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
|
package org.apache.hadoop.mapred
import org.apache.hadoop.fs.FileSystem
import org.apache.hadoop.fs.Path
import java.text.SimpleDateFormat
import java.text.NumberFormat
import java.io.IOException
import java.util.Date
import spark.Logging
import spark.SerializableWritable
/**
* Internal helper class that saves an RDD using a Hadoop OutputFormat. This is only public
* because we need to access this class from the `spark` package to use some package-private Hadoop
* functions, but this class should not be used directly by users.
*
* Saves the RDD using a JobConf, which should contain an output key class, an output value class,
* a filename to write to, etc, exactly like in a Hadoop MapReduce job.
*/
class HadoopWriter(@transient jobConf: JobConf) extends Logging with HadoopMapRedUtil with Serializable {
private val now = new Date()
private val conf = new SerializableWritable(jobConf)
private var jobID = 0
private var splitID = 0
private var attemptID = 0
private var jID: SerializableWritable[JobID] = null
private var taID: SerializableWritable[TaskAttemptID] = null
@transient private var writer: RecordWriter[AnyRef,AnyRef] = null
@transient private var format: OutputFormat[AnyRef,AnyRef] = null
@transient private var committer: OutputCommitter = null
@transient private var jobContext: JobContext = null
@transient private var taskContext: TaskAttemptContext = null
def preSetup() {
setIDs(0, 0, 0)
setConfParams()
val jCtxt = getJobContext()
getOutputCommitter().setupJob(jCtxt)
}
def setup(jobid: Int, splitid: Int, attemptid: Int) {
setIDs(jobid, splitid, attemptid)
setConfParams()
}
def open() {
val numfmt = NumberFormat.getInstance()
numfmt.setMinimumIntegerDigits(5)
numfmt.setGroupingUsed(false)
val outputName = "part-" + numfmt.format(splitID)
val path = FileOutputFormat.getOutputPath(conf.value)
val fs: FileSystem = {
if (path != null) {
path.getFileSystem(conf.value)
} else {
FileSystem.get(conf.value)
}
}
getOutputCommitter().setupTask(getTaskContext())
writer = getOutputFormat().getRecordWriter(
fs, conf.value, outputName, Reporter.NULL)
}
def write(key: AnyRef, value: AnyRef) {
if (writer!=null) {
//println (">>> Writing ("+key.toString+": " + key.getClass.toString + ", " + value.toString + ": " + value.getClass.toString + ")")
writer.write(key, value)
} else {
throw new IOException("Writer is null, open() has not been called")
}
}
def close() {
writer.close(Reporter.NULL)
}
def commit() {
val taCtxt = getTaskContext()
val cmtr = getOutputCommitter()
if (cmtr.needsTaskCommit(taCtxt)) {
try {
cmtr.commitTask(taCtxt)
logInfo (taID + ": Committed")
} catch {
case e: IOException => {
logError("Error committing the output of task: " + taID.value, e)
cmtr.abortTask(taCtxt)
throw e
}
}
} else {
logWarning ("No need to commit output of task: " + taID.value)
}
}
def commitJob() {
// always ? Or if cmtr.needsTaskCommit ?
val cmtr = getOutputCommitter()
cmtr.commitJob(getJobContext())
}
def cleanup() {
getOutputCommitter().cleanupJob(getJobContext())
}
// ********* Private Functions *********
private def getOutputFormat(): OutputFormat[AnyRef,AnyRef] = {
if (format == null) {
format = conf.value.getOutputFormat()
.asInstanceOf[OutputFormat[AnyRef,AnyRef]]
}
return format
}
private def getOutputCommitter(): OutputCommitter = {
if (committer == null) {
committer = conf.value.getOutputCommitter
}
return committer
}
private def getJobContext(): JobContext = {
if (jobContext == null) {
jobContext = newJobContext(conf.value, jID.value)
}
return jobContext
}
private def getTaskContext(): TaskAttemptContext = {
if (taskContext == null) {
taskContext = newTaskAttemptContext(conf.value, taID.value)
}
return taskContext
}
private def setIDs(jobid: Int, splitid: Int, attemptid: Int) {
jobID = jobid
splitID = splitid
attemptID = attemptid
jID = new SerializableWritable[JobID](HadoopWriter.createJobID(now, jobid))
taID = new SerializableWritable[TaskAttemptID](
new TaskAttemptID(new TaskID(jID.value, true, splitID), attemptID))
}
private def setConfParams() {
conf.value.set("mapred.job.id", jID.value.toString)
conf.value.set("mapred.tip.id", taID.value.getTaskID.toString)
conf.value.set("mapred.task.id", taID.value.toString)
conf.value.setBoolean("mapred.task.is.map", true)
conf.value.setInt("mapred.task.partition", splitID)
}
}
object HadoopWriter {
def createJobID(time: Date, id: Int): JobID = {
val formatter = new SimpleDateFormat("yyyyMMddHHmm")
val jobtrackerID = formatter.format(new Date())
return new JobID(jobtrackerID, id)
}
def createPathFromString(path: String, conf: JobConf): Path = {
if (path == null) {
throw new IllegalArgumentException("Output path is null")
}
var outputPath = new Path(path)
val fs = outputPath.getFileSystem(conf)
if (outputPath == null || fs == null) {
throw new IllegalArgumentException("Incorrectly formatted output path")
}
outputPath = outputPath.makeQualified(fs)
return outputPath
}
}
|
