List of usage examples for com.amazonaws.services.s3.model PartETag getETag
public String getETag()
From source file:c3.ops.priam.aws.S3PartUploader.java
License:Apache License
private Void uploadPart() throws AmazonClientException, BackupRestoreException { UploadPartRequest req = new UploadPartRequest(); req.setBucketName(dataPart.getBucketName()); req.setKey(dataPart.getS3key());/*from w ww.j av a 2 s. co m*/ req.setUploadId(dataPart.getUploadID()); req.setPartNumber(dataPart.getPartNo()); req.setPartSize(dataPart.getPartData().length); req.setMd5Digest(SystemUtils.toBase64(dataPart.getMd5())); req.setInputStream(new ByteArrayInputStream(dataPart.getPartData())); UploadPartResult res = client.uploadPart(req); PartETag partETag = res.getPartETag(); if (!partETag.getETag().equals(SystemUtils.toHex(dataPart.getMd5()))) throw new BackupRestoreException("Unable to match MD5 for part " + dataPart.getPartNo()); partETags.add(partETag); return null; }
From source file:com.netflix.bdp.s3.S3Util.java
License:Apache License
public static PendingUpload multipartUpload(AmazonS3 client, File localFile, String partition, String bucket, String key, long uploadPartSize) { InitiateMultipartUploadResult initiate = client .initiateMultipartUpload(new InitiateMultipartUploadRequest(bucket, key)); String uploadId = initiate.getUploadId(); boolean threw = true; try {//ww w . ja v a 2 s . c o m Map<Integer, String> etags = Maps.newLinkedHashMap(); long offset = 0; long numParts = (localFile.length() / uploadPartSize + ((localFile.length() % uploadPartSize) > 0 ? 1 : 0)); Preconditions.checkArgument(numParts > 0, "Cannot upload 0 byte file: " + localFile); for (int partNumber = 1; partNumber <= numParts; partNumber += 1) { long size = Math.min(localFile.length() - offset, uploadPartSize); UploadPartRequest part = new UploadPartRequest().withBucketName(bucket).withKey(key) .withPartNumber(partNumber).withUploadId(uploadId).withFile(localFile) .withFileOffset(offset).withPartSize(size).withLastPart(partNumber == numParts); UploadPartResult partResult = client.uploadPart(part); PartETag etag = partResult.getPartETag(); etags.put(etag.getPartNumber(), etag.getETag()); offset += uploadPartSize; } PendingUpload pending = new PendingUpload(partition, bucket, key, uploadId, etags); threw = false; return pending; } finally { if (threw) { try { client.abortMultipartUpload(new AbortMultipartUploadRequest(bucket, key, uploadId)); } catch (AmazonClientException e) { LOG.error("Failed to abort multi-part upload", e); } } } }
From source file:com.upplication.s3fs.S3OutputStream.java
License:Open Source License
private void uploadPart(final InputStream content, final long contentLength, final byte[] checksum, final int partNumber, final boolean lastPart) throws IOException { if (aborted)/* w ww . ja va2s. c o m*/ return; final UploadPartRequest request = new UploadPartRequest(); request.setBucketName(objectId.getBucket()); request.setKey(objectId.getKey()); request.setUploadId(uploadId); request.setPartNumber(partNumber); request.setPartSize(contentLength); request.setInputStream(content); request.setLastPart(lastPart); request.setMd5Digest(Base64.encodeAsString(checksum)); final PartETag partETag = s3.uploadPart(request).getPartETag(); log.trace("Uploaded part {} with length {} for {}: {}", partETag.getPartNumber(), contentLength, objectId, partETag.getETag()); partETags.add(partETag); }
From source file:org.apache.flink.fs.s3.common.writer.S3Recoverable.java
License:Apache License
@Override public String toString() { StringBuilder buf = new StringBuilder(128); buf.append("S3Recoverable: "); buf.append("key=").append(objectName); buf.append(", uploadId=").append(uploadId); buf.append(", bytesInParts=").append(numBytesInParts); buf.append(", parts=["); int num = 0;//w w w . j a v a 2 s .co m for (PartETag part : parts) { if (0 != num++) { buf.append(", "); } buf.append(part.getPartNumber()).append('=').append(part.getETag()); } buf.append("], trailingPart=").append(lastPartObject); buf.append("trailingPartLen=").append(lastPartObjectLength); return buf.toString(); }
From source file:org.apache.nifi.processors.aws.s3.PutS3Object.java
License:Apache License
@Override public void onTrigger(final ProcessContext context, final ProcessSession session) { FlowFile flowFile = session.get();/*from w w w. j a v a 2 s .c om*/ 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); } // initiate multipart upload or find position in file //------------------------------------------------------------ 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); } } }