1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
|
/*
* 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.ui
import java.util.concurrent.atomic.AtomicLong
import scala.collection.mutable
import org.apache.commons.lang3.StringEscapeUtils
import org.apache.spark.sql.execution.{SparkPlanInfo, WholeStageCodegen}
import org.apache.spark.sql.execution.metric.SQLMetrics
/**
* A graph used for storing information of an executionPlan of DataFrame.
*
* Each graph is defined with a set of nodes and a set of edges. Each node represents a node in the
* SparkPlan tree, and each edge represents a parent-child relationship between two nodes.
*/
private[ui] case class SparkPlanGraph(
nodes: Seq[SparkPlanGraphNode], edges: Seq[SparkPlanGraphEdge]) {
def makeDotFile(metrics: Map[Long, String]): String = {
val dotFile = new StringBuilder
dotFile.append("digraph G {\n")
nodes.foreach(node => dotFile.append(node.makeDotNode(metrics) + "\n"))
edges.foreach(edge => dotFile.append(edge.makeDotEdge + "\n"))
dotFile.append("}")
dotFile.toString()
}
/**
* All the SparkPlanGraphNodes, including those inside of WholeStageCodegen.
*/
val allNodes: Seq[SparkPlanGraphNode] = {
nodes.flatMap {
case cluster: SparkPlanGraphCluster => cluster.nodes :+ cluster
case node => Seq(node)
}
}
}
private[sql] object SparkPlanGraph {
/**
* Build a SparkPlanGraph from the root of a SparkPlan tree.
*/
def apply(planInfo: SparkPlanInfo): SparkPlanGraph = {
val nodeIdGenerator = new AtomicLong(0)
val nodes = mutable.ArrayBuffer[SparkPlanGraphNode]()
val edges = mutable.ArrayBuffer[SparkPlanGraphEdge]()
val exchanges = mutable.HashMap[SparkPlanInfo, SparkPlanGraphNode]()
buildSparkPlanGraphNode(planInfo, nodeIdGenerator, nodes, edges, null, null, exchanges)
new SparkPlanGraph(nodes, edges)
}
private def buildSparkPlanGraphNode(
planInfo: SparkPlanInfo,
nodeIdGenerator: AtomicLong,
nodes: mutable.ArrayBuffer[SparkPlanGraphNode],
edges: mutable.ArrayBuffer[SparkPlanGraphEdge],
parent: SparkPlanGraphNode,
subgraph: SparkPlanGraphCluster,
exchanges: mutable.HashMap[SparkPlanInfo, SparkPlanGraphNode]): Unit = {
planInfo.nodeName match {
case "WholeStageCodegen" =>
val metrics = planInfo.metrics.map { metric =>
SQLPlanMetric(metric.name, metric.accumulatorId,
SQLMetrics.getMetricParam(metric.metricParam))
}
val cluster = new SparkPlanGraphCluster(
nodeIdGenerator.getAndIncrement(),
planInfo.nodeName,
planInfo.simpleString,
mutable.ArrayBuffer[SparkPlanGraphNode](),
metrics)
nodes += cluster
buildSparkPlanGraphNode(
planInfo.children.head, nodeIdGenerator, nodes, edges, parent, cluster, exchanges)
case "InputAdapter" =>
buildSparkPlanGraphNode(
planInfo.children.head, nodeIdGenerator, nodes, edges, parent, null, exchanges)
case "Subquery" if subgraph != null =>
// Subquery should not be included in WholeStageCodegen
buildSparkPlanGraphNode(planInfo, nodeIdGenerator, nodes, edges, parent, null, exchanges)
case "ReusedExchange" =>
// Point to the re-used exchange
val node = exchanges(planInfo.children.head)
edges += SparkPlanGraphEdge(node.id, parent.id)
case name =>
val metrics = planInfo.metrics.map { metric =>
SQLPlanMetric(metric.name, metric.accumulatorId,
SQLMetrics.getMetricParam(metric.metricParam))
}
val node = new SparkPlanGraphNode(
nodeIdGenerator.getAndIncrement(), planInfo.nodeName,
planInfo.simpleString, planInfo.metadata, metrics)
if (subgraph == null) {
nodes += node
} else {
subgraph.nodes += node
}
if (name.contains("Exchange")) {
exchanges += planInfo -> node
}
if (parent != null) {
edges += SparkPlanGraphEdge(node.id, parent.id)
}
planInfo.children.foreach(
buildSparkPlanGraphNode(_, nodeIdGenerator, nodes, edges, node, subgraph, exchanges))
}
}
}
/**
* Represent a node in the SparkPlan tree, along with its metrics.
*
* @param id generated by "SparkPlanGraph". There is no duplicate id in a graph
* @param name the name of this SparkPlan node
* @param metrics metrics that this SparkPlan node will track
*/
private[ui] class SparkPlanGraphNode(
val id: Long,
val name: String,
val desc: String,
val metadata: Map[String, String],
val metrics: Seq[SQLPlanMetric]) {
def makeDotNode(metricsValue: Map[Long, String]): String = {
val builder = new mutable.StringBuilder(name)
val values = for {
metric <- metrics
value <- metricsValue.get(metric.accumulatorId)
} yield {
metric.name + ": " + value
}
if (values.nonEmpty) {
// If there are metrics, display each entry in a separate line.
// Note: whitespace between two "\n"s is to create an empty line between the name of
// SparkPlan and metrics. If removing it, it won't display the empty line in UI.
builder ++= "\n \n"
builder ++= values.mkString("\n")
}
s""" $id [label="${StringEscapeUtils.escapeJava(builder.toString())}"];"""
}
}
/**
* Represent a tree of SparkPlan for WholeStageCodegen.
*/
private[ui] class SparkPlanGraphCluster(
id: Long,
name: String,
desc: String,
val nodes: mutable.ArrayBuffer[SparkPlanGraphNode],
metrics: Seq[SQLPlanMetric])
extends SparkPlanGraphNode(id, name, desc, Map.empty, metrics) {
override def makeDotNode(metricsValue: Map[Long, String]): String = {
val duration = metrics.filter(_.name.startsWith(WholeStageCodegen.PIPELINE_DURATION_METRIC))
val labelStr = if (duration.nonEmpty) {
require(duration.length == 1)
val id = duration(0).accumulatorId
if (metricsValue.contains(duration(0).accumulatorId)) {
name + "\n\n" + metricsValue.get(id).get
} else {
name
}
} else {
name
}
s"""
| subgraph cluster${id} {
| label="${StringEscapeUtils.escapeJava(labelStr)}";
| ${nodes.map(_.makeDotNode(metricsValue)).mkString(" \n")}
| }
""".stripMargin
}
}
/**
* Represent an edge in the SparkPlan tree. `fromId` is the parent node id, and `toId` is the child
* node id.
*/
private[ui] case class SparkPlanGraphEdge(fromId: Long, toId: Long) {
def makeDotEdge: String = s""" $fromId->$toId;\n"""
}
|
