ClickHouse是“战斗民族”俄罗斯搜索巨头Yandex公司开源的一个极具“战斗力”的实时数据分析数据库，是面向OLAP的分布式列式DBMS，圈内人戏称为“喀秋莎数据库”。ClickHouse简称"CH"，但在中文社区里大家更偏爱"CK"，反馈是因为有"AK"的感觉！与Hadoop、Spark这些巨无霸组件相比，ClickHouse很轻量级，其特点：列式存储数据库，数据压缩；关系型、支持SQL；分布式并行计算，把单机性能压榨到极限；高可用；数据量级在PB级别。——这段话引用自一篇公众号文章《百分点大数据技术团队：ClickHouse国家级项目最佳实践》，文章地址：https://mp.weixin.qq.com/s/9S6DhDhv7lFbMaFScf137Q。

下面的内容我就把ClickHouse简称为"CK"，官方的"C"和"H"字母是大写，对于有中度强迫症的我来说，简称为"CH"更符合我的胃口，但是为了方便大家区分和易读，所以就委曲求全简称为CK吧，ZooKeeper就简称为ZK。

约定好简称，下面就开始今天的内容吧。

01

02

mkdir zookeepercd zookeepervim ip_list[zookeeper]172.xx.xx.x HOSTNAME=sz-pg-adt-apps-002.xxxxxx.com ID=1172.xx.xx.x HOSTNAME=sz-pg-adt-apps-003.xxxxxx.com ID=2172.xx.xx.x HOSTNAME=sz-pg-adt-apps-004.xxxxxx.com ID=3[clickhouse]172.xx.xx.x

vim hosts.j2127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4::1 localhost localhost.localdomain localhost6 localhost6.localdomain6{% for item in groups['zookeeper'] %}{{ item }} {{ hostvars[item]['HOSTNAME'] }}{% endfor %}

vim zoo.cfg.j2tickTime=2000initLimit=10syncLimit=5dataDir= /usr/local/zookeeper/datadataLogDir= /usr/local/zookeeper/logclientPort=2182autopurge.purgeInterval=0globalOutstandingLimit=200{% for item in groups['zookeeper'] %}server.{{ hostvars[item]['ID'] }}={{ item }}:2999:3999{% endfor %}

vim deploy.zookeeper.yaml---- hosts: zookeeper  remote_user: root  gather_facts: true  # 定义ZK相关变量  vars:    zookeeper_version: apache-zookeeper-3.7.0-bin.tar.gz    zookeeper_dir: apache-zookeeper-3.7.0-bin  # 开始定义任务  tasks:  # 1、修改所有节点/etc/hosts文件  - name: "1 | update the /etc/hosts"    template:      src: ./hosts.j2      dest: /etc/hosts  # 2、创建ZK安装目录，并赋予相关用户权限  - name: "2 | create the work_dir"    file:      path: /usr/local/zookeeper      state: directory      owner: hadoop      group: hadoop      mode: 0755  # 3、从HTTP服务器下载安装包并解压  - name: "3 | download and unpack the zookeeper package"    unarchive:      src: http://10.xx.x.xx/td-configuration/deploy/clickhouse/{{ zookeeper_version }}      dest: /tmp/      remote_src: yes      owner: hadoop      group: hadoop  # 4、 判断解压后的目录中是否有bin目录，有则为true  - name: "4 | stat /tmp/{{zookeeper_dir}}/bin"    stat:      path: /tmp/{{zookeeper_dir}}/bin    register: zookeeper_stat  # 5、再次判断目标端目录中是否有bin目录，无则为false  - name: "5 | stat /usr/local/zookeeper/bin"    stat:      path: /usr/local/zookeeper/bin    register: bin_stat  # 6、判断解压目录、移动目标目录，如果有解压目录且目标目录没有bin目录，可以执行移动目录操作（即第4步为true，第5步为false执行移动操作）  - name: "6 | move directory |"    shell: mv /tmp/{{zookeeper_dir}}/* /usr/local/zookeeper/    args:      executable: /bin/bash    # 条件判断通过后执行移动目录操作    when:      - zookeeper_stat.stat.exists      - not bin_stat.stat.exists  # 7、设置ZK配置文件  - name: "7 | update the /usr/local/zookeeper/conf/zoo.cfg"    template:      src: ./zoo.cfg.j2      dest: /usr/local/zookeeper/conf/zoo.cfg      owner: hadoop      group: hadoop      mode: 0644  # 8、创建ZK数据、日志目录并赋予用户权限  - name: "8 | create the other directories"    file:      path: /usr/loacl/zookeeper/{{ item }}      state: directory      owner: hadoop      group: hadoop      mode: 0755    with_items:      - 'data'      - 'log'  # 9、创建各个节点myid文件  - name: "9 | create the myid"    lineinfile:      path: /usr/local/zookeeper/data/myid      state: present      create: yes      # 获取ip_list中定义的ID变量值，实现赋值      regexp: '^\s*{{ ID }}'      line: "{{ ID |string }}"  # 10、设置ZK环境变量  - name: "10 | env"    lineinfile:      path: /etc/profile      state: present      regexp: '{{ item.S }}'      line: "{{ item.D }}"    with_items:      - { S: '^\s*export\s+ZK_HOME=.*',D: 'export ZK_HOME=/usr/local/zookeeper' }      - { S: '^\s*export\s+PATH=.*ZK_HOME.*',D: 'export PATH=$PATH:$ZK_HOME/bin' }  # 11、再次递归相应目录用户权限  - name: "11 | chown"    file:      path: /usr/local/zookeeper      owner: hadoop      group: hadoop      recurse: yes  # 12、启动ZK服务  - name: "12 | start service"    shell: |      source /etc/profile      /usr/local/zookeeper/bin/zkServer.sh start    ignore_errors: yes

ansible-playbook -i ip_list deploy.zookeeper.yaml

ansible-playbook -i ip_list deploy.zookeeper.yaml --key-file=sz-key

03

mkdir clickhousecd clickhousevim ip_list[clickhouse]172.xx.xx.x172.xx.xx.x172.xx.xx.x172.xx.xx.x

vim deploy.clickhouse.yaml---- hosts: clickhouse  remote_user: root  gather_facts: true  tasks:  # 1、下载相关安装包并安装  - name: "1 | download and installed"    yum:      name:      - http://10.xx.x.xx/td-configuration/deploy/clickhouse/v21.9.5.16-stable/clickhouse-common-static-21.9.5.16-2.x86_64.rpm      - http://10.xx.x.xx/td-configuration/deploy/clickhouse/v21.9.5.16-stable/clickhouse-common-static-dbg-21.9.5.16-2.x86_64.rpm      - http://10.xx.x.xx/td-configuration/deploy/clickhouse/v21.9.5.16-stable/clickhouse-server-21.9.5.16-2.noarch.rpm      - http://10.xx.x.xx/td-configuration/deploy/clickhouse/v21.9.5.16-stable/clickhouse-client-21.9.5.16-2.noarch.rpm      state: present  # 2、创建CK相关目录  - name: "2 | create directories"    file:      path: /data/clickhouse-server/{{ item }}      state: directory    with_items:      - 'access'      - 'data'      - 'format_schemas'      - 'logs'      - 'tmp'      - 'user_files'  # 3、CK目录赋予对应用户权限  - name: "3 | chown"    file:      path: /data/clickhouse-server      owner: clickhouse      group: clickhouse      recurse: yes

ansible-playbook -i ip_list deploy.clickhouse.yaml --key-file=sz-key

vim /etc/clickhouse-server/config.xml<!-- 日志目录 --><log>/data/clickhouse-server/logs/clickhouse-server.log</log><errorlog>/data/clickhouse-server/logs/clickhouse-server.err.log</errorlog><!-- CK内部HTTP通信HOST，填写本机IP，每个节点都不一样 --><interserver_http_host>172.xx.xx.x</interserver_http_host><!-- 监听HOST --><listen_host>0.0.0.0</listen_host><!-- 软件的目录配置，同Ansible安装创建的目录 --><path>/data/clickhouse-server/data/</path><tmp_path>/data/clickhouse-server/tmp/</tmp_path><user_files_path>/data/clickhouse-server/user_files/</user_files_path><local_directory><path>/data/clickhouse-server/access/</path></local_directory><format_schema_path>/data/clickhouse-server/format_schemas/</format_schema_path><!-- 指定ZK配置 --><zookeeper incl="zookeeper-servers" optional="true" /><!-- 指定外部CK集群配置文件 --><include_from>/etc/clickhouse-server/config.d/metrika.xml</include_from>

vim /etc/clickhouse-server/config.d/metrika.xml<?xml version="1.0"?><yandex><!--zookeeper相关配置-->    <zookeeper-servers>     <!--需要与config.xml中zookerper的incl名称匹配-->        <node index="1">    <!--node节点配置-->            <host>172.xx.xx.x</host>   <!--zk_host-->            <port>2182</port>   <!--zk_port-->        </node>        <node index="2">            <host>172.xx.xx.x</host>            <port>2182</port>        </node>        <node index="3">            <host>172.xx.xx.x</host>            <port>2182</port>        </node>    </zookeeper-servers>    <remote_servers>        <!-- 集群名称 -->        <test_cluster>            <!-- 集群第1个分片 -->            <shard>                <internal_replication>true</internal_replication>                <!-- 集群第1个分片的第1个副本 -->                <replica>                    <host>172.xx.xx.x</host>                    <port>9000</port>                </replica>                <!-- 集群第1个分片的第2个副本 -->                <replica>                    <host>172.xx.xx.x</host>                    <port>9000</port>                </replica>            </shard>            <!-- 集群第2个分片 -->            <shard>                <internal_replication>true</internal_replication>                <!-- 集群第2个分片的第1个副本 -->                <replica>                    <host>172.xx.xx.x</host>                    <port>9000</port>                </replica>                <!-- 集群第2个分片的第2个副本 -->                <replica>                    <host>172.xx.xx.x</host>                    <port>9000</port>                </replica>            </shard>        </test_cluster>    </remote_servers>    <macros>        <layer>01</layer>       <!-- 集群标识 -->        <shard>01</shard>       <!-- 分片标识：不同机器放的分片数不一样 -->        <replica>rep_1_1</replica>       <!-- 副本标识：不同机器放的副本数不一样 -->    </macros></yandex>

# 启动服务，所有节点都需要执行systemctl start clickhouse-server# 查看CK服务状态systemctl status clickhouse-server

# 登录SQL对话框clickhouse-client -m# 查看当前集群状态SELECT * FROM system.clusters;# 查看ZK信息SELECT * FROM system.zookeeper where path='/clickhouse';

04

zkCli.sh -server 172.xx.xx.x:2182ls /ls -s /clickhousels -w /clickhousels -R /clickhouse