public void DataLakeUploader_ResumePartialUpload()
{
//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 DataLakeStoreUploader(up, frontEnd);
uploader.DeleteMetadataFile();
Assert.Throws <AggregateException>(() => uploader.Execute());
Assert.False(frontEnd.StreamExists(up.TargetStreamPath), "Target stream should not have been created");
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 DataLakeStoreUploader(up, backingFrontEnd);
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
}
public void DataLakeUploader_ResumeUploadWithAllMissingFiles()
{
//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) => { throw new IntentionalException(); }; //fail the concatenation
//attempt full upload
var up = CreateParameters(isResume: false);
var uploader = new DataLakeStoreUploader(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 DataLakeStoreUploader(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_ResumePartialUpload()
{
//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 DataLakeStoreUploader(up, frontEnd);
uploader.DeleteMetadataFile();
Assert.Throws<AggregateException>(() => uploader.Execute());
Assert.False(frontEnd.StreamExists(up.TargetStreamPath), "Target stream should not have been created");
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 DataLakeStoreUploader(up, backingFrontEnd);
try
{
uploader.Execute();
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
}
public void DataLakeUploader_ResumeUploadWithAllMissingFiles()
{
//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) => { throw new IntentionalException(); }; //fail the concatenation
//attempt full upload
var up = CreateParameters(isResume: false);
var uploader = new DataLakeStoreUploader(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 DataLakeStoreUploader(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_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 DataLakeStoreUploader(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 DataLakeStoreUploader(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 DataLakeStoreUploader(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 DataLakeStoreUploader(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_ResumePartialFolderUploadWithProgress()
{
//attempt to load the file fully, but only allow creating 1 target stream
var backingFrontEnd = new InMemoryFrontEnd();
var frontEnd = new MockableFrontEnd(backingFrontEnd);
UploadFolderProgress progress = null;
var syncRoot = new object();
IProgress <UploadFolderProgress> progressTracker = new Progress <UploadFolderProgress>(
(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.UploadedByteCount < p.UploadedByteCount)
{
progress = p;
}
}
});
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 DataLakeStoreUploader(up, frontEnd, folderProgressTracker: progressTracker);
uploader.DeleteMetadataFile();
// Verifies that a bug in folder upload with progress hung on failure is fixed.
try
{
var uploadTask = Task.Run(() =>
{
uploader.Execute();
});
uploadTask.Wait(TimeSpan.FromSeconds(60));
Assert.True(false, "Folder upload did not fail after error in less than 60 seconds");
}
catch (Exception ex)
{
Assert.True(ex is AggregateException, "The exception thrown by upload was not the expected aggregate exception.");
}
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 DataLakeStoreUploader(up, backingFrontEnd, folderProgressTracker: progressTracker);
try
{
var uploadTask = Task.Run(() =>
{
uploader.Execute();
});
uploadTask.Wait(TimeSpan.FromSeconds(60));
Assert.True(uploadTask.IsCompleted, "Folder upload did not complete after error in less than 60 seconds");
}
finally
{
uploader.DeleteMetadataFile();
}
VerifyFileUploadedSuccessfully(up, backingFrontEnd);
VerifyFolderProgressStatus(progress, _largeFileData.Length + (_smallFileData.Length * 2), 3);
}
