You are viewing a plain text version of this content. The canonical link for it is here.
Posted to reviews@spark.apache.org by GitBox <gi...@apache.org> on 2022/07/12 06:18:59 UTC
[GitHub] [spark] wangyum commented on a diff in pull request #37163: [SPARK-39750][SQL] Enable `spark.sql.cbo.enabled` by default
wangyum commented on code in PR #37163:
URL: https://github.com/apache/spark/pull/37163#discussion_r918586900
##########
sql/core/src/test/scala/org/apache/spark/sql/StatisticsCollectionSuite.scala:
##########
@@ -353,48 +353,65 @@ class StatisticsCollectionSuite extends StatisticsCollectionTestBase with Shared
test("invalidation of tableRelationCache after inserts") {
val table = "invalidate_catalog_cache_table"
- Seq(false, true).foreach { autoUpdate =>
- withSQLConf(SQLConf.AUTO_SIZE_UPDATE_ENABLED.key -> autoUpdate.toString) {
- withTable(table) {
- spark.range(100).write.saveAsTable(table)
- sql(s"ANALYZE TABLE $table COMPUTE STATISTICS")
- spark.table(table)
- val initialSizeInBytes = getTableFromCatalogCache(table).stats.sizeInBytes
- spark.range(100).write.mode(SaveMode.Append).saveAsTable(table)
- spark.table(table)
- assert(getTableFromCatalogCache(table).stats.sizeInBytes == 2 * initialSizeInBytes)
+ Seq(false, true).foreach { cboEnabled =>
+ Seq(false, true).foreach { autoUpdate =>
+ withSQLConf(
+ SQLConf.CBO_ENABLED.key -> cboEnabled.toString,
+ SQLConf.AUTO_SIZE_UPDATE_ENABLED.key -> autoUpdate.toString) {
+ withTable(table) {
+ spark.range(100).write.saveAsTable(table)
+ sql(s"ANALYZE TABLE $table COMPUTE STATISTICS")
+ spark.table(table)
+ val initialSizeInBytes = getTableFromCatalogCache(table).stats.sizeInBytes
+ spark.range(100).write.mode(SaveMode.Append).saveAsTable(table)
+ spark.table(table)
+ if (!cboEnabled) {
+ assert(getTableFromCatalogCache(table).stats.sizeInBytes === 2 * initialSizeInBytes)
+ } else {
+ assert(getTableFromCatalogCache(table).stats.sizeInBytes > initialSizeInBytes)
+ }
+ }
}
}
}
}
test("invalidation of tableRelationCache after alter table add partition") {
val table = "invalidate_catalog_cache_table"
- Seq(false, true).foreach { autoUpdate =>
- withSQLConf(SQLConf.AUTO_SIZE_UPDATE_ENABLED.key -> autoUpdate.toString) {
- withTempDir { dir =>
- withTable(table) {
- val path = dir.getCanonicalPath
- sql(s"""
- |CREATE TABLE $table (col1 int, col2 int)
- |USING PARQUET
- |PARTITIONED BY (col2)
- |LOCATION '${dir.toURI}'""".stripMargin)
- sql(s"ANALYZE TABLE $table COMPUTE STATISTICS")
- spark.table(table)
- assert(getTableFromCatalogCache(table).stats.sizeInBytes == 0)
- spark.catalog.recoverPartitions(table)
- val df = Seq((1, 2), (1, 2)).toDF("col2", "col1")
- df.write.parquet(s"$path/col2=1")
- sql(s"ALTER TABLE $table ADD PARTITION (col2=1) LOCATION '${dir.toURI}'")
- spark.table(table)
- val cachedTable = getTableFromCatalogCache(table)
- val cachedTableSizeInBytes = cachedTable.stats.sizeInBytes
- val defaultSizeInBytes = conf.defaultSizeInBytes
- if (autoUpdate) {
- assert(cachedTableSizeInBytes != defaultSizeInBytes && cachedTableSizeInBytes > 0)
- } else {
- assert(cachedTableSizeInBytes == defaultSizeInBytes)
+ Seq(false, true).foreach { cboEnabled =>
+ Seq(false, true).foreach { autoUpdate =>
+ withSQLConf(
+ SQLConf.CBO_ENABLED.key -> cboEnabled.toString,
+ SQLConf.AUTO_SIZE_UPDATE_ENABLED.key -> autoUpdate.toString) {
+ withTempDir { dir =>
+ withTable(table) {
+ val path = dir.getCanonicalPath
+ sql(
+ s"""
+ |CREATE TABLE $table (col1 int, col2 int)
+ |USING PARQUET
+ |PARTITIONED BY (col2)
+ |LOCATION '${dir.toURI}'""".stripMargin)
+ sql(s"ANALYZE TABLE $table COMPUTE STATISTICS")
+ spark.table(table)
+ if (!cboEnabled) {
+ assert(getTableFromCatalogCache(table).stats.sizeInBytes === 0)
+ } else {
+ assert(getTableFromCatalogCache(table).stats.sizeInBytes === 1)
Review Comment:
The size is 1, not 0:
https://github.com/apache/spark/blob/ddc61e62b9af5deff1b93e22f466f2a13f281155/sql/catalyst/src/main/scala/org/apache/spark/sql/catalyst/plans/logical/statsEstimation/EstimationUtils.scala#L116-L118
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscribe@spark.apache.org
For additional commands, e-mail: reviews-help@spark.apache.org