概述
ClickHouse 的 集群发现 功能可让节点自动发现并注册自身,无需在配置文件中逐一定义,从而简化集群配置。在必须手动定义每个节点时,这项功能尤其有用。
远程服务器配置
传统的手动配置
以往,在 ClickHouse 中,集群中的每个分片和副本都需要在配置中手动指定:
<remote_servers>
<cluster_name>
<shard>
<replica>
<host>node1</host>
<port>9000</port>
</replica>
<replica>
<host>node2</host>
<port>9000</port>
</replica>
</shard>
<shard>
<replica>
<host>node3</host>
<port>9000</port>
</replica>
<replica>
<host>node4</host>
<port>9000</port>
</replica>
</shard>
</cluster_name>
</remote_servers>使用集群发现
使用 Cluster Discovery 时,无需逐个显式定义节点,只需在 ZooKeeper 中指定一个 path。所有在 ZooKeeper 中注册到该 path 下的节点都会被自动发现并添加到 cluster 中。
<remote_servers>
<cluster_name>
<discovery>
<path>/clickhouse/discovery/cluster_name</path>
<!-- # 可选配置参数: -->
<!-- ## 访问集群中其他节点所需的身份验证凭据: -->
<!-- <user>user1</user> -->
<!-- <password>pass123</password> -->
<!-- ### 也可使用节点间 secret 代替密码: -->
<!-- <secret>secret123</secret> -->
<!-- ## 当前节点的分片(见下文): -->
<!-- <shard>1</shard> -->
<!-- ## 观察者模式(见下文): -->
<!-- <observer/> -->
</discovery>
</cluster_name>
</remote_servers>如果你想为某个节点指定分片编号,可以在 <discovery> 部分内添加 <shard> 标签:
对于 node1 和 node2:
<discovery>
<path>/clickhouse/discovery/cluster_name</path>
<shard>1</shard>
</discovery>对于 node3 和 node4:
<discovery>
<path>/clickhouse/discovery/cluster_name</path>
<shard>2</shard>
</discovery>观察者模式
配置为观察者模式的节点不会将自己注册为副本。
它们只会观察并发现集群中其他处于活动状态的副本,而不会主动参与。
要启用观察者模式,请在 <discovery> 部分中添加 <observer/> 标签:
<discovery>
<path>/clickhouse/discovery/cluster_name</path>
<observer/>
</discovery>集群发现
有时,您可能不仅需要添加和移除集群中的主机,还需要添加和移除集群本身。您可以使用 <multicluster_root_path> 节点,为多个集群指定根路径:
<remote_servers>
<some_unused_name>
<discovery>
<multicluster_root_path>/clickhouse/discovery</multicluster_root_path>
<observer/>
</discovery>
</some_unused_name>
</remote_servers>在这种情况下,当其他主机通过路径 /clickhouse/discovery/some_new_cluster 注册自身时,就会添加一个名为 some_new_cluster 的集群。
你可以同时使用这两种功能:主机既可以在集群 my_cluster 中注册自身,也可以发现其他任意集群:
<remote_servers>
<my_cluster>
<discovery>
<path>/clickhouse/discovery/my_cluster</path>
</discovery>
</my_cluster>
<some_unused_name>
<discovery>
<multicluster_root_path>/clickhouse/discovery</multicluster_root_path>
<observer/>
</discovery>
</some_unused_name>
</remote_servers>限制:
- 不能在同一个
remote_servers子树中同时使用<path>和<multicluster_root_path>。 <multicluster_root_path>只能与<observer/>搭配使用。- 集群名称使用 Keeper 中路径的最后一部分;而在注册时,该名称取自 XML 标签。
用例和限制
当节点被添加到指定的 ZooKeeper 路径或从中移除时,无需更改配置或重启服务器,系统就会自动发现这些节点,或将其从集群中移除。
但是,这些变更只会影响集群配置,不会影响数据或现有的数据库和表。
请考虑以下一个包含 3 个节点的集群示例:
<remote_servers>
<default>
<discovery>
<path>/clickhouse/discovery/default_cluster</path>
</discovery>
</default>
</remote_servers>SELECT * EXCEPT (default_database, errors_count, slowdowns_count, estimated_recovery_time, database_shard_name, database_replica_name)
FROM system.clusters WHERE cluster = 'default';
┌CREATE TABLE event_table ON CLUSTER default (event_time DateTime, value String)
ENGINE = ReplicatedMergeTree('/clickhouse/tables/event_table', '{replica}')
ORDER BY event_time PARTITION BY toYYYYMM(event_time);
INSERT INTO event_table ...然后,我们向集群中添加一个新节点:在配置文件的 remote_servers 部分中为其使用相同的条目来启动该新节点:
┌─cluster─┬─shard_num─┬─shard_weight─┬─replica_num─┬─host_name────┬─host_address─┬─port─┬─is_local─┬─user─┬─is_active─┐
│ default │ 1 │ 1 │ 1 │ 92d3c04025e8 │ 172.26.0.5 │ 9000 │ 0 │ │ ᴺᵁᴸᴸ │
│ default │ 1 │ 1 │ 2 │ a6a68731c21b │ 172.26.0.4 │ 9000 │ 1 │ │ ᴺᵁᴸᴸ │
│ default │ 1 │ 1 │ 3 │ 8e62b9cb17a1 │ 172.26.0.2 │ 9000 │ 0 │ │ ᴺᵁᴸᴸ │
│ default │ 1 │ 1 │ 4 │ b0df3669b81f │ 172.26.0.6 │ 9000 │ 0 │ │ ᴺᵁᴸᴸ │
└─────────┴───────────┴──────────────┴─────────────┴──────────────┴──────────────┴──────┴──────────┴──────┴───────────┘第四个节点已加入集群,但表 event_table 仍然只存在于前三个节点上:
SELECT hostname(), database, table FROM clusterAllReplicas(default, system.tables) WHERE table = 'event_table' FORMAT PrettyCompactMonoBlock
┌如果您需要在所有节点上复制表,可使用 Replicated 数据库引擎,作为集群发现的替代方案。