AWS S3JavaSDK上传失败，Groovy中的连接池关闭

Question

我正在使用AmazonJavaSDK（截至本文撰写时，最新版本为1.11.147）和Groovy（Groovy版本： 2.4.11 JVM： 1.8。0_112供应商：甲骨文公司操作系统：苹果操作系统X）将文件上传到S3。使用亚马逊的传输管理器留档说明，我能够将内容从一个存储桶复制到另一个存储桶。但是，上传总是失败。我尝试了以下3种方法。我让葡萄抓取了httpclient和httpcore的新版本，因为我在这个堆栈溢出后看到的Apache PoolingHttpClientConnectionManager抛出非法状态异常

@Grapes([
  @Grab(group='com.amazonaws', module='aws-java-sdk', version='1.11.147'),
  // https://mvnrepository.com/artifact/org.apache.commons/commons-compress
  @Grab(group='org.apache.commons', module='commons-compress', version='1.13'),
  // https://mvnrepository.com/artifact/org.apache.httpcomponents/httpclient
  @Grab(group='org.apache.httpcomponents', module='httpclient', version='4.5.3'),
  // https://mvnrepository.com/artifact/org.apache.httpcomponents/httpcore
  @Grab(group='org.apache.httpcomponents', module='httpcore', version='4.4.6')
])

// creds is String saying which ~/.aws/credentials profile to use
def credentials = new ProfileCredentialsProvider(creds)

def s3client  = AmazonS3ClientBuilder.standard().
                withCredentials(credentials).
                withRegion(region).
                build()
def tx = TransferManagerBuilder.standard().
         withS3Client(s3client).
         build()

def config_path = "/path/to/my/root"
def dir = "key_path"
def f = new File("${config_path}/${dir}/")

// Method 1, upload whole directory in one call
def mfu = tx.uploadDirectory(data_bucket, dir, f, true)
mfu.waitForCompletion() // <-- throws exception, w/o this line, just doesn't upload, but script continues


// Method 2, upload each file separately
def ld
ld = { File file ->
  file.listFiles().each { g ->
    if (g.isDirectory()) {
      ld.call(g)
    } else {
      def key = g.toString().substring(config_path_length)
        def fu = tx.upload(data_bucket, key, g)
      def fu = tx.upload(data_bucket, key, g)
      fu.waitForCompletion() // <-- throws exception, w/o this line, just doesn't upload, but script continues
    }
  }
}
ld.call(f)

// Finally Method 3, avoiding TransferManager altogether, and call putObject directly on each file
def ld
ld = { File file ->
  file.listFiles().each { g ->
    if (g.isDirectory()) {
      ld.call(g)
    } else {
      def key = g.toString().substring(config_path_length)
        def fu = tx.upload(data_bucket, key, g)
      s3client.putObject(new PutObjectRequest(data_bucket, key, g)) // <-- throws exception
    }
  }
}
ld.call(f)

然而，无论我尝试哪种方法，我都会得到以下stacktrace：

捕获：xec.executeIllegalStateException：连接池关闭xec.java:184IllegalStateException：连接池关闭在org.apache.http.impl.client.lient.do（lient.java:184）在org.apache.http.impl.client.AbstractConnPlient.execute（AbstractConnPlient.java:82）在org.apache.http.impl.conn.PoolingHttpClientConnectionManager.request连接（PoolingHttpClientConnectionManager.java:251）在com.amazonaws.http.apache.client.impl.ClientConnectionManagerFactory$Hlient.execute（ClientConnectionManagerFactory.java:76）在org.apache.http.impl.execchain.MainClientExec.execute（MainClientExec.java:175）在org.apache.http.impl.execchain.协议java.lang.（协议java.lang.）在org.apache.http.util.InternalHttpCsserts.check执行（InternalHttpCsserts.java:34）在org.apache.http.pool.CloseableHttpCool.lease（CloseableHttpCool.java:184）在com.amazonaws.http.conn.SdkHttpCandler.invoke（SdkHttpClient. java： 72）在com. amazonaws. http。Amazon HttpClient$亚马逊3Client. invoke（亚马逊3Client. java： 4221）在com. amazonaws. services. s3。亚马逊3Client. invoke（亚马逊3Client. java： 4168）在com. amazonaws. services. s3。亚马逊3Client. putObject（亚马逊3Client. java： 1718）在com. amazonaws. services. s3。亚马逊3$putObject. call（未知来源）...

我目前无法确认groovy是否正在使用更新的httpcomponent库，或者这是否是问题所在。安装在1.8 JDK中的是httpclient_4.2.6和httpclient_4.3.5。任何关于如何摆脱困境的建议都将不胜感激。谢谢你-文森特

Answer 1

我想出来了。s3client在代码的其他地方附加了另一个TransferManager。当调用tx.shutdownNow（）时，它也关闭了这个TransferManager。