前言 最近在做 TiDB 的恢复演练，需要在单台 Linux 服务器上部署一套 TiDB 最小的完整拓扑的集群，本文记录一下安装过程。
环境准备 开始部署 TiDB 集群前，准备一台部署主机，确保其软件满足需求：
推荐安装 CentOS 7.3 及以上版本
运行环境可以支持互联网访问，用于下载 TiDB 及相关软件安装包
注意 ：TiDB 从 v8.5.1 版本起重新适配 glibc 2.17，恢复了对 CentOS Linux 7 的兼容性支持。
环境信息 最小规模的 TiDB 集群拓扑包含以下实例：
组件
数量
IP
端口配置
PD
1
192.168.31.79
2379/2380
TiDB
1
192.168.31.79
4000/10080
TiKV
3
192.168.31.79
20160-20162/20180-20182
TiFlash
1
192.168.31.79
9000/3930/20170/20292/8234/8123
Prometheus
1
192.168.31.79
9090/12020
Grafana
1
192.168.31.79
3000
安装依赖库 编译和构建 TiDB 所需的依赖库：
Golang 1.23 及以上版本
Rust nightly-2023-12-28 及以上版本
LLVM 17.0 及以上版本
sshpass 1.06 及以上
GCC 7.x（不满足）
glibc 2.28-151.el8 版本（不满足）
下载所需依赖包：
Golang 安装：
1 2 3 4 5 6 7 [root@test soft]@test ~]PATH =$PATH :/usr/local/go/bin EOF @test ~]@test ~]25.0 linux/amd64
Rust 安装：
1 2 3 4 5 [root@test soft]# tar -xf rust-1.89.0-x86_64-unknown-linux-gnu.tar.tar @test soft]# cd rust-1.89.0-x86_64-unknown-linux-gnu/ @test rust-1.89 .0 -x86_64-unknown-linux-gnu]# ./install.sh @test ~]# rustc --version 1.89 .0 (29483883 e 2025 -08 -04 )
sshpass 安装：
1 2 3 4 5 [root@test soft]# tar -xf sshpass-1.10.tar.gz @test soft]# cd sshpass-1.10/ @test sshpass-1.10 ]# ./configure && make && make install @test ~]# sshpass -V 1.10
关闭防火墙 1 2 3 4 5 6 7 [root@test ~]@test ~]@test ~]Loaded: loaded (/usr/lib /systemd/system /firewalld.service; disabled; vendor preset: enabled) Active: inactive (dead) Docs: man:firewalld(1)
检测及关闭 swap 1 2 3 4 [root@test ~] [root@test ~] [root@test ~] vm.swappiness = 0
记得修改 /etc/fstab 配置，注释掉 swap 分区：
1 #/dev/mapper/centos-swap swap swap defaults 0 0
检查和配置操作系统优化参数 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 [root@test ~]# echo never > /sys/kernel/mm/transparent_hugepage/enabled<<EOF>> /etc/sysctl.conf1000000 32768 0 0 1 <<EOF>> /etc/security /limits.conf1000000 1000000 32768 32768
调整 MaxSessions 由于模拟多机部署，需要通过 root 用户调大 sshd 服务的连接数限制：
1 2 3 [root@test ~]# vim /etc/ssh/sshd_config ## 调整 MaxSessions 20 @test ~]# systemctl restart sshd.service
创建 TiDB 用户 1 2 3 4 5 6 7 [root@test ~ ]# useradd tidb@test ~ ]# echo "Tidb@123" | passwd tidb for user tidb.all authentication tokens updated successfully.@test ~ ]# cat<< - EOF>> / etc/ sudoersALL = (ALL ) NOPASSWD: ALL
实施部署 本文是内网环境，不使用官方在线源安装，使用本地镜像源进行部署，本地镜像源部署请参考：TiDB 离线部署 TiUP 组件 。
tiup 已部署完成：
1 2 3 4 5 6 7 8 [root@test ~]/root/tidb-community-server-v8.5.3-linux-amd64 1.16 .2 tiupVersion: go1.21.13 Ref: v1.16.2 GitHash: 678 c52de0c0ef30634b8ba7302a8376caa95d50d
创建并启动集群：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 [root@test ~]global: user: "tidb" ssh_port: 11122 deploy_dir: "/data/tidb-deploy" data_dir: "/data/tidb-data" monitored: node_exporter_port: 9100 blackbox_exporter_port: 9115 server_configs: tidb: tidb_slow_log_threshold: 300 tikv: use-unified-pool: false use-unified-pool: true pd: enable-placement-rules: true location-labels: ["host" ]tiflash: level: "info" pd_servers: - host: 192.168 .31.79 tidb_servers: - host: 192.168 .31.79 tikv_servers: - host: 192.168 .31.79 port: 20160 status_port: 20180 config: labels: { host: "logic-host-1" }- host: 192.168 .31.79 port: 20161 status_port: 20181 config: labels: { host: "logic-host-2" }- host: 192.168 .31.79 port: 20162 status_port: 20182 config: labels: { host: "logic-host-3" }tiflash_servers: - host: 192.168 .31.79 monitoring_servers: - host: 192.168 .31.79 grafana_servers: - host: 192.168 .31.79
安装前预检查：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 [root@test ~] # tiup cluster check topo.yaml --user root -p192 .168 .31 .79 Arch info ... Done192 .168 .31 .79 OS info ... Done192 .168 .31 .79 :11122 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 ... Done192 .168 .31 .79 :11122 ... Done192.168.31.79 os-version Fail CentOS Linux 7 (Core) 7 .9 .2009 not supported, use version 9 or higher192.168.31.79 cpu-cores Pass number of CPU cores / threads: 4 192.168.31.79 ntp Warn The NTPd daemon may be not start192.168.31.79 disk Warn mount point /data does not have 'noatime' option set192.168.31.79 selinux Pass SELinux is disabled192.168.31.79 thp Pass THP is disabled192.168.31.79 command Pass numactl: policy: default192.168.31.79 cpu-governor Warn Unable to determine current CPU frequency governor policy192.168.31.79 memory Pass memory size is 8192MB192.168.31.79 network Pass network speed of ens192 is 10000MB192.168.31.79 disk Fail multiple components tikv:/data/tidb-data/tikv-20160 ,tikv:/data/tidb-data/tikv-20161 ,tikv:/data/tidb-data/tikv-20162 ,tiflash:/data/tidb-data/tiflash-9000 are using the same partition 192 .168 .31 .79 :/data as data dir192.168.31.79 disk Fail mount point /data does not have 'nodelalloc' option set
部署集群：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 [root@test ~] not what you expected, check your yaml file. instance instance configs monitor configs instance 192.168.31.79:2379 instance 192.168.31.79:2379 success instance 192.168.31.79:20162 instance 192.168.31.79:20160 instance 192.168.31.79:20161 instance 192.168.31.79:20162 success instance 192.168.31.79:20161 success instance 192.168.31.79:20160 success instance 192.168.31.79:4000 instance 192.168.31.79:4000 success instance 192.168.31.79:9000 instance 192.168.31.79:9000 success instance 192.168.31.79:9090 instance 192.168.31.79:9090 success instance 192.168.31.79:3000 instance 192.168.31.79:3000 success instance 192.168.31.79 instance 192.168.31.79
启动集群：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 [root@test ~] private Key=/root/.tiup/storage/cluster/clusters/lucifer/ssh/id_rsa, public Key=/root/.tiup/storage/cluster/clusters/lucifer/ssh/id_rsa.pub instance 192.168.31.79:2379 instance 192.168.31.79:2379 success instance 192.168.31.79:20162 instance 192.168.31.79:20160 instance 192.168.31.79:20161 instance 192.168.31.79:20162 success instance 192.168.31.79:20161 success instance 192.168.31.79:20160 success instance 192.168.31.79:4000 instance 192.168.31.79:4000 success instance 192.168.31.79:9000 instance 192.168.31.79:9000 success instance 192.168.31.79:9090 instance 192.168.31.79:9090 success instance 192.168.31.79:3000 instance 192.168.31.79:3000 success instance 192.168.31.79 instance 192.168.31.79 new password is: 'm+92G0Q3eNR4^6cq*@'. and record it to somewhere safe, it is only displayed once, and will not be stored. and shown again.
查看集群：
1 2 3 4 [root@test ~]# tiup cluster listName User Version Path PrivateKey.5 .3 /root/.tiup/storage /cluster /clusters/lucifer /root/.tiup/storage /cluster /clusters/lucifer/ssh/id_rsa
检查集群状态：
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 [root@test ~]type: tidbname: luciferversion: v8.5.3 user: tidbtype: builtinURL: http:// 192.168 .31.79 :2379 /dashboard URLs: http:// 192.168 .31.79 :2379 /dashboard URL: http:// 192.168 .31.79 :3000 /Arch Status Data Dir Deploy Dir- - - --- - --- - ---- - ------ - ----- - ------- - ---------192.168 .31.79 :3000 grafana 192.168 .31.79 3000 linux/x86_64 Up - /data/tidb-deploy/grafana-3000 192.168 .31.79 :2379 pd 192.168 .31.79 2379 /2380 linux/x86_64 Up|L|UI /data/tidb-data/pd-2379 /data/tidb-deploy/pd-2379 192.168 .31.79 :9090 prometheus 192.168 .31.79 9090 /12020 linux/x86_64 Up /data/tidb-data/prometheus-9090 /data/tidb-deploy/prometheus-9090 192.168 .31.79 :4000 tidb 192.168 .31.79 4000 /10080 linux/x86_64 Up - /data/tidb-deploy/tidb-4000 192.168 .31.79 :9000 tiflash 192.168 .31.79 9000 /3930/20170/20292/8234/8123 linux/x86_64 Up /data/tidb-data/tiflash-9000 /data/tidb-deploy/tiflash-9000 192.168 .31.79 :20160 tikv 192.168 .31.79 20160 /20180 linux/x86_64 Up /data/tidb-data/tikv-20160 /data/tidb-deploy/tikv-20160 192.168 .31.79 :20161 tikv 192.168 .31.79 20161 /20181 linux/x86_64 Up /data/tidb-data/tikv-20161 /data/tidb-deploy/tikv-20161 192.168 .31.79 :20162 tikv 192.168 .31.79 20162 /20182 linux/x86_64 Up /data/tidb-data/tikv-20162 /data/tidb-deploy/tikv-20162 nodes: 8
安装 MySQL 客户端 TiDB 兼容 MySQL 协议，故需要 MySQL 客户端连接，则需安装 MySQL 客户端，Linux7 版本的系统默认自带安装了 MariaDB，需要先清理：
1 [root@test ~] # rpm -e --nodeps $(rpm -qa | grep mariadb)
找个有网的环境下载：
1 2 [root@lucifer ~]# wget https :~]# wget http :
安装 MySQL 客户端：
1 2 3 [root@test ~]# yum -y install mysql80-community-release-el7-10.noarch.rpm @test ~]# rpm --import RPM-GPG-KEY-mysql-2023 @test ~]# yum -y install mysql
连接数据库：
1 2 3 ## 这里的 root 初始密码在 tidb 集群初始化时日志中输出的密码 m+92G0Q3eNR4^6cq*@ @test ~]# mysql -h 192.168.31.79 -P 4000 -uroot –p
修改初始 root 密码：
1 2 mysql> use mysqlalter user 'root' @'%' identified by 'tidb' ;
集群监控：
写在最后 至此，TiDB 单机集群部署完成，可用于开发测试和学习研究。生产环境建议参考官方推荐的多机部署方案。
