手把手教你在MRS集群部署ClickHouse服务

举报
he lifu 发表于 2020/10/21 19:32:40 2020/10/21
【摘要】 继上篇在华为云编译和使用单点的ClickHouse,本篇博文会向大家分享MRS集群上配置部署ClickHouse服务,是ClickHouse集群:)

1 前言

继上篇在华为云编译和使用单点的ClickHouse,本篇博文会向大家分享MRS集群上配置部署ClickHouse服务,是ClickHouse集群:)

2 准备工作

因为是在MRS集群上部署,所以需要先购买一个MRS集群,这里以“MRS 2.1.0”版本的流式集群为例,注意下图中的箭头:

MRS集群购买以后,记得配置“管理安全组规则”,把前面博文中编译机IP地址添加到“入方向规则”中,这样可以从编译机scp拷贝ClickHouse二进制文件传送到MRS集群上。我这里为了省点费用,选择了非高可用的集群,所以是1个master和3个core节点:

我计划把ClickHouse集群部署在3个core节点上。

3 部署集群

3.1 上传文件

  • 上传clickhouse二进制文件到上面的三个core节点上,只需要clickhouse这个文件;

  • 上传config.xml、metrika.xml、user.xml文件到上面三个core节点上;

[root@node-str-coreCWtk0002 ~]# ls -lrt
total 3170348
-rw-r--r--. 1 root root         79 Nov  8  2019 env_file
-rwx------. 1 root root 3246411984 Oct 21 17:54 clickhouse
-rw-r--r--. 1 root root       1483 Oct 21 18:36 users.xml
-rw-r--r--. 1 root root       1603 Oct 21 18:39 metrika.xml
-rw-r--r--. 1 root root       3396 Oct 21 18:45 config.xml
[root@node-str-coreCWtk0002 ~]#

3.2 获取zookeeper信息

可以有2个方式获取到zk的信息:

  • 读取配置文件

[root@node-master1azib ~]# cat /opt/client/ZooKeeper/zookeeper/conf/zoo.cfg
  • 在MRS集群的每台机器上执行

[root@node-str-coreCWtk0002 ~]# netstat -an | grep 2181
tcp        0      0 192.168.0.220:2181      0.0.0.0:*               LISTEN

侦听2181端口的话表示该节点部署有zookeeper节点,然后把他们配置到metrika.xml中的zookeeper-servers中。

3.3 修改配置文件

3个配置文件的配置信息如下,大家只需要关注并修改其中的IP地址即可。

  • config.xml:

<?xml version="1.0"?>
<yandex>
    <logger>
      <level>information</level>
      <log>/var/log/Bigdata/clickhouse/clickhouse-server.log</log>
      <errorlog>/var/log/Bigdata/clickhouse/clickhouse-server.err.log</errorlog>
      <size>1000M</size>
      <count>10</count>
    </logger>
    <http_port>8123</http_port>
    <tcp_port>9000</tcp_port>
    <interserver_http_port>9009</interserver_http_port>
    <interserver_http_host>192.168.0.17</interserver_http_host>
    <listen_host>192.168.0.17</listen_host>
    <max_connections>4096</max_connections>
    <keep_alive_timeout>600</keep_alive_timeout>
    <max_concurrent_queries>150</max_concurrent_queries>
    <uncompressed_cache_size>8589934592</uncompressed_cache_size>
    <mark_cache_size>5368709120</mark_cache_size>
    <path>/srv/BigData/streaming/data1/clickhouse/</path>
    <tmp_path>/srv/BigData/streaming/data1/clickhouse/tmp/</tmp_path>
    <users_config>/root/users.xml</users_config>
    <default_profile>default</default_profile>
    <default_database>default</default_database>
    <mlock_executable>false</mlock_executable>
    <include_from>/root/metrika.xml</include_from>
    <remote_servers incl="clickhouse_remote_servers" />
    <zookeeper incl="zookeeper-servers" optional="true" />
    <macros incl="macros" optional="true" />
    <builtin_dictionaries_reload_interval>3600</builtin_dictionaries_reload_interval>
    <max_session_timeout>3600</max_session_timeout>
    <default_session_timeout>60</default_session_timeout>
    <query_log>
      <database>system</database>
      <table>query_log</table>
      <partition_by>toYYYYMM(event_date)</partition_by>
      <flush_interval_milliseconds>7500</flush_interval_milliseconds>
    </query_log>
    <trace_log>
      <database>system</database>
      <table>trace_log</table>
      <partition_by>toYYYYMM(event_date)</partition_by>
      <flush_interval_milliseconds>7500</flush_interval_milliseconds>
    </trace_log>
    <query_thread_log>
      <database>system</database>
      <table>query_thread_log</table>
      <partition_by>toYYYYMM(event_date)</partition_by>
      <flush_interval_milliseconds>7500</flush_interval_milliseconds>
    </query_thread_log>
    <dictionaries_config>*_dictionary.xml</dictionaries_config>
    <compression incl="clickhouse_compression"></compression>
    <distributed_ddl>
      <path>/clickhouse/task_queue/ddl</path>
    </distributed_ddl>
    <graphite_rollup_example>
        <pattern>
            <regexp>click_cost</regexp>
            <function>any</function>
            <retention>
                <age>0</age>
                <precision>3600</precision>
            </retention>
            <retention>
                <age>86400</age>
                <precision>60</precision>
            </retention>
        </pattern>
        <default>
            <function>max</function>
            <retention>
                <age>0</age>
                <precision>60</precision>
            </retention>
            <retention>
                <age>3600</age>
                <precision>300</precision>
            </retention>
            <retention>
                <age>86400</age>
                <precision>3600</precision>
            </retention>
        </default>
    </graphite_rollup_example>
    <format_schema_path>/srv/BigData/streaming/data1/clickhouse/format_schemas/</format_schema_path>
</yandex>

只需要修改:interserver_http_host、listen_host为本机的IP地址;

  • metrika.xml:

<yandex>
  <clickhouse_remote_servers>
    <cluster1>
      <shard>
        <replica>
          <host>192.168.0.17</host>
          <port>9000</port>
        </replica>
      </shard>
      <shard>
        <replica>
          <host>192.168.0.57</host>
          <port>9000</port>
        </replica>
      </shard>
      <shard>
        <replica>
          <host>192.168.0.220</host>
          <port>9000</port>
        </replica>
      </shard>
    </cluster1>
    <cluster2>
      <shard>
        <internal_replication>true</internal_replication>
        <replica>
          <host>192.168.0.17</host>
          <port>9000</port>
        </replica>
        <replica>
          <host>192.168.0.57</host>
          <port>9000</port>
        </replica>
        <replica>
          <host>192.168.0.220</host>
          <port>9000</port>
        </replica>
      </shard>
    </cluster2>
  </clickhouse_remote_servers>
  <macros>
    <cluster>cluster2</cluster>
    <shard>1</shard>
    <replica>192.168.0.17</replica>
  </macros>
  <zookeeper-servers>
    <node index="1">
      <host>192.168.0.57</host>
      <port>2181</port>
    </node>
    <node index="2">
      <host>192.168.0.134</host>
      <port>2181</port>
    </node>
    <node index="3">
      <host>192.168.0.220</host>
      <port>2181</port>
    </node>
  </zookeeper-servers>
  <allow_ip_list>
    <ip>::/0</ip>
  </allow_ip_list>
  <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>

这里我定义了2个cluster:cluster1表示3shard1replica,cluster2表示1shard3replica,大家可以自行修改;macros宏定义部分修改为本机的IP地址;zookeeper-servers配置成前面找的zookeeper地址;

  • users.xml:

<?xml version="1.0"?>
<yandex>
    <profiles>
      <default>
        <log_queries>1</log_queries>
        <max_memory_usage>66000000000</max_memory_usage>
        <max_threads>14</max_threads>
        <background_pool_size>32</background_pool_size>
        <max_pipeline_depth>2000</max_pipeline_depth>
        <load_balancing>random</load_balancing>
      </default>
      <readonly>
        <readonly>1</readonly>
      </readonly>
      <web>
        <max_execution_time>300</max_execution_time>
        <max_columns_to_read>30</max_columns_to_read>
        <max_rows_to_read>10000000000</max_rows_to_read>
        <max_rows_to_group_by>10000000000</max_rows_to_group_by>
        <max_rows_to_sort>10000000000</max_rows_to_sort>
        <max_rows_in_distinct>10000000000</max_rows_in_distinct>
        <max_rows_to_transfer>10000000000</max_rows_to_transfer>
        <readonly>1</readonly>
      </web>
    </profiles>
    <users>
    <default>
        <password></password>
        <networks>
        <ip>::/0</ip>
        </networks>
        <profile>default</profile>
        <quota>default</quota>
      </default>
</users>
    <quotas>
      <default>
        <interval>
          <duration>0</duration>
            <queries>0</queries>
              <errors>0</errors>
              <result_rows>0</result_rows>
              <read_rows>0</read_rows>
              <execution_time>0</execution_time>
            </interval>
        </default>
    </quotas>
</yandex>

这个不用改。

3.4 启动服务端

在3个core节点上分别执行:

[root@node-str-coreCWtk0002 ~]# nohup ./clickhouse server --config-file=config.xml &

3.5 启动客户端

[root@node-str-coreCWtk0002 ~]# ./clickhouse client --host=192.168.0.220
ClickHouse client version 20.9.4.1.
Connecting to 192.168.0.220:9000 as user default.
Connected to ClickHouse server version 20.9.4 revision 54439.

node-str-coreCWtk0002 :) show databases;

SHOW DATABASES

┌─name───────────────────────────┐
│ _temporary_and_external_tables │
│ default                        │
│ system                         │
└────────────────────────────────┘

3 rows in set. Elapsed: 0.001 sec. 

node-str-coreCWtk0002 :) select * from system.clusters

SELECT *
FROM system.clusters

┌─cluster──┬─shard_num─┬─shard_weight─┬─replica_num─┬─host_name─────┬─host_address──┬─port─┬─is_local─┬─user────┬─default_database─┬─errors_count─┬─estimated_recovery_time─┐
│ cluster1 │         1 │            1 │           1 │ 192.168.0.17  │ 192.168.0.17  │ 9000 │        0 │ default │                  │            0 │                       0 │
│ cluster1 │         2 │            1 │           1 │ 192.168.0.57  │ 192.168.0.57  │ 9000 │        0 │ default │                  │            0 │                       0 │
│ cluster1 │         3 │            1 │           1 │ 192.168.0.220 │ 192.168.0.220 │ 9000 │        1 │ default │                  │            0 │                       0 │
│ cluster2 │         1 │            1 │           1 │ 192.168.0.17  │ 192.168.0.17  │ 9000 │        0 │ default │                  │            0 │                       0 │
│ cluster2 │         1 │            1 │           2 │ 192.168.0.57  │ 192.168.0.57  │ 9000 │        0 │ default │                  │            0 │                       0 │
│ cluster2 │         1 │            1 │           3 │ 192.168.0.220 │ 192.168.0.220 │ 9000 │        1 │ default │                  │            0 │                       0 │
└──────────┴───────────┴──────────────┴─────────────┴───────────────┴───────────────┴──────┴──────────┴─────────┴──────────────────┴──────────────┴─────────────────────────┘

6 rows in set. Elapsed: 0.002 sec. 

node-str-coreCWtk0002 :)

3.6 测试验证

[root@node-str-coreCWtk0002 ~]# ./clickhouse client --host=192.168.0.220 --multiline
ClickHouse client version 20.9.4.1.
Connecting to 192.168.0.220:9000 as user default.
Connected to ClickHouse server version 20.9.4 revision 54439.
node-str-coreCWtk0002 :) create table table2 on cluster cluster2 (
:-]   `EventDate` DateTime,
:-]   `id` UInt64
:-] ) engine = ReplicatedMergeTree('/clickhouse/{cluster}/{shard}/table2', '{replica}')
:-] partition by toYYYYMM(EventDate)
:-] order by id;
CREATE TABLE table2 ON CLUSTER cluster2
(
    `EventDate` DateTime,
    `id` UInt64
)
ENGINE = ReplicatedMergeTree('/clickhouse/{cluster}/{shard}/table2', '{replica}')
PARTITION BY toYYYYMM(EventDate)
ORDER BY id
┌─host──────────┬─port─┬─status─┬─error─┬─num_hosts_remaining─┬─num_hosts_active─┐
│ 192.168.0.57  │ 9000 │      0 │       │                   2 │                0 │
│ 192.168.0.17  │ 9000 │      0 │       │                   1 │                0 │
│ 192.168.0.220 │ 9000 │      0 │       │                   0 │                0 │
└───────────────┴──────┴────────┴───────┴─────────────────────┴──────────────────┘
3 rows in set. Elapsed: 0.121 sec. 
node-str-coreCWtk0002 :) 
node-str-coreCWtk0002 :) create table table2_all on cluster cluster2 (
:-]   `EventDate` DateTime,
:-]   `id` UInt64
:-] ) engine = Distributed(cluster2, default, table2, rand());
CREATE TABLE table2_all ON CLUSTER cluster2
(
    `EventDate` DateTime,
    `id` UInt64
)
ENGINE = Distributed(cluster2, default, table2, rand())
┌─host──────────┬─port─┬─status─┬─error─┬─num_hosts_remaining─┬─num_hosts_active─┐
│ 192.168.0.57  │ 9000 │      0 │       │                   2 │                0 │
│ 192.168.0.17  │ 9000 │      0 │       │                   1 │                0 │
│ 192.168.0.220 │ 9000 │      0 │       │                   0 │                0 │
└───────────────┴──────┴────────┴───────┴─────────────────────┴──────────────────┘
3 rows in set. Elapsed: 0.117 sec. 
node-str-coreCWtk0002 :) show tables;
SHOW TABLES
┌─name───────┐
│ table2     │
│ table2_all │
└────────────┘
2 rows in set. Elapsed: 0.002 sec. 
node-str-coreCWtk0002 :) INSERT INTO table2 VALUES ('2020-01-01 01:01:01', 1);
INSERT INTO table2 VALUES
Ok.
1 rows in set. Elapsed: 0.012 sec. 
node-str-coreCWtk0002 :) INSERT INTO table2_all VALUES ('2020-02-02 02:02:02', 2);
INSERT INTO table2_all VALUES
Ok.
1 rows in set. Elapsed: 0.009 sec. 
node-str-coreCWtk0002 :) select * from table2;
SELECT *
FROM table2
┌───────────EventDate─┬─id─┐
│ 2020-02-02 02:02:02 │  2 │
└─────────────────────┴────┘
┌───────────EventDate─┬─id─┐
│ 2020-01-01 01:01:01 │  1 │
└─────────────────────┴────┘
2 rows in set. Elapsed: 0.002 sec. 
node-str-coreCWtk0002 :) select * from table2_all;
SELECT *
FROM table2_all
┌───────────EventDate─┬─id─┐
│ 2020-01-01 01:01:01 │  1 │
└─────────────────────┴────┘
┌───────────EventDate─┬─id─┐
│ 2020-02-02 02:02:02 │  2 │
└─────────────────────┴────┘
2 rows in set. Elapsed: 0.002 sec. 
node-str-coreCWtk0002 :)

至此,MRS集群上的ClickHouse服务部署完成,且能正常工作^_^

【版权声明】本文为华为云社区用户原创内容,转载时必须标注文章的来源(华为云社区)、文章链接、文章作者等基本信息, 否则作者和本社区有权追究责任。如果您发现本社区中有涉嫌抄袭的内容,欢迎发送邮件进行举报,并提供相关证据,一经查实,本社区将立刻删除涉嫌侵权内容,举报邮箱: cloudbbs@huaweicloud.com
  • 点赞
  • 收藏
  • 关注作者

评论(0

0/1000
抱歉,系统识别当前为高风险访问,暂不支持该操作

全部回复

上滑加载中

设置昵称

在此一键设置昵称,即可参与社区互动!

*长度不超过10个汉字或20个英文字符,设置后3个月内不可修改。

*长度不超过10个汉字或20个英文字符,设置后3个月内不可修改。

举报
请填写举报理由
0/200