主启动无法进行，保持模式直到区域上线

Question

我已经设置了一个具有两个节点的 Hbase 集群，并且我注意到警告“AssignmentManager：STUCK Region-In-Transition”，该警告不允许主节点启动。

• 节点1：observepreserve.corp.com（Master / Zookeeper）
• 节点2：knowshoe.corp.com（区域服务器）

为什么会发生这种情况以及如何解决？

在 Hbase UI 下我可以看到以下消息。

b94eb458bf643b46deaf6b00998d1f95 hbase：命名空间，，1542792846910.b94eb458bf643b46deaf6b00998d1f95。 state=OPENING，ts=Wed Nov 21 09:39:46 UTC 2018（PT18M9.696S 前），server=knewshoe.corp.com,16020,1542792833282

日志：

2018-11-21 09:40:45,900 INFO  [ReadOnlyZKClient-observepreserve.corp.com:2181@0x4068418f] zookeeper.ZooKeeper: Session: 0x167359e5ad60006 closed
2018-11-21 09:40:45,900 INFO  [ReadOnlyZKClient-observepreserve.corp.com:2181@0x4068418f-EventThread] zookeeper.ClientCnxn: EventThread shut down for session: 0x167359e5ad60006
2018-11-21 09:40:49,266 WARN  [master/observepreserve:16000:becomeActiveMaster] master.HMaster: hbase:namespace,,1542792846910.b94eb458bf643b46deaf6b00998d1f95. is NOT online; state={b94eb458bf643b46deaf6b00998d1f95 state=OPENING, ts=1542793186164, server=knewshoe.corp.com,16020,1542792833282}; ServerCrashProcedures=false. Master startup cannot progress, in holding-pattern until region onlined.
2018-11-21 09:41:46,095 WARN  [ProcExecTimeout] assignment.AssignmentManager: STUCK Region-In-Transition rit=OPENING, location=knewshoe.corp.com,16020,1542792833282, table=hbase:namespace, region=b94eb458bf643b46deaf6b00998d1f95
2018-11-21 09:41:53,267 WARN  [master/observepreserve:16000:becomeActiveMaster] master.HMaster: hbase:namespace,,1542792846910.b94eb458bf643b46deaf6b00998d1f95. is NOT online; state={b94eb458bf643b46deaf6b00998d1f95 state=OPENING, ts=1542793186164, server=knewshoe.corp.com,16020,1542792833282}; ServerCrashProcedures=false. Master startup cannot progress, in holding-pattern until region onlined.

Answer 1

是的，重新安装hbase会导致这个问题！ 这是因为旧的元数据没有被删除，您需要尝试从zk中删除hbase元数据 并重新启动hbase，一切都会好的，祝你好运。

Answer 2

我遇到了这样的问题： 警告 [...host:16000:becomeActiveMaster] master.HMaster (HMaster.java:isRegionOnline(1350)) - hbase:meta,,1.1588230740 不在线；状态={1588230740状态=打开，ts=1720351000617，服务器=wsl2host.localdomain,16201,1719929681295}; ServerCrashProcedures=true。主启动无法继续，处于保持模式，直到区域上线。

解决者：

1. zookeeper-shell.sh 本地主机

删除所有/hbase

2. hadoop fs -rm -r /hbase