Cloudera Manager + CDH安装

鉴于apache版本的Hadoop监控不完善，我们生产使用了CDH，同时使用Cloudera Manager监控，Cloudera Manager不但能监控集群还能监控机器，免费版也够用真心值得推荐。

离线安装Cloudera Manager和CDH

楼主自测很靠谱：最靠谱的CDH5安装指南

备注：通过修改每个主机/etc/cloudera-scm-agent/config.ini的server_host为安装Cloudera Manager的机器的主机名来直接识别需要管理的主机，这样你可以避免为CDH集群按照指定主机。

Cloudera Manager配置MySQL

1. 安装MySQL（忽略）

2. 创建cm用户赋予权限(为了避免问题赋予最大权限)
   create user 'cm' identified by 'cm';
   grant all on . to 'cm'@'master01' identified by 'cm' with grant option;
   flush privileges;

3. 初始化Cloudera Manager的数据库
   /usr/share/cmf/schema/scm_prepare_database.sh -h master01 -ucm -pcm mysql --scm-host master01 scm cm cm

   执行完命令后查看/etc/cloudera-scm-server/db.properties如下

   com.cloudera.cmf.db.type=mysql
   com.cloudera.cmf.db.host=master01
   com.cloudera.cmf.db.name=scm
   com.cloudera.cmf.db.user=cm
   com.cloudera.cmf.db.password=cm
   

4. 关闭内置数据库PostgreSQL（一定要停止）
   /etc/init.d/cloudera-scm-server-db stop

5. 重启Cloudera Manager
   /etc/init.d/cloudera-scm-server stop
   /etc/init.d/cloudera-scm-server start

配置时钟同步（使用NTP）

假设现在有3个节点(hadoop01、hadoop02、hadoop03)：
hadoop01节点作为ntp服务器与外界对时中心同步时间，随后对hadoop02、hadoop03节点提供时间同步服务。
hadoop02、hadoop03节点以hadoop01节点为基础同步时间。

1. 每个节点安装ntp
   yum install ntp

2. 配置hadoop01的/etc/ntp.conf

    server 0.cn.pool.ntp.org
    server 0.asia.pool.ntp.org
    server 3.asia.pool.ntp.org
   
    # allow update time by the upper server
    # 允许上层时间服务器主动修改本机时间
    restrict 0.cn.pool.ntp.org nomodify notrap noquery
    restrict 0.asia.pool.ntp.org nomodify notrap noquery
    restrict 3.asia.pool.ntp.org nomodify notrap noquery
   
    # Undisciplined Local Clock. This is a fake driver intended for backup
    # and when no outside source of synchronized time is available.
    # 外部时间服务器不可用时，以本地时间作为时间服务
    server  127.127.1.0     # local clock
    fudge   127.127.1.0 stratum 10
   

3. 配置hadoop02、hdaoop03的/etc/ntp.conf

   server 0.cn.pool.ntp.org
   server 0.asia.pool.ntp.org
   server 3.asia.pool.ntp.org
   
   # allow update time by the upper server
   # 允许上层时间服务器主动修改本机时间
   restrict 0.cn.pool.ntp.org nomodify notrap noquery
   restrict 0.asia.pool.ntp.org nomodify notrap noquery
   restrict 3.asia.pool.ntp.org nomodify notrap noquery
   
   # Undisciplined Local Clock. This is a fake driver intended for backup
   # and when no outside source of synchronized time is available.
   # 外部时间服务器不可用时，以本地时间作为时间服务
   server  127.127.1.0     # local clock
   fudge   127.127.1.0 stratum 10
   server hadoop01  prefer
   

4. 启动 ntp
   hadoop01 启动之前先手动同步下ntpdate 0.cn.pool.ntp.org
   hadoop02-03 启动之前先手动同步下ntpdate hadoop01

   service ntpd start

5. 设置开机启动(每个节点)
   chkconfig ntpd on

安装CDH中遇到的问题

Cloudera Manager安装无法登陆

Cloudera Manager安装后访问问题
备注：问题没有彻底解决，通过远程访问阿里windows系统

Hive问题

通过CM启动hive报错

Hive出现的问题作者的文章中都有，我居然其中的问题全出现了。
mysql-connector-java.jar 位置在：/usr/share/java/mysql-connector-java.jar
且名字一定要为：mysql-connector-java.jar

配置HA

官网配置

Hive

Hive HA

完全卸载Cloudera Manager

卸载Cloudera Manager

配置ngxin访问CM

问题

Cloudera_Server启动: com.cloudera.server.web.cmf.csrf.CsrfRefererInterceptor: Rejecting request originating from

解决

grep安装目录的csrf，直接注释掉然后重启cloudera-manager-server

[[email protected] cloudera-scm-server]# cd /usr/share/cmf/
[[email protected] cmf]# grep -i -r  csrf  ./
Binary file ./cloudera-navigator-server/libs/cdh5/hadoop-yarn-server-nodemanager-2.6.0-cdh5.5.0.jar matches
Binary file ./common_jars/hadoop-yarn-server-nodemanager-2.6.0-cdh5.5.0.jar matches
Binary file ./common_jars/server-5.6.0.jar matches
Binary file ./common_jars/hadoop-yarn-server-resourcemanager-2.5.0-cdh5.3.2.jar matches
./webapp/WEB-INF/spring/mvc-config.xml:                
Binary file ./lib/cdh5-java6/hadoop-yarn-server-resourcemanager-2.5.0-cdh5.3.2.jar matches
Binary file ./lib/server-5.6.0.jar matches
Binary file ./lib/cdh5/hadoop-yarn-server-nodemanager-2.6.0-cdh5.5.0.jar matches
[[email protected] cmf]# vi ./webapp/WEB-INF/spring/mvc-config.xml
注释掉这个bean，然后重启server，在访问nginx，整个世界清净了