【使用时发生的意外】HDFS 分布式写入问题 AlreadyBeingCreatedException

进行追加文件时出现AlreadyBeingCreatedException错误

堆栈信息大致如下：

org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.hdfs.protocol.AlreadyBeingCreatedException): Failed to create file [/secsight/log2//p0001] for [DFSClient_NONMAPREDUCE_200580206_1756] for client [192.168.10.117], because this file is already being created by [DFSClient_NONMAPREDUCE_-2109133545_2516] on [192.168.10.117]
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.recoverLeaseInternal(FSNamesystem.java:)
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.appendFileInternal(FSNamesystem.java:)
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.appendFileInt(FSNamesystem.java:)
    at org.apache.hadoop.hdfs.server.namenode.FSNamesystem.appendFile(FSNamesystem.java:)
    at org.apache.hadoop.hdfs.server.namenode.NameNodeRpcServer.append(NameNodeRpcServer.java:)
    at org.apache.hadoop.hdfs.server.namenode.AuthorizationProviderProxyClientProtocol.append(AuthorizationProviderProxyClientProtocol.java:)
    at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolServerSideTranslatorPB.append(ClientNamenodeProtocolServerSideTranslatorPB.java:)
    at org.apache.hadoop.hdfs.protocol.proto.ClientNamenodeProtocolProtos$ClientNamenodeProtocol$.callBlockingMethod(ClientNamenodeProtocolProtos.java)
    at org.apache.hadoop.ipc.ProtobufRpcEngine$Server$ProtoBufRpcInvoker.call(ProtobufRpcEngine.java:)
    at org.apache.hadoop.ipc.RPC$Server.call(RPC.java:)
    at org.apache.hadoop.ipc.Server$Handler$.run(Server.java:)
    at org.apache.hadoop.ipc.Server$Handler$.run(Server.java:)
    at java.security.AccessController.doPrivileged(Native Method)
    at javax.security.auth.Subject.doAs(Subject.java:)
    at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:)
    at org.apache.hadoop.ipc.Server$Handler.run(Server.java:)

    at org.apache.hadoop.ipc.Client.call(Client.java:)
    at org.apache.hadoop.ipc.Client.call(Client.java:)
    at org.apache.hadoop.ipc.ProtobufRpcEngine$Invoker.invoke(ProtobufRpcEngine.java:)
    at com.sun.proxy.$Proxy9.append(Unknown Source)
    at org.apache.hadoop.hdfs.protocolPB.ClientNamenodeProtocolTranslatorPB.append(ClientNamenodeProtocolTranslatorPB.java:)
    at sun.reflect.GeneratedMethodAccessor27.invoke(Unknown Source)
    at sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:)
    at java.lang.reflect.Method.invoke(Method.java:)
    at org.apache.hadoop.io.retry.RetryInvocationHandler.invokeMethod(RetryInvocationHandler.java:)
    at org.apache.hadoop.io.retry.RetryInvocationHandler.invoke(RetryInvocationHandler.java:)
    at com.sun.proxy.$Proxy10.append(Unknown Source)
    at org.apache.hadoop.hdfs.DFSClient.callAppend(DFSClient.java:)
    at org.apache.hadoop.hdfs.DFSClient.append(DFSClient.java:)
    at org.apache.hadoop.hdfs.DFSClient.append(DFSClient.java:)
    at org.apache.hadoop.hdfs.DistributedFileSystem$.doCall(DistributedFileSystem.java:)
    at org.apache.hadoop.hdfs.DistributedFileSystem$.doCall(DistributedFileSystem.java:)
    at org.apache.hadoop.fs.FileSystemLinkResolver.resolve(FileSystemLinkResolver.java:)
    at org.apache.hadoop.hdfs.DistributedFileSystem.append(DistributedFileSystem.java:)
    at org.apache.hadoop.fs.FileSystem.append(FileSystem.java:)
    at com.ultrapower.hdfs.HdfsUtils.appendFile(HdfsUtils.java:)
    at com.ultrapower.secsight.Runner.lambda$main$(Runner.java:)
    at java.lang.Thread.run(Thread.java:)

目前得到的可能原因：

　　多进程进行同一文件的写入在HDFS中是可能引发这种错误的。

　　hadoop 的dfs里边有个lease manager 维护了文件path －> lease和 DFSClient name -> lease -> path (多个) 的映射关系，我估计是这个lease的问题，看下是不是被close（），而未来的及释放的lease造成的。

可能引起该错误的代码：https://www.programcreek.com/java-api-examples/index.php?api=org.apache.hadoop.hdfs.protocol.AlreadyBeingCreatedException

/** Test two consecutive appends on a file with a full block. */
@Test
public void testAppendTwice() throws Exception {
  Configuration conf = new HdfsConfiguration();
  MiniDFSCluster cluster = new MiniDFSCluster.Builder(conf).build();
  final FileSystem fs1 = cluster.getFileSystem();
  final FileSystem fs2 = AppendTestUtil.createHdfsWithDifferentUsername(conf);
  try {

    final Path p = new Path("/testAppendTwice/foo");
    final int len =  << ;
    final byte[] fileContents = AppendTestUtil.initBuffer(len);

    {
      // create a new file with a full block.
      FSDataOutputStream out = fs2.create(p, true, , (short), len);
      out.write(fileContents, , len);
      out.close();
    }

    //1st append does not add any data so that the last block remains full
    //and the last block in INodeFileUnderConstruction is a BlockInfo
    //but not BlockInfoUnderConstruction.
    fs2.append(p);

    //2nd append should get AlreadyBeingCreatedException
    fs1.append(p);
    Assert.fail();
  } catch(RemoteException re) {
    AppendTestUtil.LOG.info("Got an exception:", re);
    Assert.assertEquals(AlreadyBeingCreatedException.class.getName(),
        re.getClassName());
  } finally {
    fs2.close();
    fs1.close();
    cluster.shutdown();
  }
}
 

https://issues.apache.org/jira/browse/HDFS-11367

https://issues.apache.org/jira/browse/HDFS-7203