Skip to content

[SPARK-6910] [SQL] Support for pushing predicates down to metastore for partition pruning #6921

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
piaozhexiu wants to merge 6 commits into from
Closed

[SPARK-6910] [SQL] Support for pushing predicates down to metastore for partition pruning #6921

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
ca46042
Implement predicate pushdown for hive metastore catalog
Jun 27, 2015
cf7c6e5
Check data types of expressions and push down string and integral types
Jun 27, 2015
2ff6277
Fix stylecheck
Jun 28, 2015
ead139a
Disable predicate pushdown when hive metastore version is older than
Jun 29, 2015
095672f
Push down predicates when convertMetastoreParquet=true
Jun 29, 2015
05f8e16
Remove an empty line
Jun 29, 2015
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
Prev Previous commit
Next Next commit
Push down predicates when convertMetastoreParquet=true
  • Loading branch information
Cheolsoo Park committed Jun 29, 2015
commit 095672f00ca23c5d071650f3d00657dfe20f38be
2 changes: 2 additions & 0 deletions sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveMetastoreCatalog.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -300,6 +300,8 @@ private[hive] class HiveMetastoreCatalog(val client: ClientInterface, hive: Hive
val result = if (metastoreRelation.hiveQlTable.isPartitioned) {
val partitionSchema = StructType.fromAttributes(metastoreRelation.partitionKeys)
val partitionColumnDataTypes = partitionSchema.map(_.dataType)
// This is for non-partitioned tables, so partition pruning predicates
// cannot be pushed into Hive metastore.
val partitions = metastoreRelation.getHiveQlPartitions(None).map { p =>
val location = p.getLocation
val values = InternalRow.fromSeq(p.getValues.zip(partitionColumnDataTypes).map {
Expand Down
6 changes: 3 additions & 3 deletions sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveShim.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,9 +109,9 @@ private[hive] object HiveShim {
partitionKeys: List[FieldSchema],
hiveMetastoreVersion: String): Option[String] = {

// Binary comparison operators such as >, <, >=, and <= started being supported by
// getPartitionsByFilter() in 0.13. So if Hive matastore version is older than 0.13,
// no predicate is pushed down. See HIVE-4888.
// Binary comparison operators have been supported in getPartitionsByFilter() since 0.13.
// So if Hive matastore version is older than 0.13, predicates cannot be pushed down.
// See HIVE-4888.
val versionPattern = "([\\d]+\\.[\\d]+).*".r
hiveMetastoreVersion match {
case versionPattern(version) => if (version.toDouble < 0.13) return None
Expand Down
9 changes: 8 additions & 1 deletion sql/hive/src/main/scala/org/apache/spark/sql/hive/HiveStrategies.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -106,6 +106,13 @@ private[hive] trait HiveStrategies {

try {
if (relation.hiveQlTable.isPartitioned) {

val metastoreFilter =
HiveShim.toMetastoreFilter(
pruningPredicates,
relation.hiveQlTable.getPartitionKeys,
hiveContext.hiveMetastoreVersion)

val rawPredicate = pruningPredicates.reduceOption(And).getOrElse(Literal(true))
// Translate the predicate so that it automatically casts the input values to the
// correct data types during evaluation.
Expand All @@ -124,7 +131,7 @@ private[hive] trait HiveStrategies {
InterpretedPredicate.create(castedPredicate)
}

val partitions = relation.getHiveQlPartitions(None).filter { part =>
val partitions = relation.getHiveQlPartitions(metastoreFilter).filter { part =>
val partitionValues = part.getValues
var i = 0
while (i < partitionValues.size()) {
Expand Down