CDH5-CDH Management

CDH配置文件深度解析 
1.cmf
2.
/var/run/cloudera-scm-agent/process/19-zookeeper-server/zoo.cfg  : service


/etc/zookeeper/conf/ : client


/var/run/cloudera-scm-agent/process/32-zookeeper-server/zoo.cfg 






CM的常用命令 
service  cloudera-scm-server status|start|stop|restart
service  cloudera-scm-agent status|start|stop|restart

jps命令
1.正常表现
[root@i-ucebspgm ~]# jps
8256 -- process information unavailable
10215 Jps
9224 -- process information unavailable
8042 -- process information unavailable
8218 -- process information unavailable
[root@i-ucebspgm ~]# su - hdfs
[hdfs@i-ucebspgm ~]$ jps 
10339 Jps
8042 DataNode
[hdfs@i-ucebspgm ~]$ 


        2.假象:
1101 -- process information unavailable
ps -ef|grep 1101 找不到该进程，那么就是残留文件
        cd /tmp/hsperfdata_hdfs/
ll 找到 1101文件


3.集群 机器上所有服务重启前，将/tmp/hsperfdata_hdfs/残留文件
        /tmp/hsperfdata_yarn
/tmp/hsperfdata_zookeeper

4.其他命令
jps -v
jps -l

CDH集群正确启动和停止顺序  
启动:
1.su - mysqladmin
service mysql start
2.service cloudera-scm-server start
3.service cloudera-scm-agent start (每台)
4.web: cms
5.web: cluster1

关闭:
1.web: cluster1
2.web: cms
3.service cloudera-scm-agent stop (每台)
4.service cloudera-scm-server stop
5.su - mysqladmin
service mysql stop

注意点: cms 和 cluster1 无论哪个关闭、启动等等操作，会造成web界面的图表的 不绘制
1. cms 和 cluster1 都要启动 ，稍微等待 ，正在恢复中(30s)
2. 先cluster1关闭,再重启cms服务 

CDH Tsquery Language 

CDH常规管理(监控/预警/配置/资源/日志) 
1.监控
1.1 自带的图表
1.1 CDH api

日志:
/var/log/hadoop/  10份  200M
/var/log/zookeeper/

/var/log/cloudera-scm-server/
/var/log/cloudera-scm-agent/

/var/log/hadoop-hdfs/ 

若泽大数据交流群：671914634