HDFs Hyper-Lease Exception Summary (org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException)

HDFs Hyper-Lease Exception Summary (org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException)reprinted February 22, 2014 14:40:58

• 9686

Exception information:

13/09/11 12:12:06 INFO HDFs. Dfsclient:small_buffer_size is 512org.apache.hadoop.ipc.remoteexception: Org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException:No Lease On/tmp/put_dir/20130911-121205-858/ie_ Ping1_access_log.2013091111.lzo File does not exist. Holder dfsclient_-2082829022 does not has any open files.        At Org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease (fsnamesystem.java:1623) at        Org.apache.hadoop.hdfs.server.namenode.FSNamesystem.checkLease (fsnamesystem.java:1614) at        Org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFileInternal (fsnamesystem.java:1669) at        Org.apache.hadoop.hdfs.server.namenode.FSNamesystem.completeFile (fsnamesystem.java:1657) at        Org.apache.hadoop.hdfs.server.namenode.NameNode.complete (namenode.java:714)

The literal understanding of the file operation over the lease period, is actually the data stream during the operation of the file was deleted. It has been encountered before, usually because mapred multiple task operations are the same file, and a task is deleted after it is finished.

This time, however, this exception occurred while uploading the file in HDFs, which caused the upload to fail. Google A, some feedback with the dfs.datanode.max.xcievers parameters to reach the upper limit. This is the task Datanode processing the request.

The upper limit, which defaults to 256, is configured to 2048 on the cluster. So I went to all the Datanode and swept the log, and found that there was the IoE:

Java.io.IOException:xceiverCount 2049 exceeds the limit of concurrent xcievers 2048

Turn the source to find a bit xcievers, there are dataxcievers and dataxcieversserver two classes, Dataxcievers is the dataxcieversserver start a thread for processing input and output data stream, its run ()

The method has the following judgment:

1 public   void Run () {2          ...       Curxceivercount int = Datanode.getxceivercount ();       if (Curxceivercount > Dataxceiverserver.maxxceivercount) {         IOException throw new ("Xceivercount" + curXceiverCount59                               + "exceeds The limit of concurrent xcievers "                               + dataxceiverserver.maxxceivercount);       

Xcievers over the limit to throw a ioexception, this reaction to the dfsclient end, is the operation of the file is lost response, so there is the above lease extended exception.

Solution:

Continue to change the Xceivercount to 8192 and restart the cluster to take effect.

HDFs Hyper-Lease Exception Summary (org.apache.hadoop.hdfs.server.namenode.LeaseExpiredException)