/// <exception cref="System.Exception"/>
public virtual void TestCheckOutputSpecsException()
{
Job job = Job.GetInstance();
Path outDir = new Path(Runtime.GetProperty("test.build.data", "/tmp"), "output");
FileSystem fs = outDir.GetFileSystem(new Configuration());
// Create the output dir so it already exists and set it for the job
fs.Mkdirs(outDir);
FileOutputFormat.SetOutputPath(job, outDir);
// We don't need a "full" implementation of FileOutputFormat for this test
FileOutputFormat fof = new _FileOutputFormat_54();
try
{
try
{
// This should throw a FileAlreadyExistsException because the outputDir
// already exists
fof.CheckOutputSpecs(job);
Fail("Should have thrown a FileAlreadyExistsException");
}
catch (FileAlreadyExistsException)
{
}
}
finally
{
// correct behavior
// Cleanup
if (fs.Exists(outDir))
{
fs.Delete(outDir, true);
}
}
}
/// <exception cref="System.Exception"/>
private void TestMapFileOutputCommitterInternal(int version)
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, version);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// write output
MapFileOutputFormat theOutputFormat = new MapFileOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(tContext);
WriteMapFileOutput(theRecordWriter, tContext);
// do commit
committer.CommitTask(tContext);
committer.CommitJob(jContext);
// validate output
ValidateMapFileOutputContent(FileSystem.Get(job.GetConfiguration()), outDir);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.Exception"/>
private void TestConcurrentCommitTaskWithSubDir(int version)
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, version);
conf.SetClass("fs.file.impl", typeof(TestFileOutputCommitter.RLFS), typeof(FileSystem
));
FileSystem.CloseAll();
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
FileOutputCommitter amCommitter = new FileOutputCommitter(outDir, jContext);
amCommitter.SetupJob(jContext);
TaskAttemptContext[] taCtx = new TaskAttemptContextImpl[2];
taCtx[0] = new TaskAttemptContextImpl(conf, taskID);
taCtx[1] = new TaskAttemptContextImpl(conf, taskID1);
TextOutputFormat[] tof = new TextOutputFormat[2];
for (int i = 0; i < tof.Length; i++)
{
tof[i] = new _TextOutputFormat_508(this);
}
ExecutorService executor = Executors.NewFixedThreadPool(2);
try
{
for (int i_1 = 0; i_1 < taCtx.Length; i_1++)
{
int taskIdx = i_1;
executor.Submit(new _Callable_524(this, tof, taskIdx, taCtx));
}
}
finally
{
executor.Shutdown();
while (!executor.AwaitTermination(1, TimeUnit.Seconds))
{
Log.Info("Awaiting thread termination!");
}
}
amCommitter.CommitJob(jContext);
RawLocalFileSystem lfs = new RawLocalFileSystem();
lfs.SetConf(conf);
NUnit.Framework.Assert.IsFalse("Must not end up with sub_dir/sub_dir", lfs.Exists
(new Path(OutSubDir, SubDir)));
// validate output
ValidateContent(OutSubDir);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.Exception"/>
public virtual void TestSetOutputPathException()
{
Job job = Job.GetInstance();
try
{
// Give it an invalid filesystem so it'll throw an exception
FileOutputFormat.SetOutputPath(job, new Path("foo:///bar"));
Fail("Should have thrown a RuntimeException with an IOException inside");
}
catch (RuntimeException re)
{
NUnit.Framework.Assert.IsTrue(re.InnerException is IOException);
}
}
/// <exception cref="System.IO.IOException"/>
public virtual void TestcheckOutputSpecsForbidRecordCompression()
{
Job job = Job.GetInstance();
FileSystem fs = FileSystem.GetLocal(job.GetConfiguration());
Path outputdir = new Path(Runtime.GetProperty("test.build.data", "/tmp") + "/output"
);
fs.Delete(outputdir, true);
// Without outputpath, FileOutputFormat.checkoutputspecs will throw
// InvalidJobConfException
FileOutputFormat.SetOutputPath(job, outputdir);
// SequenceFileAsBinaryOutputFormat doesn't support record compression
// It should throw an exception when checked by checkOutputSpecs
SequenceFileAsBinaryOutputFormat.SetCompressOutput(job, true);
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Block);
try
{
new SequenceFileAsBinaryOutputFormat().CheckOutputSpecs(job);
}
catch (Exception e)
{
Fail("Block compression should be allowed for " + "SequenceFileAsBinaryOutputFormat:Caught "
+ e.GetType().FullName);
}
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Record);
try
{
new SequenceFileAsBinaryOutputFormat().CheckOutputSpecs(job);
Fail("Record compression should not be allowed for " + "SequenceFileAsBinaryOutputFormat"
);
}
catch (InvalidJobConfException)
{
}
catch (Exception e)
{
// expected
Fail("Expected " + typeof(InvalidJobConfException).FullName + "but caught " + e.GetType
().FullName);
}
}
/// <exception cref="System.IO.IOException"/>
public virtual void TestInvalidVersionNumber()
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, 3);
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
try
{
new FileOutputCommitter(outDir, tContext);
Fail("should've thrown an exception!");
}
catch (IOException)
{
}
}
/// <exception cref="System.Exception"/>
public virtual void TestEmptyOutput()
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// Do not write any output
// do commit
committer.CommitTask(tContext);
committer.CommitJob(jContext);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.Exception"/>
public virtual void TestCommitter()
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// write output
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(tContext);
WriteOutput(theRecordWriter, tContext);
// do commit
committer.CommitTask(tContext);
committer.CommitJob(jContext);
// validate output
FilePath expectedFile = new FilePath(new Path(outDir, partFile).ToString());
StringBuilder expectedOutput = new StringBuilder();
expectedOutput.Append(key1).Append('\t').Append(val1).Append("\n");
expectedOutput.Append(val1).Append("\n");
expectedOutput.Append(val2).Append("\n");
expectedOutput.Append(key2).Append("\n");
expectedOutput.Append(key1).Append("\n");
expectedOutput.Append(key2).Append('\t').Append(val2).Append("\n");
string output = UtilsForTests.Slurp(expectedFile);
NUnit.Framework.Assert.AreEqual(output, expectedOutput.ToString());
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
//test passed
/// <exception cref="System.IO.IOException"/>
/// <exception cref="System.Exception"/>
private void TestAbortInternal(int version)
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, version);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// do setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// write output
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(tContext);
WriteOutput(theRecordWriter, tContext);
// do abort
committer.AbortTask(tContext);
FilePath expectedFile = new FilePath(new Path(committer.GetWorkPath(), partFile).
ToString());
NUnit.Framework.Assert.IsFalse("task temp dir still exists", expectedFile.Exists(
));
committer.AbortJob(jContext, JobStatus.State.Failed);
expectedFile = new FilePath(new Path(outDir, FileOutputCommitter.PendingDirName).
ToString());
NUnit.Framework.Assert.IsFalse("job temp dir still exists", expectedFile.Exists()
);
NUnit.Framework.Assert.AreEqual("Output directory not empty", 0, new FilePath(outDir
.ToString()).ListFiles().Length);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.IO.IOException"/>
/// <exception cref="System.Exception"/>
public virtual void TestBinary()
{
Configuration conf = new Configuration();
Job job = Job.GetInstance(conf);
Path outdir = new Path(Runtime.GetProperty("test.build.data", "/tmp"), "outseq");
Random r = new Random();
long seed = r.NextLong();
r.SetSeed(seed);
FileOutputFormat.SetOutputPath(job, outdir);
SequenceFileAsBinaryOutputFormat.SetSequenceFileOutputKeyClass(job, typeof(IntWritable
));
SequenceFileAsBinaryOutputFormat.SetSequenceFileOutputValueClass(job, typeof(DoubleWritable
));
SequenceFileAsBinaryOutputFormat.SetCompressOutput(job, true);
SequenceFileAsBinaryOutputFormat.SetOutputCompressionType(job, SequenceFile.CompressionType
.Block);
BytesWritable bkey = new BytesWritable();
BytesWritable bval = new BytesWritable();
TaskAttemptContext context = MapReduceTestUtil.CreateDummyMapTaskAttemptContext(job
.GetConfiguration());
OutputFormat <BytesWritable, BytesWritable> outputFormat = new SequenceFileAsBinaryOutputFormat
();
OutputCommitter committer = outputFormat.GetOutputCommitter(context);
committer.SetupJob(job);
RecordWriter <BytesWritable, BytesWritable> writer = outputFormat.GetRecordWriter(
context);
IntWritable iwritable = new IntWritable();
DoubleWritable dwritable = new DoubleWritable();
DataOutputBuffer outbuf = new DataOutputBuffer();
Log.Info("Creating data by SequenceFileAsBinaryOutputFormat");
try
{
for (int i = 0; i < Records; ++i)
{
iwritable = new IntWritable(r.Next());
iwritable.Write(outbuf);
bkey.Set(outbuf.GetData(), 0, outbuf.GetLength());
outbuf.Reset();
dwritable = new DoubleWritable(r.NextDouble());
dwritable.Write(outbuf);
bval.Set(outbuf.GetData(), 0, outbuf.GetLength());
outbuf.Reset();
writer.Write(bkey, bval);
}
}
finally
{
writer.Close(context);
}
committer.CommitTask(context);
committer.CommitJob(job);
InputFormat <IntWritable, DoubleWritable> iformat = new SequenceFileInputFormat <IntWritable
, DoubleWritable>();
int count = 0;
r.SetSeed(seed);
SequenceFileInputFormat.SetInputPaths(job, outdir);
Log.Info("Reading data by SequenceFileInputFormat");
foreach (InputSplit split in iformat.GetSplits(job))
{
RecordReader <IntWritable, DoubleWritable> reader = iformat.CreateRecordReader(split
, context);
MapContext <IntWritable, DoubleWritable, BytesWritable, BytesWritable> mcontext =
new MapContextImpl <IntWritable, DoubleWritable, BytesWritable, BytesWritable>(job
.GetConfiguration(), context.GetTaskAttemptID(), reader, null, null, MapReduceTestUtil
.CreateDummyReporter(), split);
reader.Initialize(split, mcontext);
try
{
int sourceInt;
double sourceDouble;
while (reader.NextKeyValue())
{
sourceInt = r.Next();
sourceDouble = r.NextDouble();
iwritable = reader.GetCurrentKey();
dwritable = reader.GetCurrentValue();
NUnit.Framework.Assert.AreEqual("Keys don't match: " + "*" + iwritable.Get() + ":"
+ sourceInt + "*", sourceInt, iwritable.Get());
NUnit.Framework.Assert.IsTrue("Vals don't match: " + "*" + dwritable.Get() + ":"
+ sourceDouble + "*", double.Compare(dwritable.Get(), sourceDouble) == 0);
++count;
}
}
finally
{
reader.Close();
}
}
NUnit.Framework.Assert.AreEqual("Some records not found", Records, count);
}
/// <exception cref="System.IO.IOException"/>
/// <exception cref="System.Exception"/>
private void TestFailAbortInternal(int version)
{
Job job = Job.GetInstance();
Configuration conf = job.GetConfiguration();
conf.Set(FileSystem.FsDefaultNameKey, "faildel:///");
conf.SetClass("fs.faildel.impl", typeof(TestFileOutputCommitter.FakeFileSystem),
typeof(FileSystem));
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(MRJobConfig.ApplicationAttemptId, 1);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, version);
FileOutputFormat.SetOutputPath(job, outDir);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// do setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// write output
TextOutputFormat <object, object> theOutputFormat = new TextOutputFormat();
RecordWriter <object, object> theRecordWriter = theOutputFormat.GetRecordWriter(tContext
);
WriteOutput(theRecordWriter, tContext);
// do abort
Exception th = null;
try
{
committer.AbortTask(tContext);
}
catch (IOException ie)
{
th = ie;
}
NUnit.Framework.Assert.IsNotNull(th);
NUnit.Framework.Assert.IsTrue(th is IOException);
NUnit.Framework.Assert.IsTrue(th.Message.Contains("fake delete failed"));
Path jtd = committer.GetJobAttemptPath(jContext);
FilePath jobTmpDir = new FilePath(jtd.ToUri().GetPath());
Path ttd = committer.GetTaskAttemptPath(tContext);
FilePath taskTmpDir = new FilePath(ttd.ToUri().GetPath());
FilePath expectedFile = new FilePath(taskTmpDir, partFile);
NUnit.Framework.Assert.IsTrue(expectedFile + " does not exists", expectedFile.Exists
());
th = null;
try
{
committer.AbortJob(jContext, JobStatus.State.Failed);
}
catch (IOException ie)
{
th = ie;
}
NUnit.Framework.Assert.IsNotNull(th);
NUnit.Framework.Assert.IsTrue(th is IOException);
NUnit.Framework.Assert.IsTrue(th.Message.Contains("fake delete failed"));
NUnit.Framework.Assert.IsTrue("job temp dir does not exists", jobTmpDir.Exists());
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}
/// <exception cref="System.Exception"/>
private void TestRecoveryInternal(int commitVersion, int recoveryVersion)
{
Job job = Job.GetInstance();
FileOutputFormat.SetOutputPath(job, outDir);
Configuration conf = job.GetConfiguration();
conf.Set(MRJobConfig.TaskAttemptId, attempt);
conf.SetInt(MRJobConfig.ApplicationAttemptId, 1);
conf.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, commitVersion
);
JobContext jContext = new JobContextImpl(conf, taskID.GetJobID());
TaskAttemptContext tContext = new TaskAttemptContextImpl(conf, taskID);
FileOutputCommitter committer = new FileOutputCommitter(outDir, tContext);
// setup
committer.SetupJob(jContext);
committer.SetupTask(tContext);
// write output
TextOutputFormat theOutputFormat = new TextOutputFormat();
RecordWriter theRecordWriter = theOutputFormat.GetRecordWriter(tContext);
WriteOutput(theRecordWriter, tContext);
// do commit
committer.CommitTask(tContext);
Path jobTempDir1 = committer.GetCommittedTaskPath(tContext);
FilePath jtd = new FilePath(jobTempDir1.ToUri().GetPath());
if (commitVersion == 1)
{
NUnit.Framework.Assert.IsTrue("Version 1 commits to temporary dir " + jtd, jtd.Exists
());
ValidateContent(jtd);
}
else
{
NUnit.Framework.Assert.IsFalse("Version 2 commits to output dir " + jtd, jtd.Exists
());
}
//now while running the second app attempt,
//recover the task output from first attempt
Configuration conf2 = job.GetConfiguration();
conf2.Set(MRJobConfig.TaskAttemptId, attempt);
conf2.SetInt(MRJobConfig.ApplicationAttemptId, 2);
conf2.SetInt(FileOutputCommitter.FileoutputcommitterAlgorithmVersion, recoveryVersion
);
JobContext jContext2 = new JobContextImpl(conf2, taskID.GetJobID());
TaskAttemptContext tContext2 = new TaskAttemptContextImpl(conf2, taskID);
FileOutputCommitter committer2 = new FileOutputCommitter(outDir, tContext2);
committer2.SetupJob(tContext2);
Path jobTempDir2 = committer2.GetCommittedTaskPath(tContext2);
FilePath jtd2 = new FilePath(jobTempDir2.ToUri().GetPath());
committer2.RecoverTask(tContext2);
if (recoveryVersion == 1)
{
NUnit.Framework.Assert.IsTrue("Version 1 recovers to " + jtd2, jtd2.Exists());
ValidateContent(jtd2);
}
else
{
NUnit.Framework.Assert.IsFalse("Version 2 commits to output dir " + jtd2, jtd2.Exists
());
if (commitVersion == 1)
{
NUnit.Framework.Assert.IsTrue("Version 2 recovery moves to output dir from " + jtd
, jtd.List().Length == 0);
}
}
committer2.CommitJob(jContext2);
ValidateContent(outDir);
FileUtil.FullyDelete(new FilePath(outDir.ToString()));
}