druid.io本地集群搭建 / 扩展集群搭建

druid.io 是一个比较重型的数据库查询系统，分为5种节点 。

在此就不对数据库进行介绍了，如果有疑问请参考白皮书：

http://pan.baidu.com/s/1eSFlIJS  

单台机器的集群搭建

 

首先说一下通用的集群搭建，基于 0.9.1.1

下载地址  http://pan.baidu.com/s/1hrJBjlq:

修改 conf/druid/_common 内的 common.runtime.properties，参考如下配置：

1.  
   #
2.  
   # Licensed to Metamarkets Group Inc. (Metamarkets) under one
3.  
   # or more contributor license agreements. See the NOTICE file
4.  
   # distributed with this work for additional information
5.  
   # regarding copyright ownership. Metamarkets licenses this file
6.  
   # to you under the Apache License, Version 2.0 (the
7.  
   # "License"); you may not use this file except in compliance
8.  
   # with the License. You may obtain a copy of the License at
9.  
   #
10.  
    # http://www.apache.org/licenses/LICENSE-2.0
11.  
    #
12.  
    # Unless required by applicable law or agreed to in writing,
13.  
    # software distributed under the License is distributed on an
14.  
    # "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
15.  
    # KIND, either express or implied. See the License for the
16.  
    # specific language governing permissions and limitations
17.  
    # under the License.
18.  
    #
19.  
     
20.  
    #
21.  
    # Extensions
22.  
    #
23.  
     
24.  
    # This is not the full list of Druid extensions, but common ones that people often use. You may need to change this list
25.  
    # based on your particular setup.
26.  
    #druid.extensions.loadList=["druid-kafka-eight", "druid-s3-extensions", "druid-histogram", "druid-datasketches", "druid-lookups-cached-global", "mysql-metadata-storage"]
27.  
    druid.extensions.loadList=["mysql-metadata-storage"]
28.  
     
29.  
    # If you have a different version of Hadoop, place your Hadoop client jar files in your hadoop-dependencies directory
30.  
    # and uncomment the line below to point to your directory.
31.  
    #druid.extensions.hadoopDependenciesDir=/my/dir/hadoop-dependencies
32.  
     
33.  
    #
34.  
    # Logging
35.  
    #
36.  
     
37.  
    # Log all runtime properties on startup. Disable to avoid logging properties on startup:
38.  
    druid.startup.logging.logProperties=true
39.  
     
40.  
    #
41.  
    # Zookeeper
42.  
    #
43.  
     
44.  
    druid.zk.service.host=10.202.4.22:2181
45.  
    druid.zk.paths.base=/druid
46.  
     
47.  
    #
48.  
    # Metadata storage
49.  
    #
50.  
     
51.  
    # For Derby server on your Druid Coordinator (only viable in a cluster with a single Coordinator, no fail-over):
52.  
    #druid.metadata.storage.type=derby
53.  
    #druid.metadata.storage.connector.connectURI=jdbc:derby://metadata.store.ip:1527/var/druid/metadata.db;create=true
54.  
    #druid.metadata.storage.connector.host=metadata.store.ip
55.  
    #druid.metadata.storage.connector.port=1527
56.  
     
57.  
    # For MySQL:
58.  
    druid.metadata.storage.type=mysql
59.  
    druid.metadata.storage.connector.connectURI=jdbc:mysql://10.202.4.22:3306/druid?characterEncoding=UTF-8
60.  
    druid.metadata.storage.connector.user=szh
61.  
    druid.metadata.storage.connector.password=123456
62.  
     
63.  
    # For PostgreSQL (make sure to additionally include the Postgres extension):
64.  
    #druid.metadata.storage.type=postgresql
65.  
    #druid.metadata.storage.connector.connectURI=jdbc:postgresql://db.example.com:5432/druid
66.  
    #druid.metadata.storage.connector.user=...
67.  
    #druid.metadata.storage.connector.password=...
68.  
     
69.  
    #
70.  
    # Deep storage
71.  
    #
72.  
     
73.  
    # For local disk (only viable in a cluster if this is a network mount):
74.  
    druid.storage.type=local
75.  
    druid.storage.storageDirectory=var/druid/segments
76.  
     
77.  
    # For HDFS (make sure to include the HDFS extension and that your Hadoop config files in the cp):
78.  
    #druid.storage.type=hdfs
79.  
    #druid.storage.storageDirectory=/druid/segments
80.  
     
81.  
    # For S3:
82.  
    #druid.storage.type=s3
83.  
    #druid.storage.bucket=your-bucket
84.  
    #druid.storage.baseKey=druid/segments
85.  
    #druid.s3.accessKey=...
86.  
    #druid.s3.secretKey=...
87.  
     
88.  
    #
89.  
    # Indexing service logs
90.  
    #
91.  
     
92.  
    # For local disk (only viable in a cluster if this is a network mount):
93.  
    druid.indexer.logs.type=file
94.  
    druid.indexer.logs.directory=var/druid/indexing-logs
95.  
     
96.  
    # For HDFS (make sure to include the HDFS extension and that your Hadoop config files in the cp):
97.  
    #druid.indexer.logs.type=hdfs
98.  
    #druid.indexer.logs.directory=/druid/indexing-logs
99.  
     
100.  
     # For S3:
101.  
     #druid.indexer.logs.type=s3
102.  
     #druid.indexer.logs.s3Bucket=your-bucket
103.  
     #druid.indexer.logs.s3Prefix=druid/indexing-logs
104.  
      
105.  
     #
106.  
     # Service discovery
107.  
     #
108.  
      
109.  
     druid.selectors.indexing.serviceName=druid/overlord
110.  
     druid.selectors.coordinator.serviceName=druid/coordinator
111.  
      
112.  
     #
113.  
     # Monitoring
114.  
     #
115.  
      
116.  
     druid.monitoring.monitors=["com.metamx.metrics.JvmMonitor"]
117.  
     druid.emitter=logging
118.  
     druid.emitter.logging.logLevel=info

0.9.1.1 默认是不带mysql 扩展的需要自己下载，解压后放置于extensions 下：

参考文章：

http://druid.io/docs/0.9.1.1/operations/including-extensions.html

扩展包的下载地址

http://druid.io/downloads.html

分别修改5个节点的启动配置

/conf/druid/${serviceName}/runtime.properties

broker节点：

/conf/druid/broker/runtime.properties

1.  
   druid.host=10.202.4.22:9102
2.  
   druid.service=druid/broker
3.  
   druid.port=9102
4.  
    
5.  
   # HTTP server threads
6.  
   druid.broker.http.numConnections=5
7.  
   druid.server.http.numThreads=25
8.  
    
9.  
   # Processing threads and buffers
10.  
    druid.processing.buffer.sizeBytes=32768
11.  
    druid.processing.numThreads=2
12.  
     
13.  
    # Query cache
14.  
    druid.broker.cache.useCache=true
15.  
    druid.broker.cache.populateCache=true
16.  
    druid.cache.type=local
17.  
    druid.cache.sizeInBytes=2000000000

coordinator 节点：

/conf/druid/coordinator/runtime.properties

druid.host=10.202.4.22:8082
druid.service=druid/coordinator
druid.port=8082

historical 节点：

/conf/druid/historical/runtime.properties

1.  
   druid.service=druid/historical
2.  
   druid.host=10.202.4.22:9002
3.  
   druid.port=9002
4.  
    
5.  
   # HTTP server threads
6.  
   druid.server.http.numThreads=25
7.  
    
8.  
   # Processing threads and buffers
9.  
   druid.processing.buffer.sizeBytes=6870912
10.  
    druid.processing.numThreads=7
11.  
     
12.  
    druid.historical.cache.useCache=false
13.  
    druid.historical.cache.populateCache=false
14.  
     
15.  
    # Segment storage
16.  
    druid.segmentCache.locations=[{"path":"var/druid/segment-cache","maxSize":13000000}]
17.  
    druid.server.maxSize=13000000

middleManager 节点：

/conf/druid/middleManager/runtime.properties

1.  
   druid.host=10.202.4.22:8091
2.  
   druid.service=druid/middleManager
3.  
   druid.port=8091
4.  
    
5.  
   # Number of tasks per middleManager
6.  
   druid.worker.capacity=3
7.  
    
8.  
   # Task launch parameters
9.  
   druid.indexer.runner.javaOpts=-server -Xmx256m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -Djava.util.logging.manager=org.apache.logging.log4j.jul.LogManager
10.  
    druid.indexer.task.baseTaskDir=var/druid/task
11.  
     
12.  
    # HTTP server threads
13.  
    druid.server.http.numThreads=25
14.  
     
15.  
    # Processing threads and buffers
16.  
    druid.processing.buffer.sizeBytes=65536
17.  
    druid.processing.numThreads=2
18.  
     
19.  
    # Hadoop indexing
20.  
    druid.indexer.task.hadoopWorkingPath=var/druid/hadoop-tmp
21.  
    druid.indexer.task.defaultHadoopCoordinates=["org.apache.hadoop:hadoop-client:2.3.0"]

overlord 节点：

/conf/druid/overlord/runtime.properties

1.  
   druid.service=druid/overlord
2.  
   druid.host=10.202.4.22:9100
3.  
   druid.port=9100
4.  
    
5.  
   #druid.indexer.queue.startDelay=PT30S
6.  
    
7.  
   druid.indexer.runner.type=remote
8.  
   druid.indexer.storage.type=metadata

启动集群启动  可以利用我写的脚本：

1.  
   #java `cat conf/druid/broker/jvm.config | xargs` -cp conf/druid/_common:conf/druid/broker:lib/* io.druid.cli.Main server broker
2.  
    
3.  
   function help(){
4.  
   echo "参数列表"
5.  
   echo " 参数1 参数2"
6.  
   echo " serviceName [-f]"
7.  
   echo "参数1：serviceName: 启动服务的名字"
8.  
   echo "serviceName可选项:"
9.  
   echo "1: broker"
10.  
    echo "2: coordinator"
11.  
    echo "3: historical"
12.  
    echo "4: middleManager"
13.  
    echo "5: overlord"
14.  
    echo "参数2：[-f]: 是否前台启动"
15.  
    echo "-f:前台启动，(不加)默认后台启动"
16.  
    }
17.  
     
18.  
    function startService(){
19.  
    # echo $0
20.  
    # echo $1
21.  
    # echo $2
22.  
    echo $service
23.  
    if [[ $2 == "-f" ]]; then
24.  
    echo "前台启动"
25.  
    java -Xmx256m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -classpath conf/druid/_common:conf/druid/$service:lib/* io.druid.cli.Main server $service
26.  
    else
27.  
    echo "后台启动"
28.  
    nohup java -Xmx256m -Duser.timezone=UTC -Dfile.encoding=UTF-8 -classpath conf/druid/_common:conf/druid/$service:lib/* io.druid.cli.Main server $service &
29.  
    fi;
30.  
    }
31.  
     
32.  
    function tips(){
33.  
    red=`tput setaf 1`
34.  
    reset=`tput sgr0`
35.  
    echo "${red}Not correct arguments${reset}"
36.  
    echo "please use --help or -h for help"
37.  
    }
38.  
     
39.  
    if [[ $1 == "--help" || $1 == "-h" ]]; then
40.  
    help
41.  
    exit
42.  
    fi
43.  
     
44.  
    service=$1
45.  
     
46.  
    case $service in
47.  
    "broker")
48.  
    ;;
49.  
    "coordinator")
50.  
    ;;
51.  
    "historical")
52.  
    ;;
53.  
    "middleManager")
54.  
    ;;
55.  
    "overlord")
56.  
    ;;
57.  
    *)
58.  
    tips
59.  
    exit
60.  
    esac
61.  
     
62.  
    if [[ $2 == "-f" || $2 == "" ]]; then
63.  
    startService $1 $2;
64.  
    else
65.  
    tips
66.  
    exit
67.  
    fi

将上述脚本放到 druid的根目录即可：

启动效果如图：

多台机器的集群搭建：

上面是单台机器的集群搭建， 扩展到多台

只需要修改 conf/druid/(broker | coordinator | historical | middleManager | overlord) 的 runtime.properties 中的 

druid.host=10.202.4.22:9102 改成其他机器的IP地址即可

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

最后共享下我成功搭建的本地集群 (内含个人写的集群启动脚本)：

http://pan.baidu.com/s/1bJjFzg

基于版本 0.9.1.1,  元数据存储用的mysql （扩展包已经下载好了）

下载并解压 我搭建的druid.tgz 的druid本地集群后：

1.修改各个配置文件

conf/druid/broker/runtime.properties 

conf/druid/coordinator/runtime.properties 

conf/druid/historical/runtime.properties 

conf/druid/middleManager/runtime.properties 

conf/druid/overlord/runtime.properties 

druid.host=10.202.4.22:9102 改成自己的IP地址

2.修改 conf/druid/_common/common.runtime.properties  中的mysql端口地址, 用户名,密码, zookeeper地址等 替换为自己的地址。

 

Tips:mysql 没有创建库druid 要自己先创建（安装mysql数据库，并创建druid数据库和druid用户）

3.用根目录下的 cluster_start_service.sh 启动5个节点的服务

   这时进入mysql数据库 会发现druid自动创建了几张表

4.进行测试

（1）数据导入

（2）数据查询

切换到 test 目录，下面有两个文件夹：

（1）数据导入：

切换到 test_load 的目录下

下面的 submit_csv_task.sh , submit_json_task.sh 分别是提交 csv 数据 与 json 数据的测试脚本

这里需要修改 env.sh

设置 overlord_ip 为自己的 overlord_ip 的 地址端口

这里试下执行 submit_json_task.sh 脚本，去页面上查看 导入任务执行的状态 ，如图所示：

输入的地址为 overlord 节点的地址+端口：

等待任务成功执行完成：

切换到查询测试目录，执行查询脚本：