Simple Multi Node Clickhouse Cluster

I hate those single-node clickhouse clusters and manually installation, I mean, why should we:

Running multiple clickhouse instance inside one docker-compose. (e.g: tetafro/clickhouse-cluster)
Manually install JRE, download zookeeper.tar.gz, and modifying those annoying config files like a hell as some Chinese blog/book does (e,g: ClickHouse集群多实例部署, 2021年最新大数据 Clickhouse零基础教程)

this is just weird!

So this repo tries to solve these problem.

Note

This is a simplified model of Multi Node Clickhouse Cluster, which lacks: LoadBalancer config/Automated Failover/MultiShard Config generation.
All clickhouse data is persisted under event-data, if you need to move clickhouse to some other directory, you'll just need to move the directory(that contains docker-compose.yml) and docker-compose up -d to fire it up again.
Host network mode is used to simplify the whole deploy procedure, so you might need to create addition firewall rules if you are running this on a public accessible machine.

Prerequisites

To use this, we need docker and docker-compose installed, recommended OS is ubuntu, and it's recommended to install clickhouse-client on machine, so on a typical ubuntu server, doing the following should be sufficient.

apt update
curl -fsSL https://get.docker.com -o get-docker.sh && sh get-docker.sh && rm -f get-docker.sh
apt install docker-compose clickhouse-client -y

Usage

Clone this repo
Edit the necessary server info in topo.yml
Run python3 generate.py
Your cluster info should be in the cluster directory now
Sync those files to related nodes and run docker-compose up -d on them
Your cluster is ready to go

If you still cannot understand what I'm saying above, see the example below.

Example Usage

Edit information

I've Clone the repo and would like to set a 3-master clickhouse cluster and has the following specs

3 replica(one replica on each node)
1 Shard only

So I need to edit the topo.yml as follows:

global:
  clickhouse_image: "yandex/clickhouse-server:21.3.2.5"
  zookeeper_image: "bitnami/zookeeper:3.6.1"

zookeeper_servers:
  - host: 192.168.33.101
  - host: 192.168.33.102
  - host: 192.168.33.103

clickhouse_servers:
  - host: 192.168.33.101
  - host: 192.168.33.102
  - host: 192.168.33.103

clickhouse_topology:
  - clusters:
      - name: "novakwok_cluster"
        shards:
          - name: "novakwok_shard"
            servers:
              - host: 192.168.33.101
              - host: 192.168.33.102
              - host: 192.168.33.103

Generate Config

After python3 generate.py, a structure has been generated under cluster directory, looks like this:

➜  simple-multinode-clickhouse-cluster git:(master) ✗ python3 generate.py 
Write clickhouse-config.xml to cluster/192.168.33.101/clickhouse-config.xml
Write clickhouse-config.xml to cluster/192.168.33.102/clickhouse-config.xml
Write clickhouse-config.xml to cluster/192.168.33.103/clickhouse-config.xml

➜  simple-multinode-clickhouse-cluster git:(master) ✗ tree cluster 
cluster
├── 192.168.33.101
│   ├── clickhouse-config.xml
│   ├── clickhouse-user-config.xml
│   └── docker-compose.yml
├── 192.168.33.102
│   ├── clickhouse-config.xml
│   ├── clickhouse-user-config.xml
│   └── docker-compose.yml
└── 192.168.33.103
    ├── clickhouse-config.xml
    ├── clickhouse-user-config.xml
    └── docker-compose.yml

3 directories, 9 files

Sync Config

Now we need to sync those files to related hosts(of course you can use ansible here):

rsync -aP ./cluster/192.168.33.101/ [email protected]:/root/ch/
rsync -aP ./cluster/192.168.33.102/ [email protected]:/root/ch/
rsync -aP ./cluster/192.168.33.103/ [email protected]:/root/ch/

Start Cluster

Now run docker-compose up -d on every hosts' /root/ch/ directory.

Validation

On 192.168.33.101, use clickhouse-client to connect to local instance and check if cluster is there.

[email protected]:~/ch# clickhouse-client 
ClickHouse client version 18.16.1.
Connecting to localhost:9000.
Connected to ClickHouse server version 21.3.2 revision 54447.

192-168-33-101 :) 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─┐
│ novakwok_cluster                             │         1 │            1 │           1 │ 192.168.33.101 │ 192.168.33.101 │ 9000 │        1 │ default │                  │            0 │                       0 │
│ novakwok_cluster                             │         1 │            1 │           2 │ 192.168.33.102 │ 192.168.33.102 │ 9000 │        0 │ default │                  │            0 │                       0 │
│ novakwok_cluster                             │         1 │            1 │           3 │ 192.168.33.103 │ 192.168.33.103 │ 9000 │        0 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards                      │         1 │            1 │           1 │ 127.0.0.1      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards                      │         2 │            1 │           1 │ 127.0.0.2      │ 127.0.0.2      │ 9000 │        0 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards_internal_replication │         1 │            1 │           1 │ 127.0.0.1      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards_internal_replication │         2 │            1 │           1 │ 127.0.0.2      │ 127.0.0.2      │ 9000 │        0 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards_localhost            │         1 │            1 │           1 │ localhost      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_cluster_two_shards_localhost            │         2 │            1 │           1 │ localhost      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_shard_localhost                         │         1 │            1 │           1 │ localhost      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_shard_localhost_secure                  │         1 │            1 │           1 │ localhost      │ 127.0.0.1      │ 9440 │        0 │ default │                  │            0 │                       0 │
│ test_unavailable_shard                       │         1 │            1 │           1 │ localhost      │ 127.0.0.1      │ 9000 │        1 │ default │                  │            0 │                       0 │
│ test_unavailable_shard                       │         2 │            1 │           1 │ localhost      │ 127.0.0.1      │    1 │        0 │ default │                  │            0 │                       0 │
└──────────────────────────────────────────────┴───────────┴──────────────┴─────────────┴────────────────┴────────────────┴──────┴──────────┴─────────┴──────────────────┴──────────────┴─────────────────────────┘
↘ Progress: 13.00 rows, 1.58 KB (4.39 thousand rows/s., 532.47 KB/s.) 
13 rows in set. Elapsed: 0.003 sec.

Let's create a DB with replica:

192-168-33-101 :) create database novakwok_test on cluster novakwok_cluster;

CREATE DATABASE novakwok_test ON CLUSTER novakwok_cluster

┌─host───────────┬─port─┬─status─┬─error─┬─num_hosts_remaining─┬─num_hosts_active─┐
│ 192.168.33.103 │ 9000 │      0 │       │                   2 │                0 │
│ 192.168.33.101 │ 9000 │      0 │       │                   1 │                0 │
│ 192.168.33.102 │ 9000 │      0 │       │                   0 │                0 │
└────────────────┴──────┴────────┴───────┴─────────────────────┴──────────────────┘
← Progress: 3.00 rows, 174.00 B (16.07 rows/s., 931.99 B/s.)  99%
3 rows in set. Elapsed: 0.187 sec. 

192-168-33-101 :) show databases;

SHOW DATABASES

┌─name──────────┐
│ default       │
│ novakwok_test │
│ system        │
└───────────────┘
↑ Progress: 3.00 rows, 479.00 B (855.61 rows/s., 136.61 KB/s.) 
3 rows in set. Elapsed: 0.004 sec.

Connect to another host to see if it's really working.

[email protected]:~/ch# clickhouse-client -h 192.168.33.102
ClickHouse client version 18.16.1.
Connecting to 192.168.33.102:9000.
Connected to ClickHouse server version 21.3.2 revision 54447.

192-168-33-102 :) show databases;

SHOW DATABASES

┌─name──────────┐
│ default       │
│ novakwok_test │
│ system        │
└───────────────┘
↘ Progress: 3.00 rows, 479.00 B (623.17 rows/s., 99.50 KB/s.) 
3 rows in set. Elapsed: 0.005 sec.

License

GPL

Deploy a simple Multi-Node Clickhouse Cluster with docker-compose in minutes.

Related tags

Overview

Simple Multi Node Clickhouse Cluster

Note

Prerequisites

Usage

Example Usage

Edit information

Generate Config

Sync Config

Start Cluster

Validation

License

Owner

Nova Kwok

This projects provides the documentation and the automation(code) for the Oracle EMEA WLA COA Demo UseCase.

Daemon to ban hosts that cause multiple authentication errors

🐳 Docker templates for various languages.

A cron monitoring tool written in Python & Django

A tool to clone efficiently all the repos in an organization

Ralph is the CMDB / Asset Management system for data center and back office hardware.

ServerStatus 云探针、多服务器探针、云监控、多服务器云监控

Tools and Docker images to make a fast Ruby on Rails development environment

Visual disk-usage analyser for docker images

Tencent Yun tools with python

gunicorn 'Green Unicorn' is a WSGI HTTP Server for UNIX, fast clients and sleepy applications.

Kube kombu - Running kombu consumers with support of liveness probe for kubernetes

Spinnaker is an open source, multi-cloud continuous delivery platform for releasing software changes with high velocity and confidence.

Run Oracle on Kubernetes with El Carro

ZeroMQ bindings for Twisted

Dynamic DNS service

Build and Push docker image in Python (luigi + docker-py)

framework providing automatic constructions of vulnerable infrastructures

Quick & dirty controller to schedule Kubernetes Jobs later (once)

Iris is a highly configurable and flexible service for paging and messaging.