org.apache.spark.sql.execution.QueryExecution.scala Maven / Gradle / Ivy
The newest version!
/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.spark.sql.execution
import java.io.{BufferedWriter, OutputStreamWriter}
import java.util.UUID
import java.util.concurrent.atomic.AtomicLong
import org.apache.hadoop.fs.Path
import org.apache.spark.SparkException
import org.apache.spark.internal.Logging
import org.apache.spark.rdd.RDD
import org.apache.spark.sql.{AnalysisException, Row, SparkSession}
import org.apache.spark.sql.catalyst.{InternalRow, QueryPlanningTracker}
import org.apache.spark.sql.catalyst.analysis.UnsupportedOperationChecker
import org.apache.spark.sql.catalyst.expressions.codegen.ByteCodeStats
import org.apache.spark.sql.catalyst.plans.QueryPlan
import org.apache.spark.sql.catalyst.plans.logical.{AppendData, Command, CommandResult, CreateTableAsSelect, LogicalPlan, OverwriteByExpression, OverwritePartitionsDynamic, ReplaceTableAsSelect, ReturnAnswer}
import org.apache.spark.sql.catalyst.rules.{PlanChangeLogger, Rule}
import org.apache.spark.sql.catalyst.util.StringUtils.PlanStringConcat
import org.apache.spark.sql.catalyst.util.truncatedString
import org.apache.spark.sql.execution.adaptive.{AdaptiveExecutionContext, InsertAdaptiveSparkPlan}
import org.apache.spark.sql.execution.bucketing.{CoalesceBucketsInJoin, DisableUnnecessaryBucketedScan}
import org.apache.spark.sql.execution.dynamicpruning.PlanDynamicPruningFilters
import org.apache.spark.sql.execution.exchange.EnsureRequirements
import org.apache.spark.sql.execution.reuse.ReuseExchangeAndSubquery
import org.apache.spark.sql.execution.streaming.{IncrementalExecution, OffsetSeqMetadata, WatermarkPropagator}
import org.apache.spark.sql.internal.SQLConf
import org.apache.spark.sql.streaming.OutputMode
import org.apache.spark.util.Utils
/**
* The primary workflow for executing relational queries using Spark. Designed to allow easy
* access to the intermediate phases of query execution for developers.
*
* While this is not a public class, we should avoid changing the function names for the sake of
* changing them, because a lot of developers use the feature for debugging.
*/
class QueryExecution(
val sparkSession: SparkSession,
val logical: LogicalPlan,
val tracker: QueryPlanningTracker = new QueryPlanningTracker,
val mode: CommandExecutionMode.Value = CommandExecutionMode.ALL) extends Logging {
val id: Long = QueryExecution.nextExecutionId
// TODO: Move the planner an optimizer into here from SessionState.
protected def planner = sparkSession.sessionState.planner
def assertAnalyzed(): Unit = analyzed
def assertSupported(): Unit = {
if (sparkSession.sessionState.conf.isUnsupportedOperationCheckEnabled) {
UnsupportedOperationChecker.checkForBatch(analyzed)
}
}
lazy val analyzed: LogicalPlan = {
val plan = executePhase(QueryPlanningTracker.ANALYSIS) {
// We can't clone `logical` here, which will reset the `_analyzed` flag.
sparkSession.sessionState.analyzer.executeAndCheck(logical, tracker)
}
tracker.setAnalyzed(plan)
plan
}
lazy val commandExecuted: LogicalPlan = mode match {
case CommandExecutionMode.NON_ROOT => analyzed.mapChildren(eagerlyExecuteCommands)
case CommandExecutionMode.ALL => eagerlyExecuteCommands(analyzed)
case CommandExecutionMode.SKIP => analyzed
}
private def commandExecutionName(command: Command): String = command match {
case _: CreateTableAsSelect => "create"
case _: ReplaceTableAsSelect => "replace"
case _: AppendData => "append"
case _: OverwriteByExpression => "overwrite"
case _: OverwritePartitionsDynamic => "overwritePartitions"
case _ => "command"
}
private def eagerlyExecuteCommands(p: LogicalPlan) = p transformDown {
case c: Command =>
// Since Command execution will eagerly take place here,
// and in most cases be the bulk of time and effort,
// with the rest of processing of the root plan being just outputting command results,
// for eagerly executed commands we mark this place as beginning of execution.
tracker.setReadyForExecution()
val qe = sparkSession.sessionState.executePlan(c, CommandExecutionMode.NON_ROOT)
val result = SQLExecution.withNewExecutionId(qe, Some(commandExecutionName(c))) {
qe.executedPlan.executeCollect()
}
CommandResult(
qe.analyzed.output,
qe.commandExecuted,
qe.executedPlan,
result)
case other => other
}
// The plan that has been normalized by custom rules, so that it's more likely to hit cache.
lazy val normalized: LogicalPlan = {
val normalizationRules = sparkSession.sessionState.planNormalizationRules
if (normalizationRules.isEmpty) {
commandExecuted
} else {
val planChangeLogger = new PlanChangeLogger[LogicalPlan]()
val normalized = normalizationRules.foldLeft(commandExecuted) { (p, rule) =>
val result = rule.apply(p)
planChangeLogger.logRule(rule.ruleName, p, result)
result
}
planChangeLogger.logBatch("Plan Normalization", commandExecuted, normalized)
normalized
}
}
lazy val withCachedData: LogicalPlan = sparkSession.withActive {
assertAnalyzed()
assertSupported()
// clone the plan to avoid sharing the plan instance between different stages like analyzing,
// optimizing and planning.
sparkSession.sharedState.cacheManager.useCachedData(normalized.clone())
}
def assertCommandExecuted(): Unit = commandExecuted
lazy val optimizedPlan: LogicalPlan = {
// We need to materialize the commandExecuted here because optimizedPlan is also tracked under
// the optimizing phase
assertCommandExecuted()
executePhase(QueryPlanningTracker.OPTIMIZATION) {
// clone the plan to avoid sharing the plan instance between different stages like analyzing,
// optimizing and planning.
val plan =
sparkSession.sessionState.optimizer.executeAndTrack(withCachedData.clone(), tracker)
// We do not want optimized plans to be re-analyzed as literals that have been constant
// folded and such can cause issues during analysis. While `clone` should maintain the
// `analyzed` state of the LogicalPlan, we set the plan as analyzed here as well out of
// paranoia.
plan.setAnalyzed()
plan
}
}
def assertOptimized(): Unit = optimizedPlan
lazy val sparkPlan: SparkPlan = {
// We need to materialize the optimizedPlan here because sparkPlan is also tracked under
// the planning phase
assertOptimized()
executePhase(QueryPlanningTracker.PLANNING) {
// Clone the logical plan here, in case the planner rules change the states of the logical
// plan.
QueryExecution.createSparkPlan(sparkSession, planner, optimizedPlan.clone())
}
}
def assertSparkPlanPrepared(): Unit = sparkPlan
// executedPlan should not be used to initialize any SparkPlan. It should be
// only used for execution.
lazy val executedPlan: SparkPlan = {
// We need to materialize the optimizedPlan here, before tracking the planning phase, to ensure
// that the optimization time is not counted as part of the planning phase.
assertOptimized()
val plan = executePhase(QueryPlanningTracker.PLANNING) {
// clone the plan to avoid sharing the plan instance between different stages like analyzing,
// optimizing and planning.
QueryExecution.prepareForExecution(preparations, sparkPlan.clone())
}
// Note: For eagerly executed command it might have already been called in
// `eagerlyExecutedCommand` and is a noop here.
tracker.setReadyForExecution()
plan
}
def assertExecutedPlanPrepared(): Unit = executedPlan
/**
* Internal version of the RDD. Avoids copies and has no schema.
* Note for callers: Spark may apply various optimization including reusing object: this means
* the row is valid only for the iteration it is retrieved. You should avoid storing row and
* accessing after iteration. (Calling `collect()` is one of known bad usage.)
* If you want to store these rows into collection, please apply some converter or copy row
* which produces new object per iteration.
* Given QueryExecution is not a public class, end users are discouraged to use this: please
* use `Dataset.rdd` instead where conversion will be applied.
*/
lazy val toRdd: RDD[InternalRow] = new SQLExecutionRDD(
executedPlan.execute(), sparkSession.sessionState.conf)
/** Get the metrics observed during the execution of the query plan. */
def observedMetrics: Map[String, Row] = CollectMetricsExec.collect(executedPlan)
protected def preparations: Seq[Rule[SparkPlan]] = {
QueryExecution.preparations(sparkSession,
Option(InsertAdaptiveSparkPlan(AdaptiveExecutionContext(sparkSession, this))), false)
}
protected def executePhase[T](phase: String)(block: => T): T = sparkSession.withActive {
QueryExecution.withInternalError(s"The Spark SQL phase $phase failed with an internal error.") {
tracker.measurePhase(phase)(block)
}
}
def simpleString: String = {
val concat = new PlanStringConcat()
simpleString(false, SQLConf.get.maxToStringFields, concat.append)
withRedaction {
concat.toString
}
}
private def simpleString(
formatted: Boolean,
maxFields: Int,
append: String => Unit): Unit = {
append("== Physical Plan ==\n")
if (formatted) {
try {
ExplainUtils.processPlan(executedPlan, append)
} catch {
case e: AnalysisException => append(e.toString)
case e: IllegalArgumentException => append(e.toString)
}
} else {
QueryPlan.append(executedPlan,
append, verbose = false, addSuffix = false, maxFields = maxFields)
}
append("\n")
}
def explainString(mode: ExplainMode): String = {
val concat = new PlanStringConcat()
explainString(mode, SQLConf.get.maxToStringFields, concat.append)
withRedaction {
concat.toString
}
}
private def explainString(mode: ExplainMode, maxFields: Int, append: String => Unit): Unit = {
val queryExecution = if (logical.isStreaming) {
// This is used only by explaining `Dataset/DataFrame` created by `spark.readStream`, so the
// output mode does not matter since there is no `Sink`.
new IncrementalExecution(
sparkSession, logical, OutputMode.Append(), "",
UUID.randomUUID, UUID.randomUUID, 0, None, OffsetSeqMetadata(0, 0),
WatermarkPropagator.noop())
} else {
this
}
mode match {
case SimpleMode =>
queryExecution.simpleString(false, maxFields, append)
case ExtendedMode =>
queryExecution.toString(maxFields, append)
case CodegenMode =>
try {
org.apache.spark.sql.execution.debug.writeCodegen(append, queryExecution.executedPlan)
} catch {
case e: AnalysisException => append(e.toString)
}
case CostMode =>
queryExecution.stringWithStats(maxFields, append)
case FormattedMode =>
queryExecution.simpleString(formatted = true, maxFields = maxFields, append)
}
}
private def writePlans(append: String => Unit, maxFields: Int): Unit = {
val (verbose, addSuffix) = (true, false)
append("== Parsed Logical Plan ==\n")
QueryPlan.append(logical, append, verbose, addSuffix, maxFields)
append("\n== Analyzed Logical Plan ==\n")
try {
if (analyzed.output.nonEmpty) {
append(
truncatedString(
analyzed.output.map(o => s"${o.name}: ${o.dataType.simpleString}"), ", ", maxFields)
)
append("\n")
}
QueryPlan.append(analyzed, append, verbose, addSuffix, maxFields)
append("\n== Optimized Logical Plan ==\n")
QueryPlan.append(optimizedPlan, append, verbose, addSuffix, maxFields)
append("\n== Physical Plan ==\n")
QueryPlan.append(executedPlan, append, verbose, addSuffix, maxFields)
} catch {
case e: AnalysisException => append(e.toString)
}
}
override def toString: String = withRedaction {
val concat = new PlanStringConcat()
toString(SQLConf.get.maxToStringFields, concat.append)
withRedaction {
concat.toString
}
}
private def toString(maxFields: Int, append: String => Unit): Unit = {
writePlans(append, maxFields)
}
def stringWithStats: String = {
val concat = new PlanStringConcat()
stringWithStats(SQLConf.get.maxToStringFields, concat.append)
withRedaction {
concat.toString
}
}
private def stringWithStats(maxFields: Int, append: String => Unit): Unit = {
// trigger to compute stats for logical plans
try {
// This will trigger to compute stats for all the nodes in the plan, including subqueries,
// if the stats doesn't exist in the statsCache and update the statsCache corresponding
// to the node.
optimizedPlan.collectWithSubqueries {
case plan => plan.stats
}
} catch {
case e: AnalysisException => append(e.toString + "\n")
}
// only show optimized logical plan and physical plan
append("== Optimized Logical Plan ==\n")
QueryPlan.append(optimizedPlan, append, verbose = true, addSuffix = true, maxFields)
append("\n== Physical Plan ==\n")
QueryPlan.append(executedPlan, append, verbose = true, addSuffix = false, maxFields)
append("\n")
}
/**
* Redact the sensitive information in the given string.
*/
private def withRedaction(message: String): String = {
Utils.redact(sparkSession.sessionState.conf.stringRedactionPattern, message)
}
/** A special namespace for commands that can be used to debug query execution. */
// scalastyle:off
object debug {
// scalastyle:on
/**
* Prints to stdout all the generated code found in this plan (i.e. the output of each
* WholeStageCodegen subtree).
*/
def codegen(): Unit = {
// scalastyle:off println
println(org.apache.spark.sql.execution.debug.codegenString(executedPlan))
// scalastyle:on println
}
/**
* Get WholeStageCodegenExec subtrees and the codegen in a query plan
*
* @return Sequence of WholeStageCodegen subtrees and corresponding codegen
*/
def codegenToSeq(): Seq[(String, String, ByteCodeStats)] = {
org.apache.spark.sql.execution.debug.codegenStringSeq(executedPlan)
}
/**
* Dumps debug information about query execution into the specified file.
*
* @param path path of the file the debug info is written to.
* @param maxFields maximum number of fields converted to string representation.
* @param explainMode the explain mode to be used to generate the string
* representation of the plan.
*/
def toFile(
path: String,
maxFields: Int = Int.MaxValue,
explainMode: Option[String] = None): Unit = {
val filePath = new Path(path)
val fs = filePath.getFileSystem(sparkSession.sessionState.newHadoopConf())
val writer = new BufferedWriter(new OutputStreamWriter(fs.create(filePath)))
try {
val mode = explainMode.map(ExplainMode.fromString(_)).getOrElse(ExtendedMode)
explainString(mode, maxFields, writer.write)
if (mode != CodegenMode) {
writer.write("\n== Whole Stage Codegen ==\n")
org.apache.spark.sql.execution.debug.writeCodegen(writer.write, executedPlan)
}
log.info(s"Debug information was written at: $filePath")
} finally {
writer.close()
}
}
}
}
/**
* SPARK-35378: Commands should be executed eagerly so that something like `sql("INSERT ...")`
* can trigger the table insertion immediately without a `.collect()`. To avoid end-less recursion
* we should use `NON_ROOT` when recursively executing commands. Note that we can't execute
* a query plan with leaf command nodes, because many commands return `GenericInternalRow`
* and can't be put in a query plan directly, otherwise the query engine may cast
* `GenericInternalRow` to `UnsafeRow` and fail. When running EXPLAIN, or commands inside other
* command, we should use `SKIP` to not eagerly trigger the command execution.
*/
object CommandExecutionMode extends Enumeration {
val SKIP, NON_ROOT, ALL = Value
}
object QueryExecution {
private val _nextExecutionId = new AtomicLong(0)
private def nextExecutionId: Long = _nextExecutionId.getAndIncrement
/**
* Construct a sequence of rules that are used to prepare a planned [[SparkPlan]] for execution.
* These rules will make sure subqueries are planned, make sure the data partitioning and ordering
* are correct, insert whole stage code gen, and try to reduce the work done by reusing exchanges
* and subqueries.
*/
private[execution] def preparations(
sparkSession: SparkSession,
adaptiveExecutionRule: Option[InsertAdaptiveSparkPlan] = None,
subquery: Boolean): Seq[Rule[SparkPlan]] = {
// `AdaptiveSparkPlanExec` is a leaf node. If inserted, all the following rules will be no-op
// as the original plan is hidden behind `AdaptiveSparkPlanExec`.
adaptiveExecutionRule.toSeq ++
Seq(
CoalesceBucketsInJoin,
PlanDynamicPruningFilters(sparkSession),
PlanSubqueries(sparkSession),
RemoveRedundantProjects,
EnsureRequirements(),
// `ReplaceHashWithSortAgg` needs to be added after `EnsureRequirements` to guarantee the
// sort order of each node is checked to be valid.
ReplaceHashWithSortAgg,
// `RemoveRedundantSorts` and `RemoveRedundantWindowGroupLimits` needs to be added after
// `EnsureRequirements` to guarantee the same number of partitions when instantiating
// PartitioningCollection.
RemoveRedundantSorts,
RemoveRedundantWindowGroupLimits,
DisableUnnecessaryBucketedScan,
ApplyColumnarRulesAndInsertTransitions(
sparkSession.sessionState.columnarRules, outputsColumnar = false),
CollapseCodegenStages()) ++
(if (subquery) {
Nil
} else {
Seq(ReuseExchangeAndSubquery)
})
}
/**
* Prepares a planned [[SparkPlan]] for execution by inserting shuffle operations and internal
* row format conversions as needed.
*/
private[execution] def prepareForExecution(
preparations: Seq[Rule[SparkPlan]],
plan: SparkPlan): SparkPlan = {
val planChangeLogger = new PlanChangeLogger[SparkPlan]()
val preparedPlan = preparations.foldLeft(plan) { case (sp, rule) =>
val result = rule.apply(sp)
planChangeLogger.logRule(rule.ruleName, sp, result)
result
}
planChangeLogger.logBatch("Preparations", plan, preparedPlan)
preparedPlan
}
/**
* Transform a [[LogicalPlan]] into a [[SparkPlan]].
*
* Note that the returned physical plan still needs to be prepared for execution.
*/
def createSparkPlan(
sparkSession: SparkSession,
planner: SparkPlanner,
plan: LogicalPlan): SparkPlan = {
// TODO: We use next(), i.e. take the first plan returned by the planner, here for now,
// but we will implement to choose the best plan.
planner.plan(ReturnAnswer(plan)).next()
}
/**
* Prepare the [[SparkPlan]] for execution.
*/
def prepareExecutedPlan(spark: SparkSession, plan: SparkPlan): SparkPlan = {
prepareForExecution(preparations(spark, subquery = true), plan)
}
/**
* Transform the subquery's [[LogicalPlan]] into a [[SparkPlan]] and prepare the resulting
* [[SparkPlan]] for execution.
*/
def prepareExecutedPlan(spark: SparkSession, plan: LogicalPlan): SparkPlan = {
val sparkPlan = createSparkPlan(spark, spark.sessionState.planner, plan.clone())
prepareExecutedPlan(spark, sparkPlan)
}
/**
* Prepare the [[SparkPlan]] for execution using exists adaptive execution context.
* This method is only called by [[PlanAdaptiveDynamicPruningFilters]].
*/
def prepareExecutedPlan(
session: SparkSession,
plan: LogicalPlan,
context: AdaptiveExecutionContext): SparkPlan = {
val sparkPlan = createSparkPlan(session, session.sessionState.planner, plan.clone())
val preparationRules = preparations(session, Option(InsertAdaptiveSparkPlan(context)), true)
prepareForExecution(preparationRules, sparkPlan.clone())
}
/**
* Converts asserts, null pointer exceptions to internal errors.
*/
private[sql] def toInternalError(msg: String, e: Throwable): Throwable = e match {
case e @ (_: java.lang.NullPointerException | _: java.lang.AssertionError) =>
SparkException.internalError(
msg + " You hit a bug in Spark or the Spark plugins you use. Please, report this bug " +
"to the corresponding communities or vendors, and provide the full stack trace.",
e)
case e: Throwable =>
e
}
/**
* Catches asserts, null pointer exceptions, and converts them to internal errors.
*/
private[sql] def withInternalError[T](msg: String)(block: => T): T = {
try {
block
} catch {
case e: Throwable => throw toInternalError(msg, e)
}
}
}