hadoop2: regionserver running as process 45320. Stop it first.
hadoop3: regionserver running as process 50231. Stop it first.
hadoop1: starting regionserver, logging to /usr/share/hadoop/hbase-0.98.12/logs/hbase-shartoo-regionserver-hadoop1.out
bin/stop-hbase.sh
然后分别ssh 链接到还有进程的主机,kill掉所有未停止的进程,比如此处
ssh hadoop2
kill 45320
exit
ssh hadoop3
kill 50231
exit
然后删除所有节点 /home/hdfs/Hadoop/zookeeper/ 目录下所有文件(包括Versio-2文件夹和myid 文件)。其中hdfs为当前用户.
HBase HMaster启动失败。
原因: zookeeper信息过期。
修正方法:
* 进入zookeeper控制台:…/zookeeper/bin/zkCli.sh
* 查看目录:ls /,然后删除hbase目录:rmr /hbase,重启zk和hbase。
* 如果自己制定了HBase的根目录,务必使得它的所有者是hbase
问题网址:问题网址
“`
DEBUG org.apache.zookeeper.ClientCnxn - Reading reply sessionid:0x14e3e7226a506a7, packet:: clientPath:null serverPath:null finished:false header:: 1,3 replyHeader:: 1,4294983712,0 request:: ‘/hbase/hbaseid,F response:: s{4294967333,4294971125,1435566877143,1435571857505,2,0,0,0,67,0,4294967333}
09:30:15.041 [nioEventLoopGroup-3-1-SendThread(Slave1:2181)] DEBUG org.apache.zookeeper.ClientCnxn - Reading reply sessionid:0x14e3e7226a506a7, packet:: clientPath:null serverPath:null finished:false header:: 2,4 replyHeader:: 2,4294983712,0 request:: ‘/hbase/hbaseid,F response:: #ffffffff000146d61737465723a3630303030ffffff86fffffff8ffffffb374254dffffffb4150425546a2461343965383766362d303739342d346366642d383530642d373063313832646538323638,s{4294967333,4294971125,1435566877143,1435571857505,2,0,0,0,67,0,4294967333}
09:30:15.590 [nioEventLoopGroup-3-1-SendThread(Slave1:2181)] DEBUG org.apache.zookeeper.ClientCnxn - Reading reply sessionid:0x14e3e7226a506a7, packet:: clientPath:null serverPath:null finished:false header:: 3,4 replyHeader:: 3,4294983714,0 request:: ‘/hbase/meta-region-server,F response:: #ffffffff0001a726567696f6e7365727665723a3630303230ffffffb0ffffff83fffffff1ffffff8e53ffffffc6ffffff85b50425546a13a6536c6176653110fffffff4ffffffd4318ffffffdeffffff94ffffff80fffffff6ffffffe329100,s{4294971148,4294971148,1435571867824,1435571867824,0,0,0,0,58,0,4294971148}
思路:hbase版本和java API 使用的HBase Client版本不一致。