Skip to content

[SPARK-21599][SQL] Collecting column statistics for datasource tables may fail with java.util.NoSuchElementException #18804

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
dilipbiswal wants to merge 7 commits into from
Closed

[SPARK-21599][SQL] Collecting column statistics for datasource tables may fail with java.util.NoSuchElementException #18804

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
0afefd5
[SPARK-21599] Collecting column statistics for datasource tables may …
dilipbiswal Jul 27, 2017
420be2f
Code review comments
dilipbiswal Aug 2, 2017
80b0073
Add new test for hive serde tables
dilipbiswal Aug 2, 2017
385dfbe
Code review comments.
dilipbiswal Aug 3, 2017
7a8fa2c
code review
dilipbiswal Aug 3, 2017
c1ab569
style
dilipbiswal Aug 3, 2017
a120357
Enhance test
dilipbiswal Aug 3, 2017
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
9 changes: 8 additions & 1 deletion sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveExternalCatalog.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -642,8 +642,15 @@ private[spark] class HiveExternalCatalog(conf: SparkConf, hadoopConf: Configurat
if (stats.get.rowCount.isDefined) {
statsProperties += STATISTICS_NUM_ROWS -> stats.get.rowCount.get.toString()
}

// For datasource tables the data schema is stored in the table properties.
val schema = rawTable.properties.get(DATASOURCE_PROVIDER) match {
case Some(provider) => getSchemaFromTableProperties(rawTable)
case _ => rawTable.schema

@gatorsmile gatorsmile Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For Hive serde tables that were created by Spark 2.1 or later, we should still restore it from table properties.

@gatorsmile gatorsmile Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

See the code in https://github.com/apache/spark/blob/master/sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveExternalCatalog.scala#L755-L758

@viirya viirya Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Maybe call restoreTableMetadata to avoid duplicate logic.

@dilipbiswal dilipbiswal Aug 2, 2017

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

should we call getTable().schema or you guys think its too verbose ?
val schema = getTable(db, table).schema ?

@dilipbiswal dilipbiswal Aug 2, 2017

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@viirya Actually, we do have a raw table here.. so i will just call restoreTableMetadata. Thanks a lot @gatorsmile and @viirya

@viirya viirya Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

You still need rawTable. Call getTable will incur another metastore access.

@dilipbiswal dilipbiswal Aug 2, 2017

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@viirya right. I agree. I was saying that we do have a raw table from a prior call. So here we just pass that to restoreTableMetadata like you suggested.

@viirya viirya Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I saw your comment #18804 (comment) after post #18804 (comment). :)

}

val colNameTypeMap: Map[String, DataType] =
rawTable.schema.fields.map(f => (f.name, f.dataType)).toMap
schema.fields.map(f => (f.name, f.dataType)).toMap
stats.get.colStats.foreach { case (colName, colStat) =>
colStat.toMap(colName, colNameTypeMap(colName)).foreach { case (k, v) =>
statsProperties += (columnStatKeyPropName(colName, k) -> v)
Expand Down
20 changes: 20 additions & 0 deletions sql/hive/src/test/scala/org/apache/spark/sql/hive/StatisticsSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -117,6 +117,26 @@ class StatisticsSuite extends StatisticsCollectionTestBase with TestHiveSingleto
}
}

test("analyze non hive compatible datasource tables") {
val table = "parquet_tab"
withTable(table) {
sql(
s"""
|CREATE TABLE $table (a int, b int)
|USING parquet
|OPTIONS (skipHiveMetadata true)
""".stripMargin)
sql(s"insert into $table values (1, 1)")
sql(s"insert into $table values (2, 1)")

@viirya viirya Aug 2, 2017

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: minor style issue. INSERT INTO...

sql(s"ANALYZE TABLE $table COMPUTE STATISTICS FOR COLUMNS a, b")
val fetchedStats0 =
checkTableStats(table, hasSizeInBytes = true, expectedRowCounts = Some(2))
assert(fetchedStats0.get.colStats == Map(
"a" -> ColumnStat(2, Some(1), Some(2), 0, 4, 4),
"b" -> ColumnStat(1, Some(1), Some(1), 0, 4, 4)))
}
}

test("SPARK-21079 - analyze table with location different than that of individual partitions") {
val tableName = "analyzeTable_part"
withTable(tableName) {
Expand Down