不多说，直接上干货！
Storm的版本选取
我这里，是选用apache-storm-1.0.2.tar.gz
apache-storm-0.9.6.tar.gz的集群搭建（3节点）（图文详解）
为什么我用过storm-0.9.6版本，我还要用storm-1.0.2？
storm集群也是由主节点和从节点组成的。
storm版本的变更：
storm0.9.x
storm0.10.x
storm1.x
前面这些版本里面storm的核心源码是由Java+clojule组成的。
storm2.x
后期这个版本就是全部用java重写了。
（阿里在很早的时候就对storm进程了重写，提供了jstorm，后期jstorm也加入到apachestorm
负责使用java对storm进行重写，这就是storm2.x版本的由来。）
注意：
在storm0.9.x的版本中，storm集群只支持一个nimbus节点，主节点是存在问题。
在storm0.10.x以后，storm集群可以支持多个nimbus节点，其中有一个为leader，负责真正运行，其余的为offline。
主节点(控制节点 master)【主节点可以有一个或者多个】
职责：负责分发代码，监控代码的执行。
nimbus：
ui：可以查看集群的信息以及topology的运行情况
logviewer：因为主节点会有多个，有时候也需要查看主节点的日志信息。
从节点(工作节点 worker)【从节点可以有一个或者多个】
职责：负责产生worker进程，执行任务。
supervisor：
logviewer：可以通过webui界面查看topology的运行日志
Storm的本地模式安装
本地模式在一个进程里面模拟一个storm集群的所有功能， 这对开发和测试来说非常方便。以本地模式运行topology跟在集群上运行topology类似。
要创建一个进程内“集群”，使用LocalCluster对象就可以了:
import backtype.storm.LocalCluster;
LocalCluster cluster = new LocalCluster();
然后可以通过LocalCluster对象的submitTopology方法来提交topology, 效果和StormSubmitter对应的方法是一样的。submitTopology方法需要三个参数: topology的名字， topology的配置以及topology对象本身。你可以通过killTopology方法来终止一个topology， 它需要一个topology名字作为参数。
要关闭一个本地集群，简单调用:
cluster.shutdown();
就可以了。
Storm的分布式模式安装（本博文）
官方安装文档
http://storm.apache.org/releases/current/Setting-up-a-Storm-cluster.html
机器情况：在master、slave1、slave2机器的/home/hadoop/app目录下分别下载storm安装包
本博文情况是
master nimbus
slave1 nimbus supervisor
slave2 supervisor
1、apache-storm-1.0.2.tar.gz的下载
http://archive.apache.org/dist/storm/apache-storm-1.0.2/
或者，直接在安装目录下，在线下载
wget http://apache.fayea.com/storm/apache-storm-1.0.2/apache-storm-1.0.2.tar.gz
我这里，选择先下载好，再上传安装的方式。
2、上传压缩包
[hadoop@master app]$ ll
total
drwxrwxr-x hadoop hadoop May : apache-storm-0.9.
drwxrwxr-x hadoop hadoop May : azkaban
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
lrwxrwxrwx hadoop hadoop Apr : es -> elasticsearch-2.4./
lrwxrwxrwx hadoop hadoop Apr : flume -> flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.7.
lrwxrwxrwx. hadoop hadoop Apr : hadoop -> hadoop-2.6.
drwxr-xr-x. hadoop hadoop Apr : hadoop-2.6.
lrwxrwxrwx. hadoop hadoop Apr : hbase -> hbase-0.98.
drwxrwxr-x. hadoop hadoop Apr : hbase-0.98.
lrwxrwxrwx. hadoop hadoop Apr : hive -> hive-1.0.
drwxrwxr-x. hadoop hadoop May : hive-1.0.
lrwxrwxrwx. hadoop hadoop Apr : jdk -> jdk1..0_79
drwxr-xr-x. hadoop hadoop Apr jdk1..0_79
drwxr-xr-x. hadoop hadoop Aug jdk1..0_60
lrwxrwxrwx hadoop hadoop May : kafka -> kafka_2.-0.8.2.2
drwxr-xr-x hadoop hadoop May : kafka_2.-0.8.2.2
lrwxrwxrwx hadoop hadoop Apr : kibana -> kibana-4.6.-linux-x86_64/
drwxrwxr-x hadoop hadoop Nov kibana-4.6.-linux-x86_64
lrwxrwxrwx hadoop hadoop May : snappy -> snappy-1.1.
drwxr-xr-x hadoop hadoop May : snappy-1.1.
lrwxrwxrwx. hadoop hadoop Apr : sqoop -> sqoop-1.4.
drwxr-xr-x. hadoop hadoop May : sqoop-1.4.
lrwxrwxrwx hadoop hadoop May : storm -> apache-storm-0.9./
lrwxrwxrwx. hadoop hadoop Apr : zookeeper -> zookeeper-3.4.
drwxr-xr-x. hadoop hadoop Apr : zookeeper-3.4.
[hadoop@master app]$ rz [hadoop@master app]$ ll
total
drwxrwxr-x hadoop hadoop May : apache-storm-0.9.
-rw-r--r-- hadoop hadoop May : apache-storm-1.0..tar.gz
drwxrwxr-x hadoop hadoop May : azkaban
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
lrwxrwxrwx hadoop hadoop Apr : es -> elasticsearch-2.4./
lrwxrwxrwx hadoop hadoop Apr : flume -> flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.7.
lrwxrwxrwx. hadoop hadoop Apr : hadoop -> hadoop-2.6.
drwxr-xr-x. hadoop hadoop Apr : hadoop-2.6.
lrwxrwxrwx. hadoop hadoop Apr : hbase -> hbase-0.98.
drwxrwxr-x. hadoop hadoop Apr : hbase-0.98.
lrwxrwxrwx. hadoop hadoop Apr : hive -> hive-1.0.
drwxrwxr-x. hadoop hadoop May : hive-1.0.
lrwxrwxrwx. hadoop hadoop Apr : jdk -> jdk1..0_79
drwxr-xr-x. hadoop hadoop Apr jdk1..0_79
drwxr-xr-x. hadoop hadoop Aug jdk1..0_60
lrwxrwxrwx hadoop hadoop May : kafka -> kafka_2.-0.8.2.2
drwxr-xr-x hadoop hadoop May : kafka_2.-0.8.2.2
lrwxrwxrwx hadoop hadoop Apr : kibana -> kibana-4.6.-linux-x86_64/
drwxrwxr-x hadoop hadoop Nov kibana-4.6.-linux-x86_64
lrwxrwxrwx hadoop hadoop May : snappy -> snappy-1.1.
drwxr-xr-x hadoop hadoop May : snappy-1.1.
lrwxrwxrwx. hadoop hadoop Apr : sqoop -> sqoop-1.4.
drwxr-xr-x. hadoop hadoop May : sqoop-1.4.
lrwxrwxrwx hadoop hadoop May : storm -> apache-storm-0.9./
lrwxrwxrwx. hadoop hadoop Apr : zookeeper -> zookeeper-3.4.
drwxr-xr-x. hadoop hadoop Apr : zookeeper-3.4.
[hadoop@master app]$
slave1和slave2机器同样。不多赘述。
3、解压压缩包，并赋予用户组和用户权限
[hadoop@master app]$ ll
total
drwxrwxr-x hadoop hadoop May : apache-storm-0.9.
-rw-r--r-- hadoop hadoop May : apache-storm-1.0..tar.gz
drwxrwxr-x hadoop hadoop May : azkaban
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
lrwxrwxrwx hadoop hadoop Apr : es -> elasticsearch-2.4./
lrwxrwxrwx hadoop hadoop Apr : flume -> flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.7.
lrwxrwxrwx. hadoop hadoop Apr : hadoop -> hadoop-2.6.
drwxr-xr-x. hadoop hadoop Apr : hadoop-2.6.
lrwxrwxrwx. hadoop hadoop Apr : hbase -> hbase-0.98.
drwxrwxr-x. hadoop hadoop Apr : hbase-0.98.
lrwxrwxrwx. hadoop hadoop Apr : hive -> hive-1.0.
drwxrwxr-x. hadoop hadoop May : hive-1.0.
lrwxrwxrwx. hadoop hadoop Apr : jdk -> jdk1..0_79
drwxr-xr-x. hadoop hadoop Apr jdk1..0_79
drwxr-xr-x. hadoop hadoop Aug jdk1..0_60
lrwxrwxrwx hadoop hadoop May : kafka -> kafka_2.-0.8.2.2
drwxr-xr-x hadoop hadoop May : kafka_2.-0.8.2.2
lrwxrwxrwx hadoop hadoop Apr : kibana -> kibana-4.6.-linux-x86_64/
drwxrwxr-x hadoop hadoop Nov kibana-4.6.-linux-x86_64
lrwxrwxrwx hadoop hadoop May : snappy -> snappy-1.1.
drwxr-xr-x hadoop hadoop May : snappy-1.1.
lrwxrwxrwx. hadoop hadoop Apr : sqoop -> sqoop-1.4.
drwxr-xr-x. hadoop hadoop May : sqoop-1.4.
lrwxrwxrwx hadoop hadoop May : storm -> apache-storm-0.9./
lrwxrwxrwx. hadoop hadoop Apr : zookeeper -> zookeeper-3.4.
drwxr-xr-x. hadoop hadoop Apr : zookeeper-3.4.
[hadoop@master app]$ tar -zxvf apache-storm-1.0..tar.gz
slave1和slave2机器同样。不多赘述。
4、删除压缩包，为了更好容下多版本，创建软链接
大数据各子项目的环境搭建之建立与删除软连接（博主推荐）
[hadoop@master app]$ ll
total
drwxrwxr-x hadoop hadoop May : apache-storm-0.9.
drwxrwxr-x hadoop hadoop May : apache-storm-1.0.
drwxrwxr-x hadoop hadoop May : azkaban
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
lrwxrwxrwx hadoop hadoop Apr : es -> elasticsearch-2.4./
lrwxrwxrwx hadoop hadoop Apr : flume -> flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.7.
lrwxrwxrwx. hadoop hadoop Apr : hadoop -> hadoop-2.6.
drwxr-xr-x. hadoop hadoop Apr : hadoop-2.6.
lrwxrwxrwx. hadoop hadoop Apr : hbase -> hbase-0.98.
drwxrwxr-x. hadoop hadoop Apr : hbase-0.98.
lrwxrwxrwx. hadoop hadoop Apr : hive -> hive-1.0.
drwxrwxr-x. hadoop hadoop May : hive-1.0.
lrwxrwxrwx. hadoop hadoop Apr : jdk -> jdk1..0_79
drwxr-xr-x. hadoop hadoop Apr jdk1..0_79
drwxr-xr-x. hadoop hadoop Aug jdk1..0_60
lrwxrwxrwx hadoop hadoop May : kafka -> kafka_2.-0.8.2.2
drwxr-xr-x hadoop hadoop May : kafka_2.-0.8.2.2
lrwxrwxrwx hadoop hadoop Apr : kibana -> kibana-4.6.-linux-x86_64/
drwxrwxr-x hadoop hadoop Nov kibana-4.6.-linux-x86_64
lrwxrwxrwx hadoop hadoop May : snappy -> snappy-1.1.
drwxr-xr-x hadoop hadoop May : snappy-1.1.
lrwxrwxrwx. hadoop hadoop Apr : sqoop -> sqoop-1.4.
drwxr-xr-x. hadoop hadoop May : sqoop-1.4.
lrwxrwxrwx. hadoop hadoop Apr : zookeeper -> zookeeper-3.4.
drwxr-xr-x. hadoop hadoop Apr : zookeeper-3.4.
[hadoop@master app]$ ln -s apache-storm-1.0./ storm
[hadoop@master app]$ ll
total
drwxrwxr-x hadoop hadoop May : apache-storm-0.9.
drwxrwxr-x hadoop hadoop May : apache-storm-1.0.
drwxrwxr-x hadoop hadoop May : azkaban
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
drwxrwxr-x hadoop hadoop Apr : elasticsearch-2.4.
lrwxrwxrwx hadoop hadoop Apr : es -> elasticsearch-2.4./
lrwxrwxrwx hadoop hadoop Apr : flume -> flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.6.
drwxrwxr-x hadoop hadoop Apr : flume-1.7.
lrwxrwxrwx. hadoop hadoop Apr : hadoop -> hadoop-2.6.
drwxr-xr-x. hadoop hadoop Apr : hadoop-2.6.
lrwxrwxrwx. hadoop hadoop Apr : hbase -> hbase-0.98.
drwxrwxr-x. hadoop hadoop Apr : hbase-0.98.
lrwxrwxrwx. hadoop hadoop Apr : hive -> hive-1.0.
drwxrwxr-x. hadoop hadoop May : hive-1.0.
lrwxrwxrwx. hadoop hadoop Apr : jdk -> jdk1..0_79
drwxr-xr-x. hadoop hadoop Apr jdk1..0_79
drwxr-xr-x. hadoop hadoop Aug jdk1..0_60
lrwxrwxrwx hadoop hadoop May : kafka -> kafka_2.-0.8.2.2
drwxr-xr-x hadoop hadoop May : kafka_2.-0.8.2.2
lrwxrwxrwx hadoop hadoop Apr : kibana -> kibana-4.6.-linux-x86_64/
drwxrwxr-x hadoop hadoop Nov kibana-4.6.-linux-x86_64
lrwxrwxrwx hadoop hadoop May : snappy -> snappy-1.1.
drwxr-xr-x hadoop hadoop May : snappy-1.1.
lrwxrwxrwx. hadoop hadoop Apr : sqoop -> sqoop-1.4.
drwxr-xr-x. hadoop hadoop May : sqoop-1.4.
lrwxrwxrwx hadoop hadoop May : storm -> apache-storm-1.0./
lrwxrwxrwx. hadoop hadoop Apr : zookeeper -> zookeeper-3.4.
drwxr-xr-x. hadoop hadoop Apr : zookeeper-3.4.
[hadoop@master app]$
slave1和slave2机器同样。不多赘述。
5、修改配置环境
[hadoop@master app]$ su root
Password:
[root@master app]# vim /etc/profile
slave1和slave2机器同样。不多赘述
#storm
export STORM_HOME=/home/hadoop/app/storm
export PATH=$PATH:$STORM_HOME/bin
slave1和slave2机器同样。不多赘述
[hadoop@master app]$ su root
Password:
[root@master app]# vim /etc/profile
[root@master app]# source /etc/profile
[root@master app]#
slave1和slave2机器同样。不多赘述
6、下载好Storm集群所需的其他
因为博主我的机器是CentOS6.5，已经自带了
[hadoop@master ~]$ python
Python 2.6.6 (r266:84292, Nov 22 2013, 12:16:22)
[GCC 4.4.7 20120313 (Red Hat 4.4.7-4)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>>
7、配置storm的配置文件
[hadoop@master storm]$ pwd
/home/hadoop/app/storm
[hadoop@master storm]$ ll
total
drwxrwxr-x hadoop hadoop May : bin
-rw-r--r-- hadoop hadoop Jul CHANGELOG.md
drwxrwxr-x hadoop hadoop May : conf
drwxrwxr-x hadoop hadoop Jul examples
drwxrwxr-x hadoop hadoop May : external
drwxrwxr-x hadoop hadoop Jul extlib
drwxrwxr-x hadoop hadoop Jul extlib-daemon
drwxrwxr-x hadoop hadoop May : lib
-rw-r--r-- hadoop hadoop Jul LICENSE
drwxrwxr-x hadoop hadoop May : log4j2
-rw-r--r-- hadoop hadoop Jul NOTICE
drwxrwxr-x hadoop hadoop May : public
-rw-r--r-- hadoop hadoop Jul README.markdown
-rw-r--r-- hadoop hadoop Jul RELEASE
-rw-r--r-- hadoop hadoop Jul SECURITY.md
[hadoop@master storm]$
进入storm配置目录下，修改配置文件storm.yaml
[hadoop@master conf]$ pwd
/home/hadoop/app/storm/conf
[hadoop@master conf]$ ll
total
-rw-r--r-- hadoop hadoop Jul storm_env.ini
-rwxr-xr-x hadoop hadoop Jul storm-env.sh
-rw-r--r-- hadoop hadoop Jul storm.yaml
[hadoop@master conf]$ vim storm.yaml
slave1和slave2机器同样。不多赘述
这里，教给大家一个非常好的技巧。
大数据搭建各个子项目时配置文件技巧（适合CentOS和Ubuntu系统）（博主推荐）
注意第一列需要一个空格
注意第一列需要一个空格（HA）
storm.zookeeper.servers:
- "master"
- "slave1"
- "slave2" nimbus.seeds: ["master", "slave1"]
ui.port: storm.local.dir: "/home/hadoop/data/storm" supervisor.slots.ports:
-
-
-
-
注意：我的这里ui.port选定为9999，是自定义，为了解决Storm 和spark默认的 8080 端口冲突！
slave1和slave2机器同样。不多赘述。
注意第一列需要一个空格（非HA）
storm.zookeeper.servers:
- "master"
- "slave1"
- "slave2" nimbus.seeds: ["master"]
ui.port: 9999 storm.local.dir: "/home/hadoop/data/storm" supervisor.slots.ports:
- 6700
- 6701
- 6702
- 6703
注意：我的这里ui.port选定为9999，是自定义，为了解决Storm 和spark默认的 8080 端口冲突！
slave1和slave2机器同样。不多赘述。
8、新建storm数据存储的路径目录
[hadoop@master conf]$ mkdir -p /home/hadoop/data/storm
slave1和slave2机器同样。不多赘述
9、启动storm集群（HA）
本博文情况是
master（主） nimbus
slave1（主）（从） nimbus supervisor
slave2（从） supervisor
1、先在master上启动
nohup bin/storm nimbus >/dev/null >& &
[hadoop@master storm]$ jps
QuorumPeerMain
Jps
AzkabanWebServer
ResourceManager
AzkabanExecutorServer
NameNode
SecondaryNameNode
[hadoop@master storm]$ nohup bin/storm nimbus >/dev/null >& &
[]
[hadoop@master storm]$ jps
QuorumPeerMain
Jps
config_value
AzkabanWebServer
ResourceManager
AzkabanExecutorServer
NameNode
SecondaryNameNode
[hadoop@master storm]$
2、再在slave1上启动
nohup bin/storm nimbus >/dev/null >& &
[hadoop@slave1 storm]$ jps
NodeManager
DataNode
Jps
QuorumPeerMain
[hadoop@slave1 storm]$ nohup bin/storm nimbus >/dev/null >& &
[]
[hadoop@slave1 storm]$ jps
2421 NodeManager
5244 Jps
2342 DataNode
5135 nimbus
5234 config_value
2274 QuorumPeerMain
3、先在slave1和slave2上启动
nohup bin/storm supervisor >/dev/null >& &
[hadoop@slave2 storm]$ jps
Jps
supervisor
NodeManager
DataNode
QuorumPeerMain
[hadoop@slave2 storm]$ nohup bin/storm supervisor >/dev/null >& &
[]
[hadoop@slave2 storm]$ jps
Jps
supervisor
NodeManager
DataNode
QuorumPeerMain
[hadoop@slave2 storm]$
4、在master上启动
nohup bin/storm ui>/dev/null >& &
[hadoop@master storm]$ jps
config_value
QuorumPeerMain
supervisor
AzkabanWebServer
ResourceManager
Jps
AzkabanExecutorServer
config_value
core
NameNode
SecondaryNameNode
[hadoop@master storm]$ nohup bin/storm ui>/dev/null >& &
[]
[hadoop@master storm]$ jps
QuorumPeerMain
supervisor
Jps
AzkabanWebServer
ResourceManager
AzkabanExecutorServer
core
NameNode
config_value
SecondaryNameNode
config_value
[hadoop@master storm]$
5、在master、slave1和slave2上启动
nohup bin/storm logviwer >/dev/null >& &
9、启动storm集群（非HA）
本博文情况是
master（主） nimbus
slave1（主）（从） supervisor
slave2（从） supervisor
1、先在master上启动
nohup bin/storm nimbus >/dev/null 2>&1 &
[hadoop@master storm]$ jps
2374 QuorumPeerMain
7862 Jps
3343 AzkabanWebServer
2813 ResourceManager
3401 AzkabanExecutorServer
2515 NameNode
2671 SecondaryNameNode
[hadoop@master storm]$ nohup bin/storm nimbus >/dev/null 2>&1 &
[1] 7876
[hadoop@master storm]$ jps
2374 QuorumPeerMain
7905 Jps
7910 config_value
3343 AzkabanWebServer
2813 ResourceManager
3401 AzkabanExecutorServer
2515 NameNode
2671 SecondaryNameNode
9743 nimbus
[hadoop@master storm]$
2、先在slave1和slave2上启动
nohup bin/storm supervisor >/dev/null 2>&1 &
[hadoop@slave2 storm]$ jps
4868 Jps
4089 supervisor
2365 NodeManager
2291 DataNode
2229 QuorumPeerMain
[hadoop@slave2 storm]$ nohup bin/storm supervisor >/dev/null 2>&1 &
[1] 4903
[hadoop@slave2 storm]$ jps
4918 Jps
4089 supervisor
2365 NodeManager
2291 DataNode
2229 QuorumPeerMain
[hadoop@slave2 storm]$
3、在master上启动
nohup bin/storm ui>/dev/null 2>&1 &
[hadoop@master storm]$ jps
8550 config_value
2374 QuorumPeerMain
8113 supervisor
3343 AzkabanWebServer
2813 ResourceManager
8560 Jps
3401 AzkabanExecutorServer
8524 config_value
8372 core
2515 NameNode
2671 SecondaryNameNode
[hadoop@master storm]$ nohup bin/storm ui>/dev/null 2>&1 &
[7] 8582
[hadoop@master storm]$ jps
2374 QuorumPeerMain
8113 supervisor
8623 Jps
3343 AzkabanWebServer
2813 ResourceManager
3401 AzkabanExecutorServer
8372 core
2515 NameNode
8597 config_value
2671 SecondaryNameNode
8613 config_value
[hadoop@master storm]$
4、在master、slave1和slave2上启动
nohup bin/storm logviwer >/dev/null 2>&1 &
成功！