use of com.amazonaws.services.s3.model.UploadPartResult in project elasticsearch by elastic.
the class TestAmazonS3 method uploadPart.
@Override
public UploadPartResult uploadPart(UploadPartRequest request) throws AmazonClientException, AmazonServiceException {
if (shouldFail(request.getBucketName(), request.getKey(), writeFailureRate)) {
long length = request.getPartSize();
long partToRead = (long) (length * randomDouble());
byte[] buffer = new byte[1024];
for (long cur = 0; cur < partToRead; cur += buffer.length) {
try (InputStream input = request.getInputStream()) {
input.read(buffer, 0, (int) (partToRead - cur > buffer.length ? buffer.length : partToRead - cur));
} catch (IOException ex) {
throw new ElasticsearchException("cannot read input stream", ex);
}
}
logger.info("--> random write failure on uploadPart method: throwing an exception for [bucket={}, key={}]", request.getBucketName(), request.getKey());
AmazonS3Exception ex = new AmazonS3Exception("Random S3 write exception");
ex.setStatusCode(400);
ex.setErrorCode("RequestTimeout");
throw ex;
} else {
return super.uploadPart(request);
}
}
use of com.amazonaws.services.s3.model.UploadPartResult in project nifi by apache.
the class PutS3Object method onTrigger.
@Override
public void onTrigger(final ProcessContext context, final ProcessSession session) {
FlowFile flowFile = session.get();
if (flowFile == null) {
return;
}
final long startNanos = System.nanoTime();
final String bucket = context.getProperty(BUCKET).evaluateAttributeExpressions(flowFile).getValue();
final String key = context.getProperty(KEY).evaluateAttributeExpressions(flowFile).getValue();
final String cacheKey = getIdentifier() + "/" + bucket + "/" + key;
final AmazonS3Client s3 = getClient();
final FlowFile ff = flowFile;
final Map<String, String> attributes = new HashMap<>();
final String ffFilename = ff.getAttributes().get(CoreAttributes.FILENAME.key());
attributes.put(S3_BUCKET_KEY, bucket);
attributes.put(S3_OBJECT_KEY, key);
final Long multipartThreshold = context.getProperty(MULTIPART_THRESHOLD).asDataSize(DataUnit.B).longValue();
final Long multipartPartSize = context.getProperty(MULTIPART_PART_SIZE).asDataSize(DataUnit.B).longValue();
final long now = System.currentTimeMillis();
/*
* If necessary, run age off for existing uploads in AWS S3 and local state
*/
ageoffS3Uploads(context, s3, now);
/*
* Then
*/
try {
session.read(flowFile, new InputStreamCallback() {
@Override
public void process(final InputStream rawIn) throws IOException {
try (final InputStream in = new BufferedInputStream(rawIn)) {
final ObjectMetadata objectMetadata = new ObjectMetadata();
objectMetadata.setContentDisposition(ff.getAttribute(CoreAttributes.FILENAME.key()));
objectMetadata.setContentLength(ff.getSize());
final String contentType = context.getProperty(CONTENT_TYPE).evaluateAttributeExpressions(ff).getValue();
if (contentType != null) {
objectMetadata.setContentType(contentType);
attributes.put(S3_CONTENT_TYPE, contentType);
}
final String expirationRule = context.getProperty(EXPIRATION_RULE_ID).evaluateAttributeExpressions(ff).getValue();
if (expirationRule != null) {
objectMetadata.setExpirationTimeRuleId(expirationRule);
}
final Map<String, String> userMetadata = new HashMap<>();
for (final Map.Entry<PropertyDescriptor, String> entry : context.getProperties().entrySet()) {
if (entry.getKey().isDynamic()) {
final String value = context.getProperty(entry.getKey()).evaluateAttributeExpressions(ff).getValue();
userMetadata.put(entry.getKey().getName(), value);
}
}
final String serverSideEncryption = context.getProperty(SERVER_SIDE_ENCRYPTION).getValue();
if (!serverSideEncryption.equals(NO_SERVER_SIDE_ENCRYPTION)) {
objectMetadata.setSSEAlgorithm(serverSideEncryption);
attributes.put(S3_SSE_ALGORITHM, serverSideEncryption);
}
if (!userMetadata.isEmpty()) {
objectMetadata.setUserMetadata(userMetadata);
}
if (ff.getSize() <= multipartThreshold) {
// ----------------------------------------
// single part upload
// ----------------------------------------
final PutObjectRequest request = new PutObjectRequest(bucket, key, in, objectMetadata);
request.setStorageClass(StorageClass.valueOf(context.getProperty(STORAGE_CLASS).getValue()));
final AccessControlList acl = createACL(context, ff);
if (acl != null) {
request.setAccessControlList(acl);
}
final CannedAccessControlList cannedAcl = createCannedACL(context, ff);
if (cannedAcl != null) {
request.withCannedAcl(cannedAcl);
}
try {
final PutObjectResult result = s3.putObject(request);
if (result.getVersionId() != null) {
attributes.put(S3_VERSION_ATTR_KEY, result.getVersionId());
}
if (result.getETag() != null) {
attributes.put(S3_ETAG_ATTR_KEY, result.getETag());
}
if (result.getExpirationTime() != null) {
attributes.put(S3_EXPIRATION_ATTR_KEY, result.getExpirationTime().toString());
}
if (result.getMetadata().getRawMetadata().keySet().contains(S3_STORAGECLASS_META_KEY)) {
attributes.put(S3_STORAGECLASS_ATTR_KEY, result.getMetadata().getRawMetadataValue(S3_STORAGECLASS_META_KEY).toString());
}
if (userMetadata.size() > 0) {
StringBuilder userMetaBldr = new StringBuilder();
for (String userKey : userMetadata.keySet()) {
userMetaBldr.append(userKey).append("=").append(userMetadata.get(userKey));
}
attributes.put(S3_USERMETA_ATTR_KEY, userMetaBldr.toString());
}
attributes.put(S3_API_METHOD_ATTR_KEY, S3_API_METHOD_PUTOBJECT);
} catch (AmazonClientException e) {
getLogger().info("Failure completing upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
throw (e);
}
} else {
// ----------------------------------------
// multipart upload
// ----------------------------------------
// load or create persistent state
// ------------------------------------------------------------
MultipartState currentState;
try {
currentState = getLocalStateIfInS3(s3, bucket, cacheKey);
if (currentState != null) {
if (currentState.getPartETags().size() > 0) {
final PartETag lastETag = currentState.getPartETags().get(currentState.getPartETags().size() - 1);
getLogger().info("Resuming upload for flowfile='{}' bucket='{}' key='{}' " + "uploadID='{}' filePosition='{}' partSize='{}' storageClass='{}' " + "contentLength='{}' partsLoaded={} lastPart={}/{}", new Object[] { ffFilename, bucket, key, currentState.getUploadId(), currentState.getFilePosition(), currentState.getPartSize(), currentState.getStorageClass().toString(), currentState.getContentLength(), currentState.getPartETags().size(), Integer.toString(lastETag.getPartNumber()), lastETag.getETag() });
} else {
getLogger().info("Resuming upload for flowfile='{}' bucket='{}' key='{}' " + "uploadID='{}' filePosition='{}' partSize='{}' storageClass='{}' " + "contentLength='{}' no partsLoaded", new Object[] { ffFilename, bucket, key, currentState.getUploadId(), currentState.getFilePosition(), currentState.getPartSize(), currentState.getStorageClass().toString(), currentState.getContentLength() });
}
} else {
currentState = new MultipartState();
currentState.setPartSize(multipartPartSize);
currentState.setStorageClass(StorageClass.valueOf(context.getProperty(STORAGE_CLASS).getValue()));
currentState.setContentLength(ff.getSize());
persistLocalState(cacheKey, currentState);
getLogger().info("Starting new upload for flowfile='{}' bucket='{}' key='{}'", new Object[] { ffFilename, bucket, key });
}
} catch (IOException e) {
getLogger().error("IOException initiating cache state while processing flow files: " + e.getMessage());
throw (e);
}
// ------------------------------------------------------------
if (currentState.getUploadId().isEmpty()) {
final InitiateMultipartUploadRequest initiateRequest = new InitiateMultipartUploadRequest(bucket, key, objectMetadata);
initiateRequest.setStorageClass(currentState.getStorageClass());
final AccessControlList acl = createACL(context, ff);
if (acl != null) {
initiateRequest.setAccessControlList(acl);
}
final CannedAccessControlList cannedAcl = createCannedACL(context, ff);
if (cannedAcl != null) {
initiateRequest.withCannedACL(cannedAcl);
}
try {
final InitiateMultipartUploadResult initiateResult = s3.initiateMultipartUpload(initiateRequest);
currentState.setUploadId(initiateResult.getUploadId());
currentState.getPartETags().clear();
try {
persistLocalState(cacheKey, currentState);
} catch (Exception e) {
getLogger().info("Exception saving cache state while processing flow file: " + e.getMessage());
throw (new ProcessException("Exception saving cache state", e));
}
getLogger().info("Success initiating upload flowfile={} available={} position={} " + "length={} bucket={} key={} uploadId={}", new Object[] { ffFilename, in.available(), currentState.getFilePosition(), currentState.getContentLength(), bucket, key, currentState.getUploadId() });
if (initiateResult.getUploadId() != null) {
attributes.put(S3_UPLOAD_ID_ATTR_KEY, initiateResult.getUploadId());
}
} catch (AmazonClientException e) {
getLogger().info("Failure initiating upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
throw (e);
}
} else {
if (currentState.getFilePosition() > 0) {
try {
final long skipped = in.skip(currentState.getFilePosition());
if (skipped != currentState.getFilePosition()) {
getLogger().info("Failure skipping to resume upload flowfile={} " + "bucket={} key={} position={} skipped={}", new Object[] { ffFilename, bucket, key, currentState.getFilePosition(), skipped });
}
} catch (Exception e) {
getLogger().info("Failure skipping to resume upload flowfile={} bucket={} " + "key={} position={} reason={}", new Object[] { ffFilename, bucket, key, currentState.getFilePosition(), e.getMessage() });
throw (new ProcessException(e));
}
}
}
// upload parts
// ------------------------------------------------------------
long thisPartSize;
for (int part = currentState.getPartETags().size() + 1; currentState.getFilePosition() < currentState.getContentLength(); part++) {
if (!PutS3Object.this.isScheduled()) {
throw new IOException(S3_PROCESS_UNSCHEDULED_MESSAGE + " flowfile=" + ffFilename + " part=" + part + " uploadId=" + currentState.getUploadId());
}
thisPartSize = Math.min(currentState.getPartSize(), (currentState.getContentLength() - currentState.getFilePosition()));
UploadPartRequest uploadRequest = new UploadPartRequest().withBucketName(bucket).withKey(key).withUploadId(currentState.getUploadId()).withInputStream(in).withPartNumber(part).withPartSize(thisPartSize);
try {
UploadPartResult uploadPartResult = s3.uploadPart(uploadRequest);
currentState.addPartETag(uploadPartResult.getPartETag());
currentState.setFilePosition(currentState.getFilePosition() + thisPartSize);
try {
persistLocalState(cacheKey, currentState);
} catch (Exception e) {
getLogger().info("Exception saving cache state processing flow file: " + e.getMessage());
}
getLogger().info("Success uploading part flowfile={} part={} available={} " + "etag={} uploadId={}", new Object[] { ffFilename, part, in.available(), uploadPartResult.getETag(), currentState.getUploadId() });
} catch (AmazonClientException e) {
getLogger().info("Failure uploading part flowfile={} part={} bucket={} key={} " + "reason={}", new Object[] { ffFilename, part, bucket, key, e.getMessage() });
throw (e);
}
}
// complete multipart upload
// ------------------------------------------------------------
CompleteMultipartUploadRequest completeRequest = new CompleteMultipartUploadRequest(bucket, key, currentState.getUploadId(), currentState.getPartETags());
try {
CompleteMultipartUploadResult completeResult = s3.completeMultipartUpload(completeRequest);
getLogger().info("Success completing upload flowfile={} etag={} uploadId={}", new Object[] { ffFilename, completeResult.getETag(), currentState.getUploadId() });
if (completeResult.getVersionId() != null) {
attributes.put(S3_VERSION_ATTR_KEY, completeResult.getVersionId());
}
if (completeResult.getETag() != null) {
attributes.put(S3_ETAG_ATTR_KEY, completeResult.getETag());
}
if (completeResult.getExpirationTime() != null) {
attributes.put(S3_EXPIRATION_ATTR_KEY, completeResult.getExpirationTime().toString());
}
if (currentState.getStorageClass() != null) {
attributes.put(S3_STORAGECLASS_ATTR_KEY, currentState.getStorageClass().toString());
}
if (userMetadata.size() > 0) {
StringBuilder userMetaBldr = new StringBuilder();
for (String userKey : userMetadata.keySet()) {
userMetaBldr.append(userKey).append("=").append(userMetadata.get(userKey));
}
attributes.put(S3_USERMETA_ATTR_KEY, userMetaBldr.toString());
}
attributes.put(S3_API_METHOD_ATTR_KEY, S3_API_METHOD_MULTIPARTUPLOAD);
} catch (AmazonClientException e) {
getLogger().info("Failure completing upload flowfile={} bucket={} key={} reason={}", new Object[] { ffFilename, bucket, key, e.getMessage() });
throw (e);
}
}
}
}
});
if (!attributes.isEmpty()) {
flowFile = session.putAllAttributes(flowFile, attributes);
}
session.transfer(flowFile, REL_SUCCESS);
final String url = s3.getResourceUrl(bucket, key);
final long millis = TimeUnit.NANOSECONDS.toMillis(System.nanoTime() - startNanos);
session.getProvenanceReporter().send(flowFile, url, millis);
getLogger().info("Successfully put {} to Amazon S3 in {} milliseconds", new Object[] { ff, millis });
try {
removeLocalState(cacheKey);
} catch (IOException e) {
getLogger().info("Error trying to delete key {} from cache: {}", new Object[] { cacheKey, e.getMessage() });
}
} catch (final ProcessException | AmazonClientException pe) {
if (pe.getMessage().contains(S3_PROCESS_UNSCHEDULED_MESSAGE)) {
getLogger().info(pe.getMessage());
session.rollback();
} else {
getLogger().error("Failed to put {} to Amazon S3 due to {}", new Object[] { flowFile, pe });
flowFile = session.penalize(flowFile);
session.transfer(flowFile, REL_FAILURE);
}
}
}
use of com.amazonaws.services.s3.model.UploadPartResult in project kafka-connect-storage-cloud by confluentinc.
the class S3SinkTaskTest method testWriteRecordsSpanningMultiplePartsWithRetry.
@Test
public void testWriteRecordsSpanningMultiplePartsWithRetry() throws Exception {
localProps.put(S3SinkConnectorConfig.FLUSH_SIZE_CONFIG, "10000");
localProps.put(S3SinkConnectorConfig.S3_PART_RETRIES_CONFIG, "3");
setUp();
List<SinkRecord> sinkRecords = createRecords(11000);
int totalBytes = calcByteSize(sinkRecords);
final int parts = totalBytes / connectorConfig.getPartSize();
// From time to time fail S3 upload part method
final AtomicInteger count = new AtomicInteger();
PowerMockito.doAnswer(new Answer<UploadPartResult>() {
@Override
public UploadPartResult answer(InvocationOnMock invocationOnMock) throws Throwable {
if (count.getAndIncrement() % parts == 0) {
throw new SdkClientException("Boom!");
} else {
return (UploadPartResult) invocationOnMock.callRealMethod();
}
}
}).when(s3).uploadPart(Mockito.isA(UploadPartRequest.class));
replayAll();
task = new S3SinkTask();
task.initialize(context);
task.start(properties);
verifyAll();
task.put(sinkRecords);
task.close(context.assignment());
task.stop();
long[] validOffsets = { 0, 10000 };
verify(sinkRecords, validOffsets);
}
use of com.amazonaws.services.s3.model.UploadPartResult in project bender by Nextdoor.
the class S3Transport method sendStream.
protected void sendStream(InputStream input, String key, long streamSize) throws TransportException {
/*
* Create metadata
*/
ObjectMetadata metadata = new ObjectMetadata();
/*
* Find if a multipart upload has already begun or start a new one.
*/
MultiPartUpload upload;
synchronized (multiPartUploads) {
if (!multiPartUploads.containsKey(key)) {
InitiateMultipartUploadRequest uploadRequest = new InitiateMultipartUploadRequest(bucketName, key);
uploadRequest.setObjectMetadata(metadata);
InitiateMultipartUploadResult res = client.initiateMultipartUpload(uploadRequest);
upload = new MultiPartUpload(bucketName, key, res.getUploadId());
multiPartUploads.put(key, upload);
} else {
upload = multiPartUploads.get(key);
}
}
/*
* Write out to S3. Note that the S3 client auto closes the input stream.
*/
UploadPartRequest req = upload.getUploadPartRequest().withInputStream(input).withPartSize(streamSize);
try {
UploadPartResult res = client.uploadPart(req);
upload.addPartETag(res.getPartETag());
} catch (AmazonClientException e) {
client.abortMultipartUpload(upload.getAbortMultipartUploadRequest());
throw new TransportException("unable to put file" + e, e);
} finally {
try {
input.close();
} catch (IOException e) {
logger.warn("error encountered while closing input stream", e);
}
}
}
use of com.amazonaws.services.s3.model.UploadPartResult in project bender by Nextdoor.
the class S3TransporterTest method testAmazonClientException.
@Test(expected = TransportException.class)
public void testAmazonClientException() throws TransportException, IllegalStateException, IOException {
/*
* Create mock client, requets, and replies
*/
AmazonS3Client mockClient = mock(AmazonS3Client.class);
UploadPartResult uploadResult = new UploadPartResult();
uploadResult.setETag("foo");
doThrow(new AmazonClientException("expected")).when(mockClient).uploadPart(any(UploadPartRequest.class));
InitiateMultipartUploadResult initUploadResult = new InitiateMultipartUploadResult();
initUploadResult.setUploadId("123");
doReturn(initUploadResult).when(mockClient).initiateMultipartUpload(any(InitiateMultipartUploadRequest.class));
/*
* Fill buffer with mock data
*/
S3TransportBuffer buffer = new S3TransportBuffer(1000, false, new S3TransportSerializer());
InternalEvent mockIevent = mock(InternalEvent.class);
doReturn("foo").when(mockIevent).getSerialized();
/*
* Create transport
*/
Map<String, MultiPartUpload> multiPartUploads = new HashMap<String, MultiPartUpload>(0);
S3Transport transport = new S3Transport(mockClient, "bucket", "basepath", false, multiPartUploads);
/*
* Do actual test
*/
buffer.add(mockIevent);
LinkedHashMap<String, String> partitions = new LinkedHashMap<String, String>();
partitions.put(S3Transport.FILENAME_KEY, "a_filename");
ArgumentCaptor<UploadPartRequest> argument = ArgumentCaptor.forClass(UploadPartRequest.class);
try {
transport.sendBatch(buffer, partitions, new TestContext());
} catch (Exception e) {
assertEquals(e.getCause().getClass(), AmazonClientException.class);
throw e;
}
}
Aggregations