hbase createtable从windows失败

Question

我在ubuntu上设置了hbase 0.98.13，它看起来工作正常。我在Windows中有一个Java客户端示例程序，用于在hbase中创建表。如果我将java程序和ftp打包到ubuntu并在本地运行java客户端，它可以创建hbase表'people‘。但是，如果我从Windows运行java程序，则不会创建hbase表。hbase日志显示从我的windows PC 10.154.14.43建立了一个会话。但在那之后什么也没发生。这是我从windows运行java客户端时的hbase日志。

2015-09-02 11:05:14,477 INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.NIOServerCnxnFactory: Accepted socket connection from /10.154.14.43:59780
2015-09-02 11:05:14,479 INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.ZooKeeperServer: Client attempting to establish new session at /10.154.14.43:59780
2015-09-02 11:05:14,500 INFO  [SyncThread:0] server.ZooKeeperServer: Established session 0x14f8bf64aa20026 with negotiated timeout 40000 for client /10.154.14.43:59780
2015-09-02 11:05:19,013 DEBUG [LruStats #0] hfile.LruBlockCache: Total=415.43 KB, free=386.27 MB, max=386.68 MB, blockCount=4, accesses=99, hits=95, hitRatio=95.96%, , cachingAccesses=99, cachingHits=95, cachingHitsRatio=95.96%, evictions=89, evicted=0, evictedPerRun=0.0
2015-09-02 11:05:25,149 DEBUG [cju-linux,33912,1441162218543-BalancerChore] balancer.BaseLoadBalancer: Not running balancer because only 1 active regionserver(s)
2015-09-02 11:10:19,013 DEBUG [LruStats #0] hfile.LruBlockCache: Total=415.43 KB, free=386.27 MB, max=386.68 MB, blockCount=4, accesses=101, hits=97, hitRatio=96.04%, , cachingAccesses=101, cachingHits=97, cachingHitsRatio=96.04%, evictions=119, evicted=0, evictedPerRun=0.0
2015-09-02 11:10:25,149 DEBUG [cju-linux,33912,1441162218543-BalancerChore] balancer.BaseLoadBalancer: Not running balancer because only 1 active regionserver(s)
2015-09-02 11:15:19,013 DEBUG [LruStats #0] hfile.LruBlockCache: Total=415.43 KB, free=386.27 MB, max=386.68 MB, blockCount=4, accesses=103, hits=99, hitRatio=96.12%, , cachingAccesses=103, cachingHits=99, cachingHitsRatio=96.12%, evictions=149, evicted=0, evictedPerRun=0.0
2015-09-02 11:15:25,150 DEBUG [cju-linux,33912,1441162218543-BalancerChore] balancer.BaseLoadBalancer: Not running balancer because only 1 active regionserver(s)
2015-09-02 11:18:01,831 WARN  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.NIOServerCnxn: Exception causing close of session 0x14f8bf64aa20026 due to java.io.IOException: Connection reset by peer
2015-09-02 11:18:01,832 INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.NIOServerCnxn: Closed socket connection for client /10.154.14.43:59780 which had sessionid 0x14f8bf64aa20026
2015-09-02 11:18:42,000 INFO  [SessionTracker] server.ZooKeeperServer: Expiring session 0x14f8bf64aa20026, timeout of 40000ms exceeded
2015-09-02 11:18:42,000 INFO  [ProcessThread(sid:0 cport:-1):] server.PrepRequestProcessor: Processed session termination for sessionid: 0x14f8bf64aa20026

这是我在同一个ubuntu上运行java客户端时的日志。

2015-09-02 10:54:22,890 INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.NIOServerCnxnFactory: Accepted socket connection from /127.0.0.1:49964
2015-09-02 10:54:22,892 INFO  [NIOServerCxn.Factory:0.0.0.0/0.0.0.0:2181] server.ZooKeeperServer: Client attempting to establish new session at /127.0.0.1:49964
2015-09-02 10:54:22,909 INFO  [SyncThread:0] server.ZooKeeperServer: Established session 0x14f8bf64aa2001b with negotiated timeout 40000 for client /127.0.0.1:49964
2015-09-02 10:54:23,267 INFO  [FifoRpcScheduler.handler1-thread-14] master.HMaster: Client=hduser/null create 'people', {NAME => 'contactinfo', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'ROW', REPLICATION_SCOPE => '0', VERSIONS => '1', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => 'FOREVER', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false', BLOCKCACHE => 'true'}, {NAME => 'name', DATA_BLOCK_ENCODING => 'NONE', BLOOMFILTER => 'ROW', REPLICATION_SCOPE => '0', VERSIONS => '1', COMPRESSION => 'NONE', MIN_VERSIONS => '0', TTL => 'FOREVER', KEEP_DELETED_CELLS => 'false', BLOCKSIZE => '65536', IN_MEMORY => 'false', BLOCKCACHE => 'true'}
2015-09-02 10:54:23,269 INFO  [ProcessThread(sid:0 cport:-1):] server.PrepRequestProcessor: Got user-level KeeperException when processing sessionid:0x14f8bf64aa20000 type:create cxid:0x206 zxid:0x21d txntype:-1 reqpath:n/a Error Path:/hbase/table-lock/people Error:KeeperErrorCode = NoNode for /hbase/table-lock/people
2015-09-02 10:54:23,342 DEBUG [FifoRpcScheduler.handler1-thread-14] lock.ZKInterProcessLockBase: Acquired a lock for /hbase/table-lock/people/write-master:339120000000000
2015-09-02 10:54:23,370 INFO  [MASTER_TABLE_OPERATIONS-cju-linux:33912-0] handler.CreateTableHandler: Create table people
2015-09-02 10:54:23,382 DEBUG [MASTER_TABLE_OPERATIONS-cju-linux:33912-0] util.FSTableDescriptors: Wrote descriptor into: file:/home/hduser/hbase/.tmp/data/default/people/.tabledesc/.tableinfo.0000000001

如您所见，当我从ubuntu运行java客户端时，将显示以下代码行。

[FifoRpcScheduler.handler1-thread-14] master.HMaster: Client=hduser/null create 'people', .....

这是我的源代码。

public class CreateTable {
  public static void main(String[] args) throws IOException {
    Configuration config = HBaseConfiguration.create();

    // Example of setting zookeeper values for HDInsight
    //   in code instead of an hbase-site.xml file
    //
    config.set("hbase.zookeeper.quorum", "cju-linux");
    //config.set("hbase.zookeeper.property.clientPort", "2181");
    //config.set("hbase.cluster.distributed", "true");

    System.out.println("before creating table people 1111111111");
    // create an admin object using the config
    HBaseAdmin admin = new HBaseAdmin(config);

    System.out.println("before creating table people 22222222");
    // create the table...
    HTableDescriptor tableDescriptor = new HTableDescriptor(TableName.valueOf("people"));
    // ... with two column families
    tableDescriptor.addFamily(new HColumnDescriptor("name"));
    tableDescriptor.addFamily(new HColumnDescriptor("contactinfo"));
    System.out.println("before creating table people 33333");
    admin.createTable(tableDescriptor);
    System.out.println("before creating table people 4444");

    // define some people
    String[][] people = {
        { "1", "Marcel", "Haddad", "marcel@fabrikam.com"},
        { "2", "Franklin", "Holtz", "franklin@contoso.com" },
        { "3", "Dwayne", "McKee", "dwayne@fabrikam.com" },
        { "4", "Rae", "Schroeder", "rae@contoso.com" },
        { "5", "Rosalie", "burton", "rosalie@fabrikam.com"},
        { "6", "Gabriela", "Ingram", "gabriela@contoso.com"} };

    System.out.println("before creating table people xxxxxxxxxxxxxxxx");

    HTable table = new HTable(config, "people1");

    System.out.println("After creating table people");
    // Add each person to the table
    //   Use the `name` column family for the name
    //   Use the `contactinfo` column family for the email
    for (int i = 0; i< people.length; i++) {
      Put person = new Put(Bytes.toBytes(people[i][0]));
      person.add(Bytes.toBytes("name"), Bytes.toBytes("first"), Bytes.toBytes(people[i][1]));
      person.add(Bytes.toBytes("name"), Bytes.toBytes("last"), Bytes.toBytes(people[i][2]));
      person.add(Bytes.toBytes("contactinfo"), Bytes.toBytes("email"), Bytes.toBytes(people[i][3]));
      table.put(person);
    }
    // flush commits and close the table
    table.flushCommits();
    table.close();
  }

当我从windows运行java客户端时，打印出消息“在创建表格people 33333之前”，但没有打印出消息“在创建表格people 4444之前”。

打开debug后，运行在windows上的java客户端程序将打印出以下消息

2015-09-02 15:41:25,737 INFO  [main] client.HConnectionManager$HConnectionImplementation: getMaster attempt 2 of 35 failed; retrying after sleep of 200, exception=com.google.protobuf.ServiceException:org.apache.hadoop.hbase.ipc.RpcClient$FailedServerException: This server is in the failed servers list: localhost/127.0.0.1:48285

我已经在谷歌上搜索并更改了我在ubuntu上的/etc/hosts文件，如下所示

10.154.14.5 localhost cju-linux

我重新启动了hbase，它仍然不能工作。

如何调试此问题？

Answer 1

看起来当你在windows上运行它的时候，它在连接到集群或者说出正确的协议时会有问题。确保您的windows机器能够连接到集群中的所有机器(检查zookeeper、hmaster和region服务器主机和端口)。如果可以正常连接，请验证客户端和服务器是否都使用兼容版本(即0.98.x)。您还可以尝试单步执行createTable命令。如果您让它运行很长很长时间，当它放弃重试时，您可能还会看到一个异常，这可能会给您提供更多线索。您可以减少客户端重试次数，以使其更快实现。祝好运。

Answer 2

您应该清除Zookeeper数据目录。为此，请停止Zookeeper，在hbase-site.xml中检查hbase.zookeeper.property.dataDir属性的值，清除此目录，然后重新启动Zookeeper。