0
0

Clickhouse keeper方式搭建集群（不再依赖zookeeper）

下面是如何在三个节点（192.168.250.140, 192.168.250.146, 192.168.250.147）上部署 ClickHouse 集群，并使用 ClickHouse Keeper 作为分布式协调服务的详细步骤。

Step 1: 安装 ClickHouse

首先，在每个节点上安装 ClickHouse。以下命令适用于 Debian 系统。

wget -O clickhouse-client_24.4.3.25_amd64.deb https://github.com/ClickHouse/ClickHouse/releases/download/v24.4.3.25-stable/clickhouse-client_24.4.3.25_amd64.deb
wget -O clickhouse-common-static_24.4.3.25_amd64.deb https://github.com/ClickHouse/ClickHouse/releases/download/v24.4.3.25-stable/clickhouse-common-static_24.4.3.25_amd64.deb
wget -O clickhouse-server_24.4.3.25_amd64.deb https://github.com/ClickHouse/ClickHouse/releases/download/v24.4.3.25-stable/clickhouse-server_24.4.3.25_amd64.deb

sudo dpkg -i clickhouse-common-static_24.4.3.25_amd64.deb
sudo dpkg -i clickhouse-client_24.4.3.25_amd64.deb
sudo dpkg -i clickhouse-server_24.4.3.25_amd64.deb
dpkg -l | grep clickhouse  --是否安装成功

Step 2: 配置 ClickHouse Keeper

在每个节点上配置 ClickHouse Keeper。编辑 /etc/clickhouse-server/config.xml 文件，添加 ClickHouse Keeper 的配置。

在每个节点上，添加以下配置：

<clickhouse>
      <keeper_server>
        <tcp_port>9181</tcp_port>
        <server_id>1</server_id>
        <log_storage_path>/var/lib/clickhouse/coordination/log</log_storage_path>
        <snapshot_storage_path>/var/lib/clickhouse/coordination/snapshots</snapshot_storage_path>
        <coordination_settings>
            <operation_timeout_ms>10000</operation_timeout_ms>
            <session_timeout_ms>30000</session_timeout_ms>
            <raft_logs_level>trace</raft_logs_level>
        </coordination_settings>
        <raft_configuration>
            <server>
                <id>1</id>
                <hostname>192.168.252.140</hostname>
                <port>9234</port>
            </server>
            <server>
                <id>2</id>
                <hostname>192.168.252.146</hostname>
                <port>9234</port>
            </server>
            <server>
                <id>3</id>
                <hostname>192.168.252.147</hostname>
                <port>9234</port>
            </server>
        </raft_configuration>
    </keeper_server>

</clickhouse>

注意：在每个节点上，server_id 应该是唯一的。例如：

● 192.168.250.140 上的 server_id 为 1
● 192.168.250.146 上的 server_id 为 2
● 192.168.250.147 上的 server_id 为 3

Step 3: 配置分片和副本

在每个节点上，编辑 /etc/clickhouse-server/config.d/clusters.xml 文件，配置分片和副本。

<clickhouse>
    <remote_servers>
        <log_cluster> <!-- 使用了自定义的集群名称 -->
            <shard>
                <replica>
                    <host>192.168.250.140</host>
                    <port>9000</port>
                </replica>
                <replica>
                    <host>192.168.250.146</host>
                    <port>9000</port>
                </replica>
            </shard>
            <shard>
                <replica>
                    <host>192.168.250.147</host>
                    <port>9000</port>
                </replica>
                <replica>
                    <host>192.168.250.140</host>
                    <port>9000</port>
                </replica>
            </shard>
        </log_cluster>
    </remote_servers>
</clickhouse>

Step 4: 配置宏

在对应的节点上加入宏

//在192.168.250.140加入这个宏
<macros>
    <shard>01</shard>
    <replica>192.168.250.140</replica>
</macros>
//在192.168.250.146加入这个宏
<macros>
    <shard>02</shard>
    <replica>192.168.250.146</replica>
</macros>
//在192.168.250.147加入这个宏
<macros>
    <shard>03</shard>
    <replica>192.168.250.147</replica>
</macros>

Step 5: 启动 ClickHouse 服务

我使用的所有配置都在/etc/clickhouse-server/config.xml里面
所以注释掉keeper_config.xml里面的 <keeper_server>标签及 <keeper_server>内的内容

Step 6: 启动 ClickHouse 服务

在每个节点上启动 ClickHouse 服务：

sudo systemctl start clickhouse-server

并确保服务已启动：

sudo systemctl status clickhouse-server

如果启动不能完成初始化如下，不妨删除旧数据

2024.06.26 21:03:29.159411 [ 94688 ] {} <Information> KeeperDispatcher: Server still not initialized, will not apply configuration until initialization finished
2024.06.26 21:03:34.159564 [ 94688 ] {} <Information> KeeperDispatcher: Server still not initialized, will not apply configuration until initialization finished
2024.06.26 21:03:39.159754 [ 94688 ] {} <Information> KeeperDispatcher: Server still not initialized, will not apply configuration until initialization finished
2024.06.26 21:03:44.159931 [ 94688 ] {} <Information> KeeperDispatcher: Server still not initialized, will not apply configuration until initialization finished
2024.06.26 21:03:49.160159 [ 94688 ] {} <Information> KeeperDispatcher: Server still not initialized, will not apply configuration until initialization finished

  sudo rm -rf /var/lib/clickhouse/coordination/log/*
  sudo rm -rf /var/lib/clickhouse/coordination/snapshots/*

Step 6: 创建分布式表

使用 ClickHouse 客户端连接到任意一个节点，创建分布式表。

clickhouse-client -h 192.168.250.140

-- 创建数据库
CREATE DATABASE IF NOT EXISTS test;

-- 创建本地表
CREATE TABLE test.local_table ON CLUSTER log_cluster (
    id UInt32,
    name String
) ENGINE = ReplicatedMergeTree('/clickhouse/tables/{shard}/test/local_table', '{replica}')
ORDER BY id;

-- 创建分布式表
CREATE TABLE test.distributed_table ON CLUSTER your_cluster_name AS test.local_table
ENGINE = Distributed(log_cluster, test, local_table, rand());

配置总结

<listen_host>0.0.0.0</listen_host>
       <log_cluster>
            <shard>
                <replica>
                    <host>192.168.250.140</host>
                    <port>9000</port>
                </replica>
            </shard>
            <shard>
                <replica>
                    <host>192.168.250.146</host>
                    <port>9000</port>
                </replica>
            </shard>
            <shard>
                <replica>
                    <host>192.168.250.147</host>
                    <port>9000</port>
                </replica>
            </shard>
        </log_cluster>
       <keeper_server>
        <tcp_port>9181</tcp_port>
        <server_id>1</server_id>
        <log_storage_path>/var/lib/clickhouse/coordination/log</log_storage_path>
        <snapshot_storage_path>/var/lib/clickhouse/coordination/snapshots</snapshot_storage_path>
        <coordination_settings>
            <operation_timeout_ms>10000</operation_timeout_ms>
            <session_timeout_ms>30000</session_timeout_ms>
            <raft_logs_level>trace</raft_logs_level>
        </coordination_settings>
        <raft_configuration>
            <server>
                <id>1</id>
                <hostname>192.168.252.140</hostname>
                <port>9234</port>
            </server>
            <server>
                <id>2</id>
                <hostname>192.168.252.146</hostname>
                <port>9234</port>
            </server>
            <server>
                <id>3</id>
                <hostname>192.168.252.147</hostname>
                <port>9234</port>
            </server>
        </raft_configuration>
    </keeper_server>

<macros>
    <shard>shardNum</shard>
    <replica>192.168.250.140</replica>
</macros>
标签： clickhouse zookeeper 分布式
本文转载自: https://blog.csdn.net/wyl15982038375/article/details/139999634
版权归原作者 wyl15982038375 所有， 如有侵权，请联系我们删除。
登录后发布评论

“Clickhouse keeper方式搭建集群（不再依赖zookeeper）”的评论:

还没有评论
关于作者
...
overfit同步小助手
文章同步
相关阅读

hive分区详细教程

使用Python实现对接Hadoop集群（通过Hive）并提供API接口

springboot整合nacos

Postman Pre-request Script

Apifox、Postman进行接口调用时如何携带cookie？

使用IDEA启动项目build时，解决Java编译时内存溢出问题：OutOfMemoryError深入解析

zookeeper实现RMI服务，高可用，HA

文章导航