hadoop - datanode DataNode失败的卷上出现致命磁盘错误:

Question

我在我的namenode上收到以下日志并将其从执行中删除我的datanode

2013-02-08 03:25:54,345 WARN  namenode.NameNode (NameNodeRpcServer.java:errorReport(825)) - Fatal disk error on xxx.xxx.xxx.xxx:50010: DataNode failed volumes:/home/srikmvm/hadoop-0.23.0/tmp/current;
2013-02-08 03:25:54,349 INFO  net.NetworkTopology (NetworkTopology.java:remove(367)) - Removing a node: /default-rack/xxx.xxx.xxx.xxx:50010

有人可以建议如何纠正这一点吗？

数据节点日志:

2013-02-08 03:25:54,718 WARN datanode.DataNode (FSDataset.java:checkDirs(871)) - Removing failed volume /home/srikmvm/hadoop-0.23.0/tmp/current:   
    org.apache.hadoop.util.DiskChecker$DiskErrorException: can not create directory: /home/srikmvm/hadoop-0.23.0/tmp/current/BP-876979163-137.132.153.125-13602411944‌​23/current/finalized 
      at org.apache.hadoop.util.DiskChecker.checkDir(DiskChecker.java:87)

Answer 1

导致此错误消息的原因:

目录/目录的路径是否存在

是否正在运行datanode进程的用户具有创建/写入此目录的权限

/home/srikmvm/hadoop-0.23.0/tmp/current/BP-876979163-137.132.153.125-13602411944‌ 23 / current / finalized