# ClickHouse集群部署 防止一些意外的情况 1、取消打开文件数限制 在/etc/security/limits.conf、/etc/security/limits.d/20-nproc.conf这2个文件的末尾加入一下 文件末尾加： ~~~ * soft nofile 65536 * hard nofile 65536 * soft nproc 131072 * hard nproc 131072 ~~~ 2、取消SELINUX 修改/etc/selinux/config 中 SELINUX=disabled 3、关闭防火墙 service iptables stop 4、安装依赖 ~~~ yum install -y libtool yum install -y *unixODBC ~~~ 5、安装三台单机clickhouse 略 6、配置metrika.xml文件 在/etc/clickhouse-server/config.d文件夹下添加一个名为metrika.xml的配置文件 ~~~XML <yandex> <clickhouse_remote_servers> <!--集群名称，clickhouse支持多集群的模式--> <clickhouse_cluster> <!--定义分片节点，这里我指定3个分片，每个分片只有1个副本，也就是它本身--> <shard> <internal_replication>true</internal_replication> <replica> <host>server1</host> <port>9000</port> </replica> </shard> <shard> <replica> <internal_replication>true</internal_replication> <host>server2</host> <port>9000</port> </replica> </shard> <shard> <internal_replication>true</internal_replication> <replica> <host>server3</host> <port>9000</port> </replica> </shard> </clickhouse_cluster> </clickhouse_remote_servers> <!--zookeeper集群的连接信息--> <zookeeper-servers> <node index="1"> <host>server1</host> <port>2181</port> </node> <node index="2"> <host>server1</host> <port>2182</port> </node> <node index="3"> <host>server1</host> <port>2183</port> </node> </zookeeper-servers> <!--定义宏变量，后面需要用--> <macros> <replica>server1</replica> </macros> <!--不限制访问来源ip地址--> <networks> <ip>::/0</ip> </networks> <!--数据压缩方式，默认为lz4--> <clickhouse_compression> <case> <min_part_size>10000000000</min_part_size> <min_part_size_ratio>0.01</min_part_size_ratio> <method>lz4</method> </case> </clickhouse_compression> </yandex> ~~~ 注意配置中变量macros为对应主机名不能一样 7、修改三台机器的config.xml的配置 vi /etc/clickhouse-server/config.xml 取消注释:: ~~~ <!--引入metrika.xml--> <include_from>/etc/clickhouse-server/config.d/metrika.xml</include_from> #引用Zookeeper配置的定义 <zookeeper incl="zookeeper-servers" optional="true" /> ~~~ 8、启动集群 1）启动Zookeeper集群 ~~~ zkServer.sh start zk1.cfg zkServer.sh start zk2.cfg zkServer.sh start zk3.cfg ~~~ 2）启动ClickHouse 分别启动ClickHouse ~~~ sudo clickhouse start ~~~ 9、验证集群 select \* from system.clusters  10、使用ClickHouse访问Zookeeper 在ClickHouse系统表中，提供了一张Zookeeper代理表，我们可以使用SQL轻松访问Zookeeper内的数据，不用再像以前一样使用客户端登录进去查看。 ~~~ -- 查询Zookeeper根目录 select * from system.zookeeper where path = '/' -- 查询ClickHouse目录 select * from system.zookeeper where path = '/clickhouse' ~~~ 文档参考：[https://blog.csdn.net/qq\_42194171/article/details/109560651](https://blog.csdn.net/qq_42194171/article/details/109560651)