amazon-s3 - 将 AWS S3 TransferManager 与 MultipartUpload 和 S3ProgressListener 用于 ResumableTransfer 时打开的进程过多
问题描述
我们已经使用 MultipartUpload 和 ResumableTransfer 实现了 AWS TransferManager 用于文件上传。
按照以下方式实施解决方案:
https ://aws.amazon.com/blogs/developer/pausing-and-resuming-transfers-using-transfer-manager/
https://docs.aws.amazon.com/sdk- for-java/v1/developer-guide/examples-s3-transfermanager.html
https://aws.amazon.com/blogs/mobile/pause-and-resume-amazon-s3-transfers-using-the-aws-mobile -sdk-for-android/
在没有 MultipartUpload 和 ResumableTransfer 的情况下上传文件时,进程计数得到了很好的控制,但在我们实施上述方法后开始呈指数增长。
下面的示例代码:
try {
AmazonS3 s3client = s3ClientFactory.createClient();
xferManager = TransferManagerBuilder.standard()
.withS3Client(s3client)
.withMinimumUploadPartSize(6291456L) //6 * 1024 * 1024(long) (represents 6MB)
.withMultipartUploadThreshold(6291456L) //6 * 1024 * 1024(long) (represents 6MB)
.withExecutorFactory(() -> Executors.newFixedThreadPool(3))
.build();
String resumableTargetFile ="/path/to/resumableTargetFile";
Upload upload = xferManager.upload(putRequest, new S3ProgressListener() {
ExecutorService executor = Executors.newFixedThreadPool(1);
@Override
public void progressChanged(ProgressEvent progressEvent) {
double pct = progressEvent.getBytesTransferred() * 100.0 / progressEvent.getBytes();
LOGGER.info("Upload status for file - " + fileName + " is: " + Double.toString(pct) + "%");
switch (progressEvent.getEventType()) {
case TRANSFER_STARTED_EVENT:
LOGGER.info("Started uploading file {} to S3", fileName);
break;
case TRANSFER_COMPLETED_EVENT:
LOGGER.info("Completed uploading file {} to S3", fileName);
break;
case TRANSFER_CANCELED_EVENT:
LOGGER.warn("Upload of file {} to S3 was aborted", fileName);
break;
case TRANSFER_FAILED_EVENT:
LOGGER.error("Failed uploading file {} to S3", fileName);
break;
default:
break;
}
}
@Override
public void onPersistableTransfer(final PersistableTransfer persistableTransfer) {
executor.submit(() -> {
saveTransferState(persistableTransfer, resumableTargetFile);
});
}
});
UploadResult uploadResult = upload.waitForUploadResult();
streamMD5 = uploadResult.getETag();
if (upload.isDone()) {
LOGGER.info("File {} uploaded successfully to S3 bucket {}",fileNameKey, bucketName);
}
} catch (AmazonServiceException ase) {
// The call was transmitted successfully, but Amazon S3 couldn't process
// it, so it returned an error response.
LOGGER.error("AmazonServiceException occurred: " + ase.getMessage());
} catch (SdkClientException sdce) {
// Amazon S3 couldn't be contacted for a response, or the client
// couldn't parse the response from Amazon S3.
LOGGER.error("SdkClientException occurred: " + sdce.getMessage());
} catch (AmazonClientException ace) {
LOGGER.error("AWS Exception occurred: " + ace.getMessage());
} catch (Exception e) {
LOGGER.error("Exception occurred during files processing: " + e.getMessage());
} finally {
xferManager.shutdownNow(true);
return streamMD5;
}
看看是否有人遇到过类似的问题以及有关此问题的任何意见
解决方案
尽管根据 AWS 文档,使用 TransferManager.shutdownNow(true) 关闭 TransferManager 应该会关闭 TransferManager 和相关的子对象,但我们发现在用于 ResumableTransfer 的 S3ProgressListener 中生成的 ExecutorService 在关闭 TransferManager 时从未关闭。
一旦我们通过调用 executor.shutdown() 显式关闭执行程序,打开的进程呈指数增长的问题就得到了解决
推荐阅读
- c++ - 保留对局部变量的 const 引用是 UB 吗?
- devops - 推断挂起的进程 STRACE -p PID 输出
- sql-server - 由于缺少对象,DACPAC 无法部署
- sqlite - SQLite numColumns > 0 但 fetchArray 返回 false
- copy-item - 将项目复制到多台计算机
- python - 自定义用户模型 Django 错误,没有这样的表
- apache-spark - 如何迭代一个组并使用 Pyspark 创建一个数组列?
- postgresql - 使用 Doobie 映射多对多关系
- rust - 如何集成测试补品应用
- ubuntu - 在 Ubuntu 上 sudo ufw 导致 ssh 停止工作