
Seq((1, 2, 3)).toDF("id", "A", "B").write.partitionBy("A", "B").saveAsTable("t")
spark.sql("alter table t partition(A=2, B=3) rename to partition(A=4, B=5)")
var table = spark.sessionState.catalog.getTableMetadata(TableIdentifier("t"))
var extraHivePath = new Path(table.location + "/a=4")
assert(!extraHivePath.getFileSystem(spark.sessionState.newHadoopConf())
.exists(extraHivePath), "partition path created by Hive should be deleted " +
"after renamePartitions with upper-case")
Seq((1, 2, 3, 4)).toDF("id", "A", "B", "C").write.partitionBy("A", "B", "C").saveAsTable("t1")
spark.sql("alter table t1 partition(A=2, B=3, C=4) rename to partition(A=5, B=6, C=7)")
table = spark.sessionState.catalog.getTableMetadata(TableIdentifier("t1"))
extraHivePath = new Path(table.location + "/a=5")
assert(!extraHivePath.getFileSystem(spark.sessionState.newHadoopConf())
.exists(extraHivePath), "partition path created by Hive should be deleted " +
"after renamePartitions with upper-case")
Seq((1, 2, 3, 4)).toDF("id", "a", "B", "C").write.partitionBy("a", "B", "C").saveAsTable("t2")
spark.sql("alter table t2 partition(a=2, B=3, C=4) rename to partition(a=4, B=5, C=6)")
table = spark.sessionState.catalog.getTableMetadata(TableIdentifier("t2"))
val partPath = new Path(table.location + "/a=4")
assert(partPath.getFileSystem(spark.sessionState.newHadoopConf())
.exists(partPath), "partition path of lower-case partition name should not be deleted")
extraHivePath = new Path(table.location + "/a=4/b=5")
assert(!extraHivePath.getFileSystem(spark.sessionState.newHadoopConf())
.exists(extraHivePath), "partition path created by Hive should be deleted " +
"after renamePartitions with upper-case")
}
}
}