[hadoop] 集群启动和内存调优

1、启动Hadoop集群

#首先查看下zoo.cfg里面配置的server.id和集群主机的id是否一致
#如果不一致会造成yarn控制不了从节点的启动

cat /home/hadoop/zookeeper/conf/zoo.cfg 
cat /home/hadoop/zookeeper/data/myid

#分别停止和启动各节点的zookeeper 
#保证每台节点的状态为follow或者leader

/home/hadoop/zookeeper/bin/zkServer.sh stop
/home/hadoop/zookeeper/bin/zkServer.sh start
/home/hadoop/zookeeper/bin/zkServer.sh status

#启动hadoop
#成功启动后可访问 
#http://master:50070/dfshealth.html#tab-overview
#http://master:8088/cluster/nodes

/home/hadoop/hadoop/sbin/start-all.sh 
/home/hadoop/hadoop/sbin/stop-all.sh 

#启动hbase
#成功启动后可访问 
#http://master:60010/master-status?filter=all#baseStats
#Region Servers 会列出hbase的节点列表

/home/hadoop/hbase/bin/start-hbase.sh 
/home/hadoop/hbase/bin/stop-hbase.sh 

2、提示信息

#hadoop

[hadoop@master sbin]$ ./start-all.sh 
This script is Deprecated. Instead use start-dfs.sh and start-yarn.sh
Starting namenodes on [master]
master: starting namenode, logging to /home/hadoop/hadoop/logs/hadoop-hadoop-namenode-master.out
localhost: starting datanode, logging to /home/hadoop/hadoop/logs/hadoop-hadoop-datanode-master.out
node1: starting datanode, logging to /home/hadoop/hadoop/logs/hadoop-hadoop-datanode-node1.out
node2: starting datanode, logging to /home/hadoop/hadoop/logs/hadoop-hadoop-datanode-node2.out
Starting secondary namenodes [master]
master: starting secondarynamenode, logging to /home/hadoop/hadoop/logs/hadoop-hadoop-secondarynamenode-master.out
starting yarn daemons
starting resourcemanager, logging to /home/hadoop/hadoop/logs/yarn-hadoop-resourcemanager-master.out
localhost: starting nodemanager, logging to /home/hadoop/hadoop/logs/yarn-hadoop-nodemanager-master.out
node1: starting nodemanager, logging to /home/hadoop/hadoop/logs/yarn-hadoop-nodemanager-node1.out
node2: starting nodemanager, logging to /home/hadoop/hadoop/logs/yarn-hadoop-nodemanager-node2.out


#hbase

[hadoop@master sbin]$ start-hbase.sh 
master: starting zookeeper, logging to /home/hadoop/hbase/bin/../logs/hbase-hadoop-zookeeper-master.out
node1: starting zookeeper, logging to /home/hadoop/hbase/bin/../logs/hbase-hadoop-zookeeper-node1.out
node2: starting zookeeper, logging to /home/hadoop/hbase/bin/../logs/hbase-hadoop-zookeeper-node2.out
starting master, logging to /home/hadoop/hbase/logs/hbase-hadoop-master-master.out
node2: starting regionserver, logging to /home/hadoop/hbase/bin/../logs/hbase-hadoop-regionserver-node2.out
node1: starting regionserver, logging to /home/hadoop/hbase/bin/../logs/hbase-hadoop-regionserver-node1.out

3、守护进程

#主节点
[hadoop@master sbin]$ jps
11178 Jps
10212 ResourceManager（资源管理者）
9691 NameNode（主节点）
10028 SecondaryNameNode（备用主节点）
11032 HMaster（Hbase管理者）
10332 NodeManager（节点管理者）
6923 QuorumPeerMain（zookeeper守护进程）
9809 DataNode（主节点也作为数据节点）

#从节点
[hadoop@node2 conf]$ jps
6782 Jps
6640 HRegionServer（Hbase从节点）
6225 NodeManager（节点管理）
6532 HQuorumPeer（zookeeper守护进程）
6103 DataNode（数据节点）

4、心得总结

#启动顺序

(1)先用jps查看是否还有没有停止的hadoop服务，有就先全部停止
(2)先启动 zookeeper：~/zookeeper/bin/zkServer.sh start
不管主从每一个节点的zk都要单独启动 启动后
用jps查看 有没有 QuorumPeerMain 进程 
或者用 ~/zookeeper/bin/zkServer.sh status 查看角色
从属：follower
主：leader

(3)再启动hadoop：~/hadoop/sbin/start-all.sh
主节点 jps查看是否有Namenode等进程
从节点 jps查看是否有Datanode等进程

(4)最后启动hbase：~/hbase/bin/start-hbase.sh
主节点 jps查看是否有HMaster等进程
从节点 jps查看是否有HRegionserver等进程

#其他

　　(1)如果从节点HBASE启动失败
　　1、检查zookeeper/data下myid是否正确
　　2、实在不行单独启动从节点
　　hbase-daemon.sh start regionserver

　　(2)namenode没启动成功先检查是不是端口被占用
　　netstat -atunp | grep 9000
　　ps -aux | grep hadoop

　　(3)如果zk的端口被绑定占用
　　ps -aux | grep zookeeper
　　看看是否有其他的用户比如【root】启动过zk

5、内存调优

【hadoop】512 - 4096

/home/hadoop/hadoop/etc/hadoop/hadoop-env.sh
export HADOOP_PORTMAP_OPTS="-Xmx4096m $HADOOP_PORTMAP_OPTS"
export HADOOP_CLIENT_OPTS="-Xmx4096m $HADOOP_CLIENT_OPTS"

【yarn】2048 - 4096
/home/hadoop/hadoop/etc/hadoop/yarn-env.sh
JAVA_HEAP_MAX=-Xmx4096m


【hbase】

#以前的
export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS $HBASE_JMX_BASE -Xmx1000m -Xms1000m -Xmn750m -XX:+UseParNewGC -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=70"
export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS $HBASE_JMX_BASE -Xmx3096m -Xms3096m -Xmn1220m -XX:+UseParNewGC -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=70"
export HBASE_THRIFT_OPTS="$HBASE_THRIFT_OPTS $HBASE_JMX_BASE -Xms100m -Xmx1024m"
export HBASE_ZOOKEEPER_OPTS="$HBASE_ZOOKEEPER_OPTS $HBASE_JMX_BASE -Xms100m -Xmx1024m"

#现在修改为
export HBASE_MASTER_OPTS="$HBASE_MASTER_OPTS $HBASE_JMX_BASE -Xmx8192m -Xms8192m -Xmn6144m -XX:+UseParNewGC -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=70"
export HBASE_REGIONSERVER_OPTS="$HBASE_REGIONSERVER_OPTS $HBASE_JMX_BASE -Xmx10240m -Xms10240m -Xmn5120m -XX:+UseParNewGC -XX:+UseConcMarkSweepGC -XX:CMSInitiatingOccupancyFraction=70"
export HBASE_THRIFT_OPTS="$HBASE_THRIFT_OPTS $HBASE_JMX_BASE -Xms1024m -Xmx4096m"
export HBASE_ZOOKEEPER_OPTS="$HBASE_ZOOKEEPER_OPTS $HBASE_JMX_BASE -Xms2048m -Xmx4096m"