kudu1.10基于cdh6.3.1搭建

1、下载kudu依赖： yum -y install cyrus-sasl-plain ntp

 

2、下载kudu rpm包：

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-client-devel-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-client0-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-debuginfo-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-master-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

wget https://archive.cloudera.com/cdh6/6.3.1/redhat7/yum/RPMS/x86_64/kudu-tserver-1.10.0+cdh6.3.1-1470567.el7.x86_64.rpm

 

3、kudu安装部署分布：

机器	服务
cdh01	kudu-master，kudu-tserver
cdh02	kudu-master，kudu-tserver
cdh03	kudu-master，kudu-tserver

 

4、进入目录：/etc/kudu/conf

修改master.gflagfile文件如下：

 

# Do not modify these two lines. If you wish to change these variables,
# modify them in /etc/default/kudu-master.
## Comma-separated list of the RPC addresses belonging to all Masters in this cluster.
## NOTE: if not specified, configures a non-replicated Master.

--fromenv=rpc_bind_addresses
--fromenv=log_dir
--fs_wal_dir=/var/lib/kudu/master
--fs_data_dirs=/var/lib/kudu/master
--master_addresses=cdh02:7051,cdh03:7051

## You can avoid the dependency on ntpd by running Kudu with --use-hybrid-clock=false
## This is not recommended for production environment.
## NOTE: If you run without hybrid time the tablet history GC will not work.
## Therefore when you delete or update a row the history of that data will be kept
## forever. Eventually you may run out of disk space.
#--use_hybrid_clock=false

 

修改tserver.gflagfile文件如下：

# Do not modify these two lines. If you wish to change these variables,
# modify them in /etc/default/kudu-tserver.
--fromenv=rpc_bind_addresses
--fromenv=log_dir
--fs_wal_dir=/var/lib/kudu/tserver
--fs_data_dirs=/var/lib/kudu/tserver
--tserver_master_addrs=node02:7051,node03:7051

5、上面两个文件修改好之后分发到上述三台机器

 

启动方式：进入/etc/init.d目录：

service kudu-master restart

service kudu-tserver restart

6、启动后在/var/log/kudu目录进行日志的查看

启动成功在后台查看端口：

master：8051

tserver：8050

进入浏览器进行IP:端口 验证kudu服务是否正常

 

kudu常见报错，摘自：

https://blog.csdn.net/m0_38003171/article/details/79789455

整理如下：

 

第一种报错：

Log file created at: 2018/04/02 10:40:33
Running on machine: mrj001
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
F0402 10:40:33.091943 51977 master_options.cc:46] At least 2 masters are required for a distributed config, but master_addresses flag ('192.168.137.6') only specifies 1 masters.

报错分析：

至少需要配置两个master，但是只配置了一个

报错处理：

    kudu-master支持多master备份机制，所以配置时需注意；

    vi /etc/kudu/kudu-master

    注释掉--master-addresses 

    该参数配置为master节点以及ip列表，不指定默认为单master节点；如果指定了该配置，但是只有一个master节点就会报错；

 

 第二种报错：

Log file created at: 2018/04/02 10:46:39
Running on machine: mrj001
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg
F0402 10:46:39.101296 52137 tablet_server_main.cc:72] Check failed: _s.ok() Bad status: IO error: Failed to load FS layout: Could not lock /soft/kudu/data/data/block_manager_instance: Could not lock /soft/kudu/data/data/block_manager_instance: lock /soft/kudu/data/data/block_manager_instance: Resource temporarily unavailable (error 11)

报错分析：

kudu-tserver启动报错，对目录/soft/kudu/data/data/block_manager_instance不能够进行锁定

 

 

 第三种报错：

Log file created at: 2018/04/06 16:13:39
Running on machine: mrj001
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg

F0406 16:13:39.828665  2572 master_main.cc:68] Check failed: _s.ok() Bad status: Service unavailable: Cannot initialize clock: Error reading clock. Clock considered unsynchronized

报错分析：

kudu在启动之前必须保证各个服务器ntp一致，当各个table server不一致时会导致此报错；

解决办法：

查看ntpd服务是否开启：service ntpd status

查看ntpd服务是否已经同步：ntpstat

如果没有安装ntp服务，请参照本人所写博客：https://blog.csdn.net/m0_38003171/article/details/79789557

 

 第四种报错：

Log file created at: 2018/04/07 13:24:52
Running on machine: mrj001
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg

F0407 13:24:52.627974  5771 master_main.cc:68] Check failed: _s.ok() Bad status: IO error: Could not create new FS layout: Unable to write instance metadata: Call to mkstemp() failed on name template /soft/kudu/master/data/instance.kudutmp.XXXXXX: Permission denied (error 13)

报错分析：

经过分析报错可以得出：/soft/kudu/master/data/目录下没有创建文件的权限，同时kudu在启动的时候默认是使用kudu用户进行创建，所以要确定kudu用户能够对该文件夹下面的所有目录进行管理

报错处理：

在没有权限的目录下面执行：chown -R kudu:kudu ./*,改变用户组和用户为kudu即可解决

第五种报错：

Log file created at: 2018/04/07 13:37:05
Running on machine: mrj001
Log line format: [IWEF]mmdd hh:mm:ss.uuuuuu threadid file:line] msg

F0407 13:37:05.960657  5931 tablet_server_main.cc:72] Check failed: _s.ok() Bad status: IO error: Failed to load FS layout: Filesystem root /soft/kudu/tserver/wal                                                                                                   contains illegal whitespace

报错分析：

/soft/kudu/tserver/wal包含不符合规则的空白

报错处理：

这个报错是因为tserver配置文件中没有指定master地址造成的；

检查/etc/kudu/conf/tserver.gflagfile配置文件，在该配置文件中添加：--tserver_master_addrs=mrj001:7051

注意：mrj001为主机名