public void DataLakeUploader_CancelUpload()
{
CancellationTokenSource myTokenSource = new CancellationTokenSource();
var cancelToken = myTokenSource.Token;
var frontEnd = new InMemoryFrontEnd();
var mockedFrontend = new MockableFrontEnd(frontEnd);
mockedFrontend.GetStreamLengthImplementation = (streamPath, isDownload) =>
{
// sleep for 2 second to allow for the cancellation to actual happen
Thread.Sleep(2000);
return(frontEnd.GetStreamLength(streamPath, isDownload));
};
mockedFrontend.StreamExistsImplementation = (streamPath, isDownload) =>
{
// sleep for 2 second to allow for the cancellation to actual happen
Thread.Sleep(2000);
return(frontEnd.StreamExists(streamPath, isDownload));
};
var up = CreateParameters(isResume: false);
TransferProgress progress = null;
var syncRoot = new object();
IProgress <TransferProgress> progressTracker = new Progress <TransferProgress>(
(p) =>
{
lock (syncRoot)
{
//it is possible that these come out of order because of race conditions (multiple threads reporting at the same time); only update if we are actually making progress
if (progress == null || progress.TransferredByteCount < p.TransferredByteCount)
{
progress = p;
}
}
});
var uploader = new DataLakeStoreTransferClient(up, mockedFrontend, cancelToken, progressTracker);
Task uploadTask = Task.Run(() =>
{
uploader.Execute();
Thread.Sleep(2000);
}, cancelToken);
myTokenSource.Cancel();
Assert.True(cancelToken.IsCancellationRequested);
while (uploadTask.Status == TaskStatus.Running || uploadTask.Status == TaskStatus.WaitingToRun)
{
Thread.Sleep(250);
}
// Verify that the file did not get uploaded completely.
Assert.False(frontEnd.StreamExists(up.TargetStreamPath), "Uploaded stream exists when it should not yet have been completely created");
}
internal void TestRetryBlock(int failCount)
{
bool expectSuccess = failCount < SingleSegmentUploader.MaxBufferUploadAttemptCount;
int callCount = 0;
var workingFrontEnd = new InMemoryFrontEnd();
var fe = new MockableFrontEnd(workingFrontEnd);
fe.CreateStreamImplementation =
(streamPath, overwrite, data, byteCount) =>
{
callCount++;
if (callCount <= failCount)
{
throw new IntentionalException();
}
workingFrontEnd.CreateStream(streamPath, overwrite, data, byteCount);
};
fe.AppendToStreamImplementation =
(streamPath, data, offset, byteCount) =>
{
callCount++;
if (callCount <= failCount)
{
throw new IntentionalException();
}
workingFrontEnd.AppendToStream(streamPath, data, offset, byteCount);
};
var metadata = CreateMetadata(_smallFilePath, _smallFileContents.Length);
var progressTracker = new TestProgressTracker();
var ssu = new SingleSegmentUploader(0, metadata, fe, progressTracker);
ssu.UseBackOffRetryStrategy = false;
if (expectSuccess)
{
ssu.Upload();
var actualContents = workingFrontEnd.GetStreamContents(StreamPath);
AssertExtensions.AreEqual(_smallFileContents, actualContents, "Unexpected uploaded stream contents.");
}
else
{
Assert.Throws <IntentionalException>(() => { ssu.Upload(); });
}
VerifyTracker(progressTracker, expectSuccess);
}
public void DataLakeUploader_ResumePartialFolderUpload()
{
//attempt to load the file fully, but only allow creating 1 target stream
var backingFrontEnd = new InMemoryFrontEnd();
var frontEnd = new MockableFrontEnd(backingFrontEnd);
int createStreamCount = 0;
frontEnd.CreateStreamImplementation = (path, overwrite, data, byteCount) =>
{
createStreamCount++;
if (createStreamCount > 1)
{
//we only allow 1 file to be created
throw new IntentionalException();
}
backingFrontEnd.CreateStream(path, overwrite, data, byteCount);
};
var up = CreateParameters(isResume: false, isRecursive: true);
var uploader = new DataLakeStoreTransferClient(up, frontEnd);
uploader.DeleteMetadataFile();
Assert.Throws <AggregateException>(() => uploader.Execute());
Assert.Equal(1, frontEnd.ListDirectory(up.TargetStreamPath, false).Keys.Count);
Assert.Equal(1, backingFrontEnd.StreamCount);
//resume the upload but point it to the real back-end, which doesn't throw exceptions
up = CreateParameters(isResume: true, isRecursive: true);
uploader = new DataLakeStoreTransferClient(up, backingFrontEnd);
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
}
public void SingleSegmentUploader_VerifyUploadStreamFails()
{
//create a mock front end which doesn't do anything
var workingFrontEnd = new InMemoryFrontEnd();
var fe = new MockableFrontEnd(workingFrontEnd);
fe.CreateStreamImplementation = (streamPath, overwrite, data, byteCount) => { };
fe.DeleteStreamImplementation = (streamPath, recurse, isDownload) => { };
fe.StreamExistsImplementation = (streamPath, isDownload) => { return(true); };
fe.AppendToStreamImplementation = (streamPath, data, offset, byteCount) => { };
fe.GetStreamLengthImplementation = (streamPath, isDownload) => { return(0); };
//upload some data
var metadata = CreateMetadata(_smallFilePath, _smallFileContents.Length);
var ssu = new SingleSegmentUploader(0, metadata, fe);
ssu.UseBackOffRetryStrategy = false;
//the Upload method should fail if it cannot verify that the stream was uploaded after the upload (i.e., it will get a length of 0 at the end)
Assert.Throws <TransferFailedException>(() => { ssu.Upload(); });
}
public void DataLakeUploader_ResumePartialUploadDownload()
{
//attempt to load the file fully, but only allow creating 1 target stream
var backingFrontEnd = new InMemoryFrontEnd();
var frontEnd = new MockableFrontEnd(backingFrontEnd);
int createStreamCount = 0;
frontEnd.CreateStreamImplementation = (path, overwrite, data, byteCount) =>
{
createStreamCount++;
if (createStreamCount > 1)
{
//we only allow 1 file to be created
throw new IntentionalException();
}
backingFrontEnd.CreateStream(path, overwrite, data, byteCount);
};
var up = CreateParameters(isResume: false);
var uploader = new DataLakeStoreTransferClient(up, frontEnd);
uploader.DeleteMetadataFile();
Assert.Throws <AggregateException>(() => uploader.Execute());
Assert.Equal(1, frontEnd.ListDirectory(up.TargetStreamPath, false).Keys.Count);
Assert.Equal(1, backingFrontEnd.StreamCount);
//resume the upload but point it to the real back-end, which doesn't throw exceptions
up = CreateParameters(isResume: true);
uploader = new DataLakeStoreTransferClient(up, backingFrontEnd);
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
// now download the same way.
var frontEnd2 = new MockableFrontEnd(backingFrontEnd); // need to have data from the successful upload available.
createStreamCount = 0;
frontEnd2.ReadStreamImplementation = (path, data, byteCount, isDownload) =>
{
createStreamCount++;
if (createStreamCount > 1)
{
//we only allow 1 file to be created
throw new IntentionalException();
}
return(backingFrontEnd.ReadStream(path, data, byteCount, isDownload));
};
up = CreateParameters(isResume: false, isDownload: true, targetStreamPath: _downloadFilePath, isOverwrite: true, filePath: up.TargetStreamPath);
uploader = new DataLakeStoreTransferClient(up, frontEnd2);
Assert.Throws <AggregateException>(() => uploader.Execute());
Assert.False(frontEnd2.StreamExists(up.TargetStreamPath), "Target stream should not have been created");
// now use the good front end
up = CreateParameters(isResume: true, isDownload: true, targetStreamPath: _downloadFilePath, isOverwrite: true, filePath: up.InputFilePath);
uploader = new DataLakeStoreTransferClient(up, backingFrontEnd);
//resume the download but point it to the real back-end, which doesn't throw exceptions
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
}
public void DataLakeUploader_ResumeUploadDownloadWithAllMissingFiles()
{
//this scenario is achieved by refusing to execute the concat command on the front end for the initial upload (which will interrupt it)
//and then resuming the upload against a fresh front-end (which obviously has no files there)
var backingFrontEnd1 = new InMemoryFrontEnd();
var frontEnd1 = new MockableFrontEnd(backingFrontEnd1);
frontEnd1.ConcatenateImplementation = (target, inputs, isDownload) => { throw new IntentionalException(); }; //fail the concatenation
//attempt full upload
var up = CreateParameters(isResume: false);
var uploader = new DataLakeStoreTransferClient(up, frontEnd1);
uploader.DeleteMetadataFile();
Assert.Throws <IntentionalException>(() => uploader.Execute());
Assert.False(frontEnd1.StreamExists(up.TargetStreamPath), "Target stream should not have been created");
Assert.True(0 < backingFrontEnd1.StreamCount, "No temporary streams seem to have been created");
//attempt to resume the upload
var frontEnd2 = new InMemoryFrontEnd();
up = CreateParameters(isResume: true);
uploader = new DataLakeStoreTransferClient(up, frontEnd2);
//at this point the metadata exists locally but there are no target files in frontEnd2
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, frontEnd2);
// now download the same way.
var frontEnd3 = new MockableFrontEnd(frontEnd2); // need to have data from the successful upload available.
frontEnd3.ConcatenateImplementation = (target, inputs, isDownload) => { throw new IntentionalException(); }; //fail the concatenation
up = CreateParameters(isResume: false, isDownload: true, targetStreamPath: _downloadFilePath, isOverwrite: true, filePath: up.TargetStreamPath);
uploader = new DataLakeStoreTransferClient(up, frontEnd3);
Assert.Throws <IntentionalException>(() => uploader.Execute());
Assert.False(frontEnd1.StreamExists(up.TargetStreamPath, true), "Target stream should not have been created");
// now use the good front end
up = CreateParameters(isResume: true, isDownload: true, targetStreamPath: _downloadFilePath, isOverwrite: true, filePath: up.InputFilePath);
uploader = new DataLakeStoreTransferClient(up, frontEnd2);
//at this point the metadata exists locally but there are no target files in frontEnd2
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, frontEnd2);
}
public void DataLakeUploader_FreshFolderUploadDownload()
{
var frontEnd = new InMemoryFrontEnd();
var up = CreateParameters(isResume: false, isRecursive: true);
// set the per file thread count to the default to validate computed values.
up.PerFileThreadCount = -1;
TransferFolderProgress progress = null;
var syncRoot = new object();
IProgress <TransferFolderProgress> progressTracker = new Progress <TransferFolderProgress>(
(p) =>
{
lock (syncRoot)
{
//it is possible that these come out of order because of race conditions (multiple threads reporting at the same time); only update if we are actually making progress
if (progress == null || progress.TransferredByteCount < p.TransferredByteCount)
{
progress = p;
}
}
});
var uploader = new DataLakeStoreTransferClient(up, frontEnd, null, progressTracker);
uploader.Execute();
VerifyFileUploadedSuccessfully(up, frontEnd);
VerifyFolderProgressStatus(progress, _largeFileData.Length + (_smallFileData.Length * 2), 3);
// verify that per file thread count is different but concurrent is the same.
Assert.True(up.PerFileThreadCount > 0 && uploader.Parameters.PerFileThreadCount > 0);
Assert.Equal(2, uploader.Parameters.ConcurrentFileCount);
// now download
var downloadFrontEnd = new MockableFrontEnd(frontEnd);
// replace the isDirectory implementation to return true
downloadFrontEnd.IsDirectoryImplementation = (streamPath) => { return(true); };
progress = null;
up = CreateParameters(isRecursive: true, isResume: false, isDownload: true, targetStreamPath: Path.GetDirectoryName(_downloadFilePath), isOverwrite: true, filePath: TargetStreamPath);
// set concurrentFileCount to default and validate that it changed.
up.ConcurrentFileCount = -1;
uploader = new DataLakeStoreTransferClient(up, downloadFrontEnd, null, progressTracker);
uploader.Execute();
VerifyFileUploadedSuccessfully(up, downloadFrontEnd.BaseAdapter);
VerifyFolderProgressStatus(progress, _largeFileData.Length + (_smallFileData.Length * 2), 3);
Assert.True(up.ConcurrentFileCount > 0 && uploader.Parameters.ConcurrentFileCount > 0);
Assert.Equal(ThreadCount, uploader.Parameters.PerFileThreadCount);
// run it one more time with both as defaults
up.PerFileThreadCount = -1;
up.ConcurrentFileCount = -1;
uploader = new DataLakeStoreTransferClient(up, downloadFrontEnd, null, progressTracker);
uploader.Execute();
VerifyFileUploadedSuccessfully(up, downloadFrontEnd.BaseAdapter);
VerifyFolderProgressStatus(progress, _largeFileData.Length + (_smallFileData.Length * 2), 3);
Assert.True(up.ConcurrentFileCount > 0 && uploader.Parameters.ConcurrentFileCount > 0);
Assert.True(up.PerFileThreadCount > 0 && uploader.Parameters.PerFileThreadCount > 0);
}
private void TestRetry(int segmentFailCount)
{
//we only have access to the underlying FrontEnd, so we need to simulate many exceptions in order to force a segment to fail the upload (multiply by SingleSegmentUploader.MaxBufferUploadAttemptAccount)
//this only works because we have a small file, which we know will fit in only one buffer (for a larger file, more complex operations are necessary)
int actualfailCount = segmentFailCount * SingleSegmentUploader.MaxBufferUploadAttemptCount;
bool expectSuccess = segmentFailCount < MultipleSegmentUploader.MaxUploadAttemptCount;
int callCount = 0;
//create a mock front end sitting on top of a working front end that simulates some erros for some time
var workingFrontEnd = new InMemoryFrontEnd();
var fe = new MockableFrontEnd(workingFrontEnd);
fe.CreateStreamImplementation =
(streamPath, overwrite, data, byteCount) =>
{
callCount++;
if (callCount <= actualfailCount)
{
throw new IntentionalException();
}
workingFrontEnd.CreateStream(streamPath, overwrite, data, byteCount);
};
fe.AppendToStreamImplementation =
(streamPath, data, offset, byteCount) =>
{
callCount++;
if (callCount <= actualfailCount)
{
throw new IntentionalException();
}
workingFrontEnd.AppendToStream(streamPath, data, offset, byteCount);
};
var metadata = CreateMetadata(1);
try
{
var msu = new MultipleSegmentUploader(metadata, 1, fe);
msu.UseSegmentBlockBackOffRetryStrategy = false;
if (expectSuccess)
{
//the Upload method should not throw any exceptions in this case
msu.Upload();
//if we are expecting success, verify that both the metadata and the target streams are complete
VerifyTargetStreamsAreComplete(metadata, workingFrontEnd);
}
else
{
//the Upload method should throw an aggregate exception in this case
Assert.Throws <AggregateException>(() => { msu.Upload(); });
//if we do not expect success, verify that at least 1 segment was marked as Failed
Assert.True(metadata.Segments.Any(s => s.Status == SegmentTransferStatus.Failed), "Could not find any failed segments");
//for every other segment, verify it was completed OK
foreach (var segment in metadata.Segments.Where(s => s.Status != SegmentTransferStatus.Failed))
{
VerifyTargetStreamIsComplete(segment, metadata, workingFrontEnd);
}
}
}
finally
{
metadata.DeleteFile();
}
}
