基于Storm-0.91的分布式流计算集群详细配置

==========================================================================================

一、服务器分布

==========================================================================================

1、相关说明

IP地址	主机名	角色
192.168.1.100	storm_zk1	leader
192.168.1.101	storm_zk2	follower
192.168.1.102	storm_zk3	follower
192.168.1.103	storm_nimbus	Nimbus
192.168.1.104	storm_supervisor1	Supervisor
192.168.1.105	storm_supervisor2	Supervisor

2、hosts信息添加

# vim /etc/hosts

192.168.1.100  storm_zk1 storm_zk1

192.168.1.101  storm_zk2 storm_zk2

192.168.1.102  storm_zk3 storm_zk3

192.168.1.103  storm_nimbus storm_nimbus

192.168.1.104  storm_supervisor1  storm_supervisor1

192.168.1.105  storm_supervisor2  storm_supervisor2

3、总体架构

==========================================================================================

二、ZooKeeper集群部署

==========================================================================================

1、JDK安装

http://download.oracle.com/otn-pub/java/jdk/7u65-b17/jdk-7u65-linux-x64.tar.gz

# tar xvzf jdk-7u65-linux-x64.gz -C /usr/local

# cd /usr/local && ln -s jdk1.7.0_65 jdk

# vim /etc/profile

export JAVA_HOME=/usr/local/jdk

export CLASSPATH=.:$JAVA_HOME/lib/dt.jar:$JAVA_HOME/lib/tools.jar

export PATH=$PATH:$JAVA_HOME/bin

# source /etc/profile

# cd /usr/local/bin && ln -s /usr/local/jdk/bin/java

2、ZooKeeper安装

http://mirror.bit.edu.cn/apache/zookeeper/zookeeper-3.4.5/zookeeper-3.4.5.tar.gz

# tar xvzf zookeeper-3.4.5.tar.gz -C /usr/local

# cd /usr/local && ln -s zookeeper-3.4.5 zookeeper

3、ZooKeeper配置

# mkdir -p /data/zookeeper/{data,logs}

# vim /usr/local/zookeeper/conf/zoo.cfg

tickTime=2000

initLimit=10

syncLimit=5

dataDir=/data/zookeeper/data

dataLogDir=/data/zookeeper/logs

clientPort=2181

maxClientCnxns=3000

server.1=192.168.1.100:2888:3888

server.2=192.168.1.101:2888:3888

server.3=192.168.1.102:2888:3888

autopurge.snapRetainCount=6

autopurge.purgeInterval=2

4、日志级别调整

# vim /usr/local/zookeeper/conf/log4j.properties

zookeeper.root.logger=ERROR,CONSOLE

zookeeper.console.threshold=ERROR

zookeeper.log.threshold=ERROR

5、新增myid文件

在 192.168.1.100 上执行：

# echo 1 > /data/zookeeper/data/myid

在 192.168.1.101 上执行：

# echo 2 > /data/zookeeper/data/myid

在 192.168.1.102 上执行：

# echo 3 > /data/zookeeper/data/myid

6、ZooKeeper集群启动

启动zookeeper集群

# /usr/local/zookeeper/bin/zkServer.sh start

测试zookeeper集群是否建立成功，如无报错表示集群创建成功

# /usr/local/zookeeper/bin/zkCli.sh -server localhost:2181

查看运行状态

# /usr/local/zookeeper/bin/zkServer.sh status

7、服务健康监控

# touch /var/run/check_zookeeper.lock

# echo 'touch /var/run/check_zookeeper.lock' >> /etc/rc.d/rc.local

# vim /usr/local/zookeeper/bin/check_service.sh

#!/bin/sh
PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/bin:/usr/local/sbin
 
SRV_PORT="2181"      ## 端口号
SRV_PROT="TCP"       ## 协议类型
 
## 是否已正确扫描
SCAN_FLAG=0
 
ETH1_ADDR=`/sbin/ifconfig eth1 | awk -F ':' '/inet addr/{print $2}' | sed 's/[a-zA-Z ]//g'`
TMP_SRV_PROT=`echo ${SRV_PROT} | tr '[A-Z]' '[a-z]'`
 
if [[ "${TMP_SRV_PROT}" == "tcp" ]]; then
    PROT_OPT="S"
elif [[ "${TMP_SRV_PROT}" == "udp" ]]; then
    PROT_OPT="U"
else
    echo "未知的协议类型！" && exit 1
fi
 
for ((i=0; i<3; i++)); do
    RETVAL=`/usr/bin/nmap -n -s${PROT_OPT} -p ${SRV_PORT} ${ETH1_ADDR} | grep open`
    [[ -n "${RETVAL}" ]] && SCAN_FLAG=1;break || sleep 10
done
 
if [[ ${SCAN_FLAG} -ne 1 ]]; then
    cd /usr/local/zookeeper/bin; ./zkServer.sh restart >/dev/null 2>&1
fi

需要添加的crontab项：

*/2 * * * * (flock --timeout=0 /var/run/check_zookeeper.lock /usr/local/zookeeper/bin/check_service.sh >/dev/null 2>&1)

==========================================================================================

三、Storm集群部署

==========================================================================================

1、Storm安装

http://mirrors.cnnic.cn/apache/incubator/storm/apache-storm-0.9.1-incubating/apache-storm-0.9.1-incubating.tar.gz

# tar xvzf apache-storm-0.9.1-incubating.tar.gz -C /usr/local

# rm -f apache-storm-0.9.1-incubating.tar.gz

# cd /usr/local

# mv apache-storm-0.9.1-incubating storm-0.9.1

# rm -f storm && ln -s storm-0.9.1 storm

# vim /etc/profile

export STORM_HOME=/usr/local/storm

export PATH=$PATH:$STORM_HOME/bin

# source /etc/profile

2、Storm配置

# mkdir -p /data/storm/{data,logs}

(1)、日志路径修改

# sed -i  's#${storm.home}#/data/storm#g' /usr/local/storm/logback/cluster.xml

(2)、主配置

# vim /usr/local/storm/conf/storm.yaml

drpc.servers:

 - "storm_nimbus"

 - "storm_supervisor1"

 - "storm_supervisor2"

storm.zookeeper.servers:

 - "storm_zk1"

 - "storm_zk2"

 - "storm_zk3"

storm.local.dir:"/data/storm/data"

nimbus.host:"storm_nimbus"

nimbus.thrift.port: 6627

nimbus.thrift.max_buffer_size:1048576

nimbus.childopts:"-Xmx1024m"

nimbus.task.timeout.secs:30

nimbus.supervisor.timeout.secs:60

nimbus.monitor.freq.secs:10

nimbus.cleanup.inbox.freq.secs:600

nimbus.inbox.jar.expiration.secs:3600

nimbus.task.launch.secs:240

nimbus.reassign: true

nimbus.file.copy.expiration.secs:600

nimbus.topology.validator:"backtype.storm.nimbus.DefaultTopologyValidator"

storm.zookeeper.port: 2181

storm.zookeeper.root:"/data/storm/zkinfo"

storm.cluster.mode:"distributed"

storm.local.mode.zmq:false

ui.port: 8080

ui.childopts:"-Xmx768m"

logviewer.port: 8000

logviewer.childopts:"-Xmx128m"

logviewer.appender.name:"A1"

supervisor.slots.ports:

    - 6700

    - 6701

    - 6702

    - 6703

supervisor.childopts:"-Xmx1024m"

supervisor.worker.start.timeout.secs:240

supervisor.worker.timeout.secs:30

supervisor.monitor.frequency.secs:3

supervisor.heartbeat.frequency.secs:5

supervisor.enable: true

worker.childopts:"-Xmx2048m"

topology.max.spout.pending:5000

storm.zookeeper.session.timeout:20

storm.zookeeper.connection.timeout:10

storm.zookeeper.retry.times:10

storm.zookeeper.retry.interval:30

storm.zookeeper.retry.intervalceiling.millis:30000

storm.thrift.transport:"backtype.storm.security.auth.SimpleTransportPlugin"

storm.messaging.transport:"backtype.storm.messaging.netty.Context"

storm.messaging.netty.server_worker_threads:1

storm.messaging.netty.client_worker_threads:1

storm.messaging.netty.buffer_size:5242880

storm.messaging.netty.max_retries:100

storm.messaging.netty.max_wait_ms:1000

storm.messaging.netty.min_wait_ms:100

storm.messaging.netty.transfer.batch.size:262144

storm.messaging.netty.flush.check.interval.ms:10

【注意】

(1)、红色1，表示可以运行的worker数量，需要根据实际的内存容量进行设置；

(2)、红色2，表示将storm的消息传送机制更改为netty。

3、服务启动

(1)、Nimbus节点

# nohup /usr/local/storm/bin/storm nimbus >/dev/null 2>&1 &

# nohup /usr/local/storm/bin/storm ui >/dev/null 2>&1 &

(2)、Supervisor节点

# nohup /usr/local/storm/bin/storm supervisor >/dev/null 2>&1 &

# nohup /usr/local/storm/bin/storm logviewer >/dev/null 2>&1 &

【注意】

如果启动不了，需要查看一下在“/etc/hosts”里，是否设置了主机名

4、健康监控

(1)、Nimbus节点设置

# touch /var/run/check_storm.lock

# echo 'touch /var/run/check_storm.lock' >> /etc/rc.d/rc.local

# vim /data/scripts/monitor_nimbus_for_storm.sh

#!/bin/sh
PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/bin:/usr/local/sbin
 
## 监控页面地址参数
MON_SRV_IPADDR=`/sbin/ifconfig eth1 | awk -F ':' '/inet addr/{print $2}' | sed 's/[a-zA-Z ]//g'`
MON_SRV_PORT="8080"
 
for ((i=0; i<3; i++)); do
    RETVAL=`/usr/bin/nmap -n -sS -p ${MON_SRV_PORT} ${MON_SRV_IPADDR} | grep open`
    [[ -n "${RETVAL}" ]] && SCAN_FLAG=1;break || sleep 10
done
 
if [[ ${SCAN_FLAG} -ne 1 ]]; then
    nohup /usr/local/storm/bin/storm nimbus >/dev/null 2>&1 &
    nohup /usr/local/storm/bin/storm ui >/dev/null 2>&1 &
fi

需要添加的crontab项：

*/2 * * * * (flock --timeout=0 /var/run/check_storm.lock /data/scripts/monitor_nimbus_for_storm.sh>/dev/null 2>&1)

(2)、Supervisor节点设置

# touch /var/run/check_storm.lock

# echo 'touch /var/run/check_storm.lock' >> /etc/rc.d/rc.local

# vim /data/scripts/monitor_supervisor_for_storm.sh

#!/bin/sh
PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/bin:/usr/local/sbin
 
## 监控页面地址参数
MON_SRV_IPADDR="192.168.1.103"
MON_SRV_PORT="8080"
 
RETVAL=$(curl -s http://${MON_SRV_IPADDR}:${MON_SRV_PORT}/ | sed 's//\n/g'| awk -F '<' '/storm_supervisor/{print $1}' | grep `hostname`)
[[-n "${RETVAL}" ]] && exit 0
 
[[ -n `ps aux | grep "storm" | grep -v grep` ]] && kill -9 `ps aux | grep "storm" | grep -v grep | awk '{print $2}'`
[[ -d /data/storm/data ]] && rm -rf /data/storm/data/*
 
nohup /usr/local/storm/bin/storm supervisor >/dev/null 2>&1 &
nohup /usr/local/storm/bin/storm logviewer >/dev/null 2>&1 &

需要添加的crontab项：

*/2 * * * * (flock --timeout=0 /var/run/check_storm.lock /data/scripts/monitor_supervisor_for_storm.sh >/dev/null 2>&1)