Hbase master 不断死亡，声称 hbase:namespace 已经存在

Question

在今天的 hbase 节目中，我们遇到了一个问题，即 hbase master 启动然后很快就死了。我的主日志是这样的：

2014-06-20 12:52:40,469 FATAL [master:hdev01:60000] master.HMaster: Master serve
r abort: loaded coprocessors are: []
2014-06-20 12:52:40,470 FATAL [master:hdev01:60000] master.HMaster: Unhandled ex
ception. Starting shutdown.
org.apache.hadoop.hbase.TableExistsException: hbase:namespace
        at org.apache.hadoop.hbase.master.handler.CreateTableHandler.prepare(Cre
ateTableHandler.java:120)
        at org.apache.hadoop.hbase.master.TableNamespaceManager.createNamespaceT
able(TableNamespaceManager.java:232)
        at org.apache.hadoop.hbase.master.TableNamespaceManager.start(TableNames
paceManager.java:86)
        at org.apache.hadoop.hbase.master.HMaster.initNamespace(HMaster.java:106
2)
        at org.apache.hadoop.hbase.master.HMaster.finishInitialization(HMaster.j
ava:926)
        at org.apache.hadoop.hbase.master.HMaster.run(HMaster.java:615)
        at java.lang.Thread.run(Thread.java:662)
2014-06-20 12:52:40,473 INFO  [master:hdev01:60000] master.HMaster: Aborting
2014-06-20 12:52:40,473 DEBUG [master:hdev01:60000] master.HMaster: Stopping ser
vice threads
2014-06-20 12:52:40,473 INFO  [master:hdev01:60000] ipc.RpcServer: Stopping serv
er on 60000
2014-06-20 12:52:40,473 INFO  [CatalogJanitor-hdev01:60000] master.CatalogJanito
r: CatalogJanitor-hdev01:60000 exiting
2014-06-20 12:52:40,473 INFO  [hdev01,60000,1403283149823-BalancerChore] balance
r.BalancerChore: hdev01,60000,1403283149823-BalancerChore exiting
2014-06-20 12:52:40,474 INFO  [RpcServer.listener,port=60000] ipc.RpcServer: Rpc
Server.listener,port=60000: stopping
2014-06-20 12:52:40,474 INFO  [RpcServer.responder] ipc.RpcServer: RpcServer.res
ponder: stopped
2014-06-20 12:52:40,474 INFO  [master:hdev01:60000] master.HMaster: Stopping inf
oServer
2014-06-20 12:52:40,474 INFO  [RpcServer.responder] ipc.RpcServer: RpcServer.res
ponder: stopping
2014-06-20 12:52:40,474 INFO  [master:hdev01:60000.oldLogCleaner] cleaner.LogCle
aner: master:hdev01:60000.oldLogCleaner exiting
2014-06-20 12:52:40,475 INFO  [hdev01,60000,1403283149823-ClusterStatusChore] ba
lancer.ClusterStatusChore: hdev01,60000,1403283149823-ClusterStatusChore exiting

2014-06-20 12:52:40,476 INFO  [master:hdev01:60000.oldLogCleaner] master.Replica
tionLogCleaner: Stopping replicationLogCleaner-0x246ba2ab1e4001c, quorum=hdev02:
5181,hdev01:5181,hdev03:5181, baseZNode=/hbase
2014-06-20 12:52:40,479 INFO  [master:hdev01:60000] mortbay.log: Stopped SelectC
hannelConnector@0.0.0.0:16010
2014-06-20 12:52:40,478 INFO  [master:hdev01:60000.archivedHFileCleaner] cleaner
.HFileCleaner: master:hdev01:60000.archivedHFileCleaner exiting
2014-06-20 12:52:40,483 INFO  [master:hdev01:60000.oldLogCleaner] zookeeper.ZooK
eeper: Session: 0x246ba2ab1e4001c closed
2014-06-20 12:52:40,484 INFO  [master:hdev01:60000-EventThread] zookeeper.Client
Cnxn: EventThread shut down
2014-06-20 12:52:40,589 DEBUG [master:hdev01:60000] catalog.CatalogTracker: Stop
ping catalog tracker org.apache.hadoop.hbase.catalog.CatalogTracker@f3f348b
2014-06-20 12:52:40,591 INFO  [master:hdev01:60000] client.HConnectionManager$HC
onnectionImplementation: Closing zookeeper sessionid=0x246ba2ab1e4001b
2014-06-20 12:52:40,592 INFO  [master:hdev01:60000] zookeeper.ZooKeeper: Session
: 0x246ba2ab1e4001b closed
2014-06-20 12:52:40,592 INFO  [master:hdev01:60000-EventThread] zookeeper.Client
Cnxn: EventThread shut down
2014-06-20 12:52:40,695 INFO  [hdev01,60000,1403283149823.splitLogManagerTimeout
Monitor] master.SplitLogManager$TimeoutMonitor: hdev01,60000,1403283149823.split
LogManagerTimeoutMonitor exiting
2014-06-20 12:52:40,696 INFO  [master:hdev01:60000] zookeeper.ZooKeeper: Session
: 0x246ba2ab1e4001a closed
2014-06-20 12:52:40,696 INFO  [main-EventThread] zookeeper.ClientCnxn: EventThre
ad shut down
2014-06-20 12:52:40,696 INFO  [master:hdev01:60000] master.HMaster: HMaster main
 thread exiting
2014-06-20 12:52:40,697 ERROR [main] master.HMasterCommandLine: Master exiting
java.lang.RuntimeException: HMaster Aborted
        at org.apache.hadoop.hbase.master.HMasterCommandLine.startMaster(HMaster
CommandLine.java:194)
        at org.apache.hadoop.hbase.master.HMasterCommandLine.run(HMasterCommandL
ine.java:135)
        at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:70)
        at org.apache.hadoop.hbase.util.ServerCommandLine.doMain(ServerCommandLi
ne.java:126)
        at org.apache.hadoop.hbase.master.HMaster.main(HMaster.java:2803)

我认为这可能是旧运行的一些残余，所以我删除了 hbases 数据目录、zookeepers 数据目录和我的 hdfs 中的文件。我仍然遇到同样的错误。奇怪的是，当我运行 stop-hbase.sh 时，我的 HMaster popper 暂时又恢复了，尽管我无能为力。

我的 Hbase 版本是 98.3，我的 hadoop 是 2.2.0。我的 hbase-site.comf 是

<configuration>
<property>
  <name>hbase.master</name>
  <value>hdev01:60000</value>
  <description>The host and port that the HBase master runs at.
                                                     A value of 'local' runs the master and a regionserver
                                                     in a single process.
                                </description>
</property>
<property>
  <name>hbase.rootdir</name>
  <value>hdfs://hdev01:9000/hbase</value>
  <description>The directory shared by region servers.</description>
</property>
<property>
  <name>hbase.cluster.distributed</name>
  <value>true</value>
  <description>The mode the cluster will be in. Possible values are
                                false: standalone and pseudo-distributed setups with managed
                                Zookeeper true: fully-distributed with unmanaged Zookeeper
                                Quorum (see hbase-env.sh)
                                </description>
</property>
<property>
  <name>hbase.zookeeper.property.clientPort</name>
  <value>5181</value>
  <description>Property from ZooKeeper's config zoo.cfg.
    The port at which the clients will connect.
    </description>
</property>
<property>
  <name>zookeeper.session.timeout</name>
  <value>10000</value>
  <description></description>
</property>
<property>
  <name>hbase.client.retries.number</name>
  <value>10</value>
  <description></description>
</property>
<property>
  <name>hbase.zookeeper.quorum</name>
  <value>hdev01,hdev02,hdev03</value>
  <description>Comma separated list of servers in the ZooKeeper Quorum. For example, "host1.mydomain.com,host2.mydomain.com". By default this is set to localhost for local and pseudo-distributed modes of operation. For a fully-distributed setup, this should be set to a full list of ZooKeeper quorum servers. If
                                     HBASE_MANAGES_ZK is set in hbase-env.sh
                                     this is the list of servers which we will start/stop
                                     ZooKeeper on.
                </description>
</property>
</configuration>

编辑 尝试了hbase org.apache.hadoop.hbase.util.hbck.OfflineMetaRepair，我现在的错误是HBase file layout needs to be upgraded. You have version null and I want version 8. Is your hbase.rootdir valid? If so, you may need to run 'hbase hbck -fixVersionFile' 这是无益的，因为没有主 hbck 将不会实际运行。 编辑编辑 我核弹并重新启动了我的 dfs，然后尝试修复并再次启动，我现在回到了我开始的地方。

Answer 1

hbase 命名空间是 HBAse 用于其自己的管理表的内部命名空间。尝试运行离线修复工具 从 $HBASE_HOME 目录：

 ./bin/hbase org.apache.hadoop.hbase.util.hbck.OfflineMetaRepair

Answer 2

su - hdfs

hbase org.apache.hadoop.hbase.util.hbck.OfflineMetaRepair

（重新启动 hbase master。如果您仍然面临问题，请执行以下操作）

zookeeper-client（回车）

rmr /hbase

退出

然后重启hbase主服务

Answer 3

@shash: 当 HBase 管理 ZooKeeper（即 HBASE_manages_ZK=true）时，访问和清理 hbase 数据的命令是： hbase zkcli。然后你使用命令rmr /hbase 清理hbae，然后你退出。