/// <exception cref="System.IO.IOException"/>
public OnDiskMapOutput(TaskAttemptID mapId, TaskAttemptID reduceId, MergeManagerImpl
<K, V> merger, long size, JobConf conf, MapOutputFile mapOutputFile, int fetcher
, bool primaryMapOutput)
: this(mapId, reduceId, merger, size, conf, mapOutputFile, fetcher, primaryMapOutput
, FileSystem.GetLocal(conf).GetRaw(), mapOutputFile.GetInputFileForWrite(mapId.GetTaskID
(), size))
{
}
public virtual void TestSucceedAndFailedCopyMap <K, V>()
{
JobConf job = new JobConf();
job.SetNumMapTasks(2);
//mock creation
TaskUmbilicalProtocol mockUmbilical = Org.Mockito.Mockito.Mock <TaskUmbilicalProtocol
>();
Reporter mockReporter = Org.Mockito.Mockito.Mock <Reporter>();
FileSystem mockFileSystem = Org.Mockito.Mockito.Mock <FileSystem>();
Type combinerClass = job.GetCombinerClass();
Task.CombineOutputCollector <K, V> mockCombineOutputCollector = (Task.CombineOutputCollector
<K, V>)Org.Mockito.Mockito.Mock <Task.CombineOutputCollector>();
// needed for mock with generic
TaskAttemptID mockTaskAttemptID = Org.Mockito.Mockito.Mock <TaskAttemptID>();
LocalDirAllocator mockLocalDirAllocator = Org.Mockito.Mockito.Mock <LocalDirAllocator
>();
CompressionCodec mockCompressionCodec = Org.Mockito.Mockito.Mock <CompressionCodec
>();
Counters.Counter mockCounter = Org.Mockito.Mockito.Mock <Counters.Counter>();
TaskStatus mockTaskStatus = Org.Mockito.Mockito.Mock <TaskStatus>();
Progress mockProgress = Org.Mockito.Mockito.Mock <Progress>();
MapOutputFile mockMapOutputFile = Org.Mockito.Mockito.Mock <MapOutputFile>();
Org.Apache.Hadoop.Mapred.Task mockTask = Org.Mockito.Mockito.Mock <Org.Apache.Hadoop.Mapred.Task
>();
MapOutput <K, V> output = Org.Mockito.Mockito.Mock <MapOutput>();
ShuffleConsumerPlugin.Context <K, V> context = new ShuffleConsumerPlugin.Context <K
, V>(mockTaskAttemptID, job, mockFileSystem, mockUmbilical, mockLocalDirAllocator
, mockReporter, mockCompressionCodec, combinerClass, mockCombineOutputCollector,
mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockTaskStatus
, mockProgress, mockProgress, mockTask, mockMapOutputFile, null);
TaskStatus status = new _TaskStatus_251();
Progress progress = new Progress();
ShuffleSchedulerImpl <K, V> scheduler = new ShuffleSchedulerImpl <K, V>(job, status
, null, null, progress, context.GetShuffledMapsCounter(), context.GetReduceShuffleBytes
(), context.GetFailedShuffleCounter());
MapHost host1 = new MapHost("host1", null);
TaskAttemptID failedAttemptID = new TaskAttemptID(new TaskID(new JobID("test", 0)
, TaskType.Map, 0), 0);
TaskAttemptID succeedAttemptID = new TaskAttemptID(new TaskID(new JobID("test", 0
), TaskType.Map, 1), 1);
// handle output fetch failure for failedAttemptID, part I
scheduler.HostFailed(host1.GetHostName());
// handle output fetch succeed for succeedAttemptID
long bytes = (long)500 * 1024 * 1024;
scheduler.CopySucceeded(succeedAttemptID, host1, bytes, 0, 500000, output);
// handle output fetch failure for failedAttemptID, part II
// for MAPREDUCE-6361: verify no NPE exception get thrown out
scheduler.CopyFailed(failedAttemptID, host1, true, false);
}
public virtual void TestConsumerApi()
{
JobConf jobConf = new JobConf();
ShuffleConsumerPlugin <K, V> shuffleConsumerPlugin = new TestShufflePlugin.TestShuffleConsumerPlugin
<K, V>();
//mock creation
ReduceTask mockReduceTask = Org.Mockito.Mockito.Mock <ReduceTask>();
TaskUmbilicalProtocol mockUmbilical = Org.Mockito.Mockito.Mock <TaskUmbilicalProtocol
>();
Reporter mockReporter = Org.Mockito.Mockito.Mock <Reporter>();
FileSystem mockFileSystem = Org.Mockito.Mockito.Mock <FileSystem>();
Type combinerClass = jobConf.GetCombinerClass();
Task.CombineOutputCollector <K, V> mockCombineOutputCollector = (Task.CombineOutputCollector
<K, V>)Org.Mockito.Mockito.Mock <Task.CombineOutputCollector>();
// needed for mock with generic
TaskAttemptID mockTaskAttemptID = Org.Mockito.Mockito.Mock <TaskAttemptID>();
LocalDirAllocator mockLocalDirAllocator = Org.Mockito.Mockito.Mock <LocalDirAllocator
>();
CompressionCodec mockCompressionCodec = Org.Mockito.Mockito.Mock <CompressionCodec
>();
Counters.Counter mockCounter = Org.Mockito.Mockito.Mock <Counters.Counter>();
TaskStatus mockTaskStatus = Org.Mockito.Mockito.Mock <TaskStatus>();
Progress mockProgress = Org.Mockito.Mockito.Mock <Progress>();
MapOutputFile mockMapOutputFile = Org.Mockito.Mockito.Mock <MapOutputFile>();
Org.Apache.Hadoop.Mapred.Task mockTask = Org.Mockito.Mockito.Mock <Org.Apache.Hadoop.Mapred.Task
>();
try
{
string[] dirs = jobConf.GetLocalDirs();
// verify that these APIs are available through super class handler
ShuffleConsumerPlugin.Context <K, V> context = new ShuffleConsumerPlugin.Context <K
, V>(mockTaskAttemptID, jobConf, mockFileSystem, mockUmbilical, mockLocalDirAllocator
, mockReporter, mockCompressionCodec, combinerClass, mockCombineOutputCollector,
mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockTaskStatus
, mockProgress, mockProgress, mockTask, mockMapOutputFile, null);
shuffleConsumerPlugin.Init(context);
shuffleConsumerPlugin.Run();
shuffleConsumerPlugin.Close();
}
catch (Exception e)
{
NUnit.Framework.Assert.IsTrue("Threw exception:" + e, false);
}
// verify that these APIs are available for 3rd party plugins
mockReduceTask.GetTaskID();
mockReduceTask.GetJobID();
mockReduceTask.GetNumMaps();
mockReduceTask.GetPartition();
mockReporter.Progress();
}
internal OnDiskMapOutput(TaskAttemptID mapId, TaskAttemptID reduceId, MergeManagerImpl
<K, V> merger, long size, JobConf conf, MapOutputFile mapOutputFile, int fetcher
, bool primaryMapOutput, FileSystem fs, Path outputPath)
: base(mapId, size, primaryMapOutput)
{
this.fs = fs;
this.merger = merger;
this.outputPath = outputPath;
tmpOutputPath = GetTempPath(outputPath, fetcher);
disk = CryptoUtils.WrapIfNecessary(conf, fs.Create(tmpOutputPath));
this.conf = conf;
}
/// <exception cref="System.Exception"/>
public virtual void TestInterruptOnDisk()
{
int Fetcher = 7;
Path p = new Path("file:///tmp/foo");
Path pTmp = OnDiskMapOutput.GetTempPath(p, Fetcher);
FileSystem mFs = Org.Mockito.Mockito.Mock <FileSystem>(ReturnsDeepStubs);
MapOutputFile mof = Org.Mockito.Mockito.Mock <MapOutputFile>();
Org.Mockito.Mockito.When(mof.GetInputFileForWrite(Matchers.Any <TaskID>(), Matchers.AnyLong
())).ThenReturn(p);
OnDiskMapOutput <Text, Text> odmo = Org.Mockito.Mockito.Spy(new OnDiskMapOutput <Text
, Text>(map1ID, id, mm, 100L, job, mof, Fetcher, true, mFs, p));
Org.Mockito.Mockito.When(mm.Reserve(Matchers.Any <TaskAttemptID>(), Matchers.AnyLong
(), Matchers.AnyInt())).ThenReturn(odmo);
Org.Mockito.Mockito.DoNothing().When(mm).WaitForResource();
Org.Mockito.Mockito.When(ss.GetHost()).ThenReturn(host);
string replyHash = SecureShuffleUtils.GenerateHash(Sharpen.Runtime.GetBytesForString
(encHash), key);
Org.Mockito.Mockito.When(connection.GetResponseCode()).ThenReturn(200);
Org.Mockito.Mockito.When(connection.GetHeaderField(SecureShuffleUtils.HttpHeaderReplyUrlHash
)).ThenReturn(replyHash);
ShuffleHeader header = new ShuffleHeader(map1ID.ToString(), 10, 10, 1);
ByteArrayOutputStream bout = new ByteArrayOutputStream();
header.Write(new DataOutputStream(bout));
TestFetcher.StuckInputStream @in = new TestFetcher.StuckInputStream(new ByteArrayInputStream
(bout.ToByteArray()));
Org.Mockito.Mockito.When(connection.GetInputStream()).ThenReturn(@in);
Org.Mockito.Mockito.When(connection.GetHeaderField(ShuffleHeader.HttpHeaderName))
.ThenReturn(ShuffleHeader.DefaultHttpHeaderName);
Org.Mockito.Mockito.When(connection.GetHeaderField(ShuffleHeader.HttpHeaderVersion
)).ThenReturn(ShuffleHeader.DefaultHttpHeaderVersion);
Org.Mockito.Mockito.DoAnswer(new _Answer_610(@in)).When(connection).Disconnect();
Fetcher <Text, Text> underTest = new TestFetcher.FakeFetcher <Text, Text>(job, id,
ss, mm, r, metrics, except, key, connection, Fetcher);
underTest.Start();
// wait for read in inputstream
@in.WaitForFetcher();
underTest.ShutDown();
underTest.Join();
// rely on test timeout to kill if stuck
NUnit.Framework.Assert.IsTrue(@in.WasClosedProperly());
Org.Mockito.Mockito.Verify(mFs).Create(Matchers.Eq(pTmp));
Org.Mockito.Mockito.Verify(mFs).Delete(Matchers.Eq(pTmp), Matchers.Eq(false));
Org.Mockito.Mockito.Verify(odmo).Abort();
}
public MergeManagerImpl(TaskAttemptID reduceId, JobConf jobConf, FileSystem localFS
, LocalDirAllocator localDirAllocator, Reporter reporter, CompressionCodec codec
, Type combinerClass, Task.CombineOutputCollector <K, V> combineCollector, Counters.Counter
spilledRecordsCounter, Counters.Counter reduceCombineInputCounter, Counters.Counter
mergedMapOutputsCounter, ExceptionReporter exceptionReporter, Progress mergePhase
, MapOutputFile mapOutputFile)
{
/* Maximum percentage of the in-memory limit that a single shuffle can
* consume*/
this.reduceId = reduceId;
this.jobConf = jobConf;
this.localDirAllocator = localDirAllocator;
this.exceptionReporter = exceptionReporter;
this.reporter = reporter;
this.codec = codec;
this.combinerClass = combinerClass;
this.combineCollector = combineCollector;
this.reduceCombineInputCounter = reduceCombineInputCounter;
this.spilledRecordsCounter = spilledRecordsCounter;
this.mergedMapOutputsCounter = mergedMapOutputsCounter;
this.mapOutputFile = mapOutputFile;
this.mapOutputFile.SetConf(jobConf);
this.localFS = localFS;
this.rfs = ((LocalFileSystem)localFS).GetRaw();
float maxInMemCopyUse = jobConf.GetFloat(MRJobConfig.ShuffleInputBufferPercent, MRJobConfig
.DefaultShuffleInputBufferPercent);
if (maxInMemCopyUse > 1.0 || maxInMemCopyUse < 0.0)
{
throw new ArgumentException("Invalid value for " + MRJobConfig.ShuffleInputBufferPercent
+ ": " + maxInMemCopyUse);
}
// Allow unit tests to fix Runtime memory
this.memoryLimit = (long)(jobConf.GetLong(MRJobConfig.ReduceMemoryTotalBytes, Runtime
.GetRuntime().MaxMemory()) * maxInMemCopyUse);
this.ioSortFactor = jobConf.GetInt(MRJobConfig.IoSortFactor, 100);
float singleShuffleMemoryLimitPercent = jobConf.GetFloat(MRJobConfig.ShuffleMemoryLimitPercent
, DefaultShuffleMemoryLimitPercent);
if (singleShuffleMemoryLimitPercent <= 0.0f || singleShuffleMemoryLimitPercent >
1.0f)
{
throw new ArgumentException("Invalid value for " + MRJobConfig.ShuffleMemoryLimitPercent
+ ": " + singleShuffleMemoryLimitPercent);
}
usedMemory = 0L;
commitMemory = 0L;
this.maxSingleShuffleLimit = (long)(memoryLimit * singleShuffleMemoryLimitPercent
);
this.memToMemMergeOutputsThreshold = jobConf.GetInt(MRJobConfig.ReduceMemtomemThreshold
, ioSortFactor);
this.mergeThreshold = (long)(this.memoryLimit * jobConf.GetFloat(MRJobConfig.ShuffleMergePercent
, 0.90f));
Log.Info("MergerManager: memoryLimit=" + memoryLimit + ", " + "maxSingleShuffleLimit="
+ maxSingleShuffleLimit + ", " + "mergeThreshold=" + mergeThreshold + ", " + "ioSortFactor="
+ ioSortFactor + ", " + "memToMemMergeOutputsThreshold=" + memToMemMergeOutputsThreshold
);
if (this.maxSingleShuffleLimit >= this.mergeThreshold)
{
throw new RuntimeException("Invalid configuration: " + "maxSingleShuffleLimit should be less than mergeThreshold "
+ "maxSingleShuffleLimit: " + this.maxSingleShuffleLimit + "mergeThreshold: " +
this.mergeThreshold);
}
bool allowMemToMemMerge = jobConf.GetBoolean(MRJobConfig.ReduceMemtomemEnabled, false
);
if (allowMemToMemMerge)
{
this.memToMemMerger = new MergeManagerImpl.IntermediateMemoryToMemoryMerger(this,
this, memToMemMergeOutputsThreshold);
this.memToMemMerger.Start();
}
else
{
this.memToMemMerger = null;
}
this.inMemoryMerger = CreateInMemoryMerger();
this.inMemoryMerger.Start();
this.onDiskMerger = new MergeManagerImpl.OnDiskMerger(this, this);
this.onDiskMerger.Start();
this.mergePhase = mergePhase;
}
public virtual void TestCorruptedIFile()
{
int fetcher = 7;
Path onDiskMapOutputPath = new Path(name.GetMethodName() + "/foo");
Path shuffledToDisk = OnDiskMapOutput.GetTempPath(onDiskMapOutputPath, fetcher);
fs = FileSystem.GetLocal(job).GetRaw();
MapOutputFile mof = Org.Mockito.Mockito.Mock <MapOutputFile>();
OnDiskMapOutput <Text, Text> odmo = new OnDiskMapOutput <Text, Text>(map1ID, id, mm
, 100L, job, mof, fetcher, true, fs, onDiskMapOutputPath);
string mapData = "MAPDATA12345678901234567890";
ShuffleHeader header = new ShuffleHeader(map1ID.ToString(), 14, 10, 1);
ByteArrayOutputStream bout = new ByteArrayOutputStream();
DataOutputStream dos = new DataOutputStream(bout);
IFileOutputStream ios = new IFileOutputStream(dos);
header.Write(dos);
int headerSize = dos.Size();
try
{
ios.Write(Sharpen.Runtime.GetBytesForString(mapData));
}
finally
{
ios.Close();
}
int dataSize = bout.Size() - headerSize;
// Ensure that the OnDiskMapOutput shuffler can successfully read the data.
MapHost host = new MapHost("TestHost", "http://test/url");
ByteArrayInputStream bin = new ByteArrayInputStream(bout.ToByteArray());
try
{
// Read past the shuffle header.
bin.Read(new byte[headerSize], 0, headerSize);
odmo.Shuffle(host, bin, dataSize, dataSize, metrics, Reporter.Null);
}
finally
{
bin.Close();
}
// Now corrupt the IFile data.
byte[] corrupted = bout.ToByteArray();
corrupted[headerSize + (dataSize / 2)] = unchecked ((int)(0x0));
try
{
bin = new ByteArrayInputStream(corrupted);
// Read past the shuffle header.
bin.Read(new byte[headerSize], 0, headerSize);
odmo.Shuffle(host, bin, dataSize, dataSize, metrics, Reporter.Null);
NUnit.Framework.Assert.Fail("OnDiskMapOutput.shuffle didn't detect the corrupted map partition file"
);
}
catch (ChecksumException e)
{
Log.Info("The expected checksum exception was thrown.", e);
}
finally
{
bin.Close();
}
// Ensure that the shuffled file can be read.
IFileInputStream iFin = new IFileInputStream(fs.Open(shuffledToDisk), dataSize, job
);
try
{
iFin.Read(new byte[dataSize], 0, dataSize);
}
finally
{
iFin.Close();
}
}
public virtual void TestAggregatedTransferRate <K, V>()
{
JobConf job = new JobConf();
job.SetNumMapTasks(10);
//mock creation
TaskUmbilicalProtocol mockUmbilical = Org.Mockito.Mockito.Mock <TaskUmbilicalProtocol
>();
Reporter mockReporter = Org.Mockito.Mockito.Mock <Reporter>();
FileSystem mockFileSystem = Org.Mockito.Mockito.Mock <FileSystem>();
Type combinerClass = job.GetCombinerClass();
Task.CombineOutputCollector <K, V> mockCombineOutputCollector = (Task.CombineOutputCollector
<K, V>)Org.Mockito.Mockito.Mock <Task.CombineOutputCollector>();
// needed for mock with generic
TaskAttemptID mockTaskAttemptID = Org.Mockito.Mockito.Mock <TaskAttemptID>();
LocalDirAllocator mockLocalDirAllocator = Org.Mockito.Mockito.Mock <LocalDirAllocator
>();
CompressionCodec mockCompressionCodec = Org.Mockito.Mockito.Mock <CompressionCodec
>();
Counters.Counter mockCounter = Org.Mockito.Mockito.Mock <Counters.Counter>();
TaskStatus mockTaskStatus = Org.Mockito.Mockito.Mock <TaskStatus>();
Progress mockProgress = Org.Mockito.Mockito.Mock <Progress>();
MapOutputFile mockMapOutputFile = Org.Mockito.Mockito.Mock <MapOutputFile>();
Org.Apache.Hadoop.Mapred.Task mockTask = Org.Mockito.Mockito.Mock <Org.Apache.Hadoop.Mapred.Task
>();
MapOutput <K, V> output = Org.Mockito.Mockito.Mock <MapOutput>();
ShuffleConsumerPlugin.Context <K, V> context = new ShuffleConsumerPlugin.Context <K
, V>(mockTaskAttemptID, job, mockFileSystem, mockUmbilical, mockLocalDirAllocator
, mockReporter, mockCompressionCodec, combinerClass, mockCombineOutputCollector,
mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockCounter, mockTaskStatus
, mockProgress, mockProgress, mockTask, mockMapOutputFile, null);
TaskStatus status = new _TaskStatus_115();
Progress progress = new Progress();
ShuffleSchedulerImpl <K, V> scheduler = new ShuffleSchedulerImpl <K, V>(job, status
, null, null, progress, context.GetShuffledMapsCounter(), context.GetReduceShuffleBytes
(), context.GetFailedShuffleCounter());
TaskAttemptID attemptID0 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 0), 0);
//adding the 1st interval, 40MB from 60s to 100s
long bytes = (long)40 * 1024 * 1024;
scheduler.CopySucceeded(attemptID0, new MapHost(null, null), bytes, 60000, 100000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(1, 1, 1), progress.ToString());
TaskAttemptID attemptID1 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 1), 1);
//adding the 2nd interval before the 1st interval, 50MB from 0s to 50s
bytes = (long)50 * 1024 * 1024;
scheduler.CopySucceeded(attemptID1, new MapHost(null, null), bytes, 0, 50000, output
);
NUnit.Framework.Assert.AreEqual(CopyMessage(2, 1, 1), progress.ToString());
TaskAttemptID attemptID2 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 2), 2);
//adding the 3rd interval overlapping with the 1st and the 2nd interval
//110MB from 25s to 80s
bytes = (long)110 * 1024 * 1024;
scheduler.CopySucceeded(attemptID2, new MapHost(null, null), bytes, 25000, 80000,
output);
NUnit.Framework.Assert.AreEqual(CopyMessage(3, 2, 2), progress.ToString());
TaskAttemptID attemptID3 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 3), 3);
//adding the 4th interval just after the 2nd interval, 100MB from 100s to 300s
bytes = (long)100 * 1024 * 1024;
scheduler.CopySucceeded(attemptID3, new MapHost(null, null), bytes, 100000, 300000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(4, 0.5, 1), progress.ToString());
TaskAttemptID attemptID4 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 4), 4);
//adding the 5th interval between after 4th, 50MB from 350s to 400s
bytes = (long)50 * 1024 * 1024;
scheduler.CopySucceeded(attemptID4, new MapHost(null, null), bytes, 350000, 400000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(5, 1, 1), progress.ToString());
TaskAttemptID attemptID5 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 5), 5);
//adding the 6th interval between after 5th, 50MB from 450s to 500s
bytes = (long)50 * 1024 * 1024;
scheduler.CopySucceeded(attemptID5, new MapHost(null, null), bytes, 450000, 500000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(6, 1, 1), progress.ToString());
TaskAttemptID attemptID6 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 6), 6);
//adding the 7th interval between after 5th and 6th interval, 20MB from 320s to 340s
bytes = (long)20 * 1024 * 1024;
scheduler.CopySucceeded(attemptID6, new MapHost(null, null), bytes, 320000, 340000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(7, 1, 1), progress.ToString());
TaskAttemptID attemptID7 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 7), 7);
//adding the 8th interval overlapping with 4th, 5th, and 7th 30MB from 290s to 350s
bytes = (long)30 * 1024 * 1024;
scheduler.CopySucceeded(attemptID7, new MapHost(null, null), bytes, 290000, 350000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(8, 0.5, 1), progress.ToString());
TaskAttemptID attemptID8 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 8), 8);
//adding the 9th interval overlapping with 5th and 6th, 50MB from 400s to 450s
bytes = (long)50 * 1024 * 1024;
scheduler.CopySucceeded(attemptID8, new MapHost(null, null), bytes, 400000, 450000
, output);
NUnit.Framework.Assert.AreEqual(CopyMessage(9, 1, 1), progress.ToString());
TaskAttemptID attemptID9 = new TaskAttemptID(new TaskID(new JobID("test", 0), TaskType
.Map, 9), 9);
//adding the 10th interval overlapping with all intervals, 500MB from 0s to 500s
bytes = (long)500 * 1024 * 1024;
scheduler.CopySucceeded(attemptID9, new MapHost(null, null), bytes, 0, 500000, output
);
NUnit.Framework.Assert.AreEqual(CopyMessage(10, 1, 2), progress.ToString());
}
