Hadoop问题解决记（1）

最近在测试HBase时遇到一个非常奇怪的问题：集群有7台机器，其中1台Master，6台RegionServer。但是Master只能控制其中1台RegionServer，而无法控制其他5台RegionServer。

打开master的日志文件，发现以下错误信息：

2011-04-22 16:37:21,242 WARN org.apache.hadoop.hbase.master.AssignmentManager: Failed assignment of -ROOT-,,0.70236052 to serverName=hp2,60020,1303461559353, load=(requests=0, regions=0, usedHeap=28, maxHeap=3979), trying to assign elsewhere instead; retry=0
org.apache.hadoop.hbase.client.RetriesExhaustedException: Failed setting up proxy interface org.apache.hadoop.hbase.ipc.HRegionInterface to /10.131.18.3:60020 after attempts=1
    at org.apache.hadoop.hbase.ipc.HBaseRPC.waitForProxy(HBaseRPC.java:355)
    at org.apache.hadoop.hbase.client.HConnectionManager$HConnectionImplementation.getHRegionConnection(HConnectionManager.java:965)
    at org.apache.hadoop.hbase.master.ServerManager.getServerConnection(ServerManager.java:606)
    at org.apache.hadoop.hbase.master.ServerManager.sendRegionOpen(ServerManager.java:541)
    at org.apache.hadoop.hbase.master.AssignmentManager.assign(AssignmentManager.java:920)
    at org.apache.hadoop.hbase.master.AssignmentManager.assign(AssignmentManager.java:730)
    at org.apache.hadoop.hbase.master.AssignmentManager.assign(AssignmentManager.java:710)
    at org.apache.hadoop.hbase.master.AssignmentManager.assignRoot(AssignmentManager.java:1189)
    at org.apache.hadoop.hbase.master.HMaster.assignRootAndMeta(HMaster.java:432)
    at org.apache.hadoop.hbase.master.HMaster.finishInitialization(HMaster.java:389)
    at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:283)
Caused by: java.net.ConnectException: Connection refused
    at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method)
    at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:574)
    at org.apache.hadoop.net.SocketIOWithTimeout.connect(SocketIOWithTimeout.java:206)
    at org.apache.hadoop.net.NetUtils.connect(NetUtils.java:408)
    at org.apache.hadoop.hbase.ipc.HBaseClient$Connection.setupIOstreams(HBaseClient.java:328)
    at org.apache.hadoop.hbase.ipc.HBaseClient.getConnection(HBaseClient.java:883)
    at org.apache.hadoop.hbase.ipc.HBaseClient.call(HBaseClient.java:750)
    at org.apache.hadoop.hbase.ipc.HBaseRPC$Invoker.invoke(HBaseRPC.java:257)
    at $Proxy7.getProtocolVersion(Unknown Source)
    at org.apache.hadoop.hbase.ipc.HBaseRPC.getProxy(HBaseRPC.java:419)
    at org.apache.hadoop.hbase.ipc.HBaseRPC.getProxy(HBaseRPC.java:393)
    at org.apache.hadoop.hbase.ipc.HBaseRPC.getProxy(HBaseRPC.java:444)
    at org.apache.hadoop.hbase.ipc.HBaseRPC.waitForProxy(HBaseRPC.java:349)
    ... 10 more

在这个日志中，master机器无法与IP地址为10.131.18.3的regionserver进行通信。

然后找到10.131.18.3机器，查看这台机器的regionserver日志，查看regionserver的启动信息：

2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 10 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 11 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 12 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 13 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 14 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 15 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 16 on 60020: starting
2011-04-14 18:32:05,122 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 17 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 18 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: IPC Server handler 19 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 0 on 60020: starting2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 1 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 2 on 60020: starting2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 3 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 4 on 60020: starting2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 5 on 60020: starting
2011-04-14 18:32:05,123 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 6 on 60020: starting2011-04-14 18:32:05,124 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 7 on 60020: starting
2011-04-14 18:32:05,124 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 8 on 60020: starting2011-04-14 18:32:05,124 INFO org.apache.hadoop.ipc.HBaseServer: PRI IPC Server handler 9 on 60020: starting2011-04-14 18:32:05,124 INFO org.apache.hadoop.hbase.regionserver.HRegionServer: Serving as dell4,60020,1302777124101, RPC listening on /127.0.0.1:60020, sessionid=0x12f535856620004

可以看出，这台regionserver机器启动成功了，但是RPC的监听ip地址却是本机的地址（127.0.0.1）。这样的话，master机器就无法与这台regionserver正常通信了，正确的监听地址应该是10.131.18.3才对。

查看代码，RPC监听地址的代码如下：

/** @return Bind address */
  public String getBindAddress() {
    final InetAddress addr = address.getAddress();
    if (addr != null) {
      return addr.getHostAddress();
    } else {
      LogFactory. getLog(HServerAddress.class).error( "Could not resolve the"
          + " DNS name of " + stringValue );
      return null;
    }
  }

代码没有错，看来是机器的某些配置导致java读取本机的ip地址出现了错误。最后查看这台机器的hosts文件：

[hadoop@hp2 logs]$ vi /etc/hosts
# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1               hp2 localhost.localdomain localhost
::1             localhost6.localdomain6 localhost6
10.131.18.8     dell1
10.131.18.5     dell2
10.131.18.6     dell3
10.131.18.7     dell4
10.131.18.2     hp1
10.131.18.3     hp2
10.131.18.4     hp3

问题找到了，其实是hosts文件的配置原因，接下来修改hosts文件为如下：

# Do not remove the following line, or various programs
# that require network functionality will fail.
127.0.0.1               localhost.localdomain localhost
::1             localhost6.localdomain6 localhost6
10.131.18.8     dell1
10.131.18.5     dell2
10.131.18.6     dell3
10.131.18.7     dell4
10.131.18.2     hp1
10.131.18.3     hp2
10.131.18.4     hp3

再次启动整个集群，问题解决。