作者: JiekeXu

TiDB 是平凯星辰 PingCAP 公司自主设计、研发的开源分布式关系型数据库,是一款同时支持在线事务处理与在线分析处理 (Hybrid Transactional and Analytical Processing, HTAP) 的融合型分布式数据库产品,具备水平扩容或者缩容、金融级高可用、实时 HTAP、云原生的分布式数据库、兼容 MySQL 协议和 MySQL 生态等重要特性。

PingCAP 成立于 2015 年,是一家企业级开源分布式数据库厂商,提供包括开源分布式数据库产品、解决方案与咨询、技术支持与培训认证服务,致力于为全球行业用户提供稳定高效、安全可靠、开放兼容的新型数据服务平台,解放企业生产力,加速企业数字化转型升级。目标是为用户提供一站式 OLTP (Online Transactional Processing)、OLAP (Online Analytical Processing)、HTAP 解决方案。TiDB 适合高可用、强一致要求较高、数据规模较大等各种应用场景。本月在墨天轮排行榜上继续保持第四名,上个月末 29 号 TiDB 也发布了一个最新的版本 8.0.0 版本,当然这个版本是 DMR(Development Milestone Releases) 版本

 

TiDB 8.0 新特性尝鲜_z



版本规则介绍

TiDB 版本的命名方式为 X.Y.Z。X.Y 代表一个版本系列。

· 从 TiDB 1.0 起,X 每年依次递增,X 的递增代表引入新的功能和改进。

· Y 从 0 开始依次递增,Y 的递增代表引入新的功能和改进。

· 一个版本系列首次发版时 Z 默认为 0,后续发补丁版本时 Z 从 1 开始依次递增。

TiDB 8.0 新特性尝鲜_tidb_02



版本支持时间表

TiDB LTS 版本的生命维护周期定义如下:

· Maintenance Support:针对 LTS 版本,自第一个 Release 版本发版日期开始计算,PingCAP 会提供最多至 3 年时间的常规版本更新,以解决版本运行过程中遇到的问题,以及安全相关的漏洞修复。

· Extended Support:在常规版本维护周期结束之后,PingCAP 会提供 1 年的额外支持服务,具体服务内容包括:

o PingCAP 仅提供严重级别的安全漏洞修复。

o PingCAP 不保证会将所有的补丁合并回当前的主干版本。

o PingCAP 官方技术支持团队会向客户提供已有的补丁或解决方案。如果没有相关方案,那么会指导并协助客户升级到已经解决该问题的主干版本。

· End of Life (EOL): 在 Extended 周期结束之后,PingCAP 就不对该版本提供修复及技术支持服务。

在 EOL 之后,针对有战略合作关系客户的关键应用,PingCAP 可视情况提供额外延长的支持服务,具体请联系对应的客户经理。在该阶段,仅提供标准的技术服务接口,不会有任何的代码级的修复。

注:TiDB DMR 版本仅供开发或测试使用,PingCAP 不提供相关技术支持。

 

TiDB 8.0 新特性尝鲜_tiflash_03



8.0 新特性简介

2024 年 3 月 29 日,TiDB 发布 8.0.0 DMR 版本,你可以获得以下关键特性,这里我将新特性分为六大板块,进行简单描述,不在一一展开说明,如果你感兴趣,可去官网查看相关内容:https://docs.pingcap.com/zh/tidb/v8.0/release-8.0.0

TiDB 8.0 新特性尝鲜_tidb_04

TiDB 这里也不是第一次介绍了,最开始接触 3.0 版本,到后面的 4.0 版本,以及21 年4月发布的TiDB 5.0 提供了一个 playground 的命令可以在个人笔记本上使用一台 Linux 机器模拟搭建一个集群环境,才算真真正正用起来了 TiDB。对于这种分布式架构和组件前面也已经介绍过了,今天这里就不介绍了,感兴趣的可查看下面相关文章,今天我们快速部署一个 8.0 的单机分布式的体验环境,体验一下部分新特性即可。

快速体验 TiDB 5.0 https://www.modb.pro/db/75269

TiDB 5.4 单机快速安装初体验 https://www.modb.pro/db/336159

分布式数据库 TiDB 6.0 集群保姆级安装手册 https://www.modb.pro/db/420603

TiDB 之 TiCDC6.0 初体验 https://www.modb.pro/db/427262** **



快速上手体验 8.0

TiDB 是一个分布式系统。最基础的 TiDB 测试集群通常由 2 个 TiDB 实例、3 个 TiKV 实例、3 个 PD 实例和可选的 TiFlash 实例构成。只要你的主机可以联网,便可通过单台 Linux 服务器,体验 TiDB 最小的完整拓扑的集群,并模拟生产环境下的部署步骤,可以快速搭建出一套基础测试集群,步骤如下:



1)最小规模拓扑结构

实例

个数

IP

配置

TiKV

3

192.168.75.88192.168.75.88192.168.75.88

避免端口和目录冲突

TiDB

1

192.168.75.88

默认端口全局目录配置

PD

1

192.168.75.88

默认端口全局目录配置

TiFlash

1

192.168.75.88

默认端口全局目录配置

Monitor

1

192.168.75.88

默认端口全局目录配置


2)下载并安装 TiUP

虚拟机使用 NAT 模式能访问互联网即可。

[root@JiekeXu ~]# ping www.baidu.com
PING www.a.shifen.com (220.181.38.149) 56(84) bytes of data.
64 bytes from 220.181.38.149 (220.181.38.149): icmp_seq=1 ttl=128 time=7.26 ms
64 bytes from 220.181.38.149 (220.181.38.149): icmp_seq=2 ttl=128 time=7.35 ms
64 bytes from 220.181.38.149 (220.181.38.149): icmp_seq=3 ttl=128 time=9.07 ms
64 bytes from 220.181.38.149 (220.181.38.149): icmp_seq=4 ttl=128 time=5.94 ms
64 bytes from 220.181.38.149 (220.181.38.149): icmp_seq=5 ttl=128 time=6.99 ms
^C
--- www.a.shifen.com ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 4001ms
rtt min/avg/max/mdev = 5.948/7.329/9.078/1.009 ms

curl --proto '=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh

[root@JiekeXu ~]#  curl --proto '=https' --tlsv1.2 -sSf https://tiup-mirrors.pingcap.com/install.sh | sh
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 5095k  100 5095k    0     0  7306k      0 --:--:-- --:--:-- --:--:-- 7310k
WARN: adding root certificate via internet: https://tiup-mirrors.pingcap.com/root.json
You can revoke this by remove /root/.tiup/bin/7b8e153f2e2d0928.root.json
Successfully set mirror to https://tiup-mirrors.pingcap.com
Detected shell: bash
Shell profile:  /root/.bash_profile
Installed path: /root/.tiup/bin/tiup
===============================================
Have a try:     tiup playground
===============================================
[root@JiekeXu ~]# source /root/.bash_profile

TiDB 8.0 新特性尝鲜_z_05



3)安装 TiUP 的 cluster 组件

tiup cluster

运行结果如下图:

[root@JiekeXu ~]# tiup cluster
Checking updates for component cluster...
A new version of cluster is available:  -> v1.15.0
 
    To update this component:   tiup update cluster
    To update all components:   tiup update --all
 
The component `cluster` version  is not installed; downloading from repository.
download https://tiup-mirrors.pingcap.com/cluster-v1.15.0-linux-amd64.tar.gz 8.76 MiB / 8.76 MiB 100.00% 41.14 MiB/s     
Deploy a TiDB cluster for production
 
Usage:
  tiup cluster [command]
 
Available Commands:
  check       Perform preflight checks for the cluster.
  deploy      Deploy a cluster for production
  start       Start a TiDB cluster
……
Use "tiup cluster help [command]" for more information about a command.
[root@JiekeXu ~]#

TiDB 8.0 新特性尝鲜_z_06



4)如果机器已经安装 TiUP cluster,需要更新软件版本

tiup update --self && tiup update cluster

[root@JiekeXu ~]# tiup update --self && tiup update cluster
download https://tiup-mirrors.pingcap.com/tiup-v1.15.0-linux-amd64.tar.gz 4.98 MiB / 4.98 MiB 100.00% 331.47 MiB/s       
Updated successfully!
component cluster version v1.15.0 is already installed
Updated successfully!



5)调大 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. 修改 /etc/ssh/sshd_config 将 MaxSessions 调至 20,并去掉注释。

2. 重启 sshd 服务:service sshd restart



6)创建并启动集群

按下面的配置模板,编辑配置文件,命名为 topo.yaml,其中:

· user: "tidb":表示通过 tidb 系统用户(部署会自动创建)来做集群的内部管理,默认使用 22 端口通过 ssh 登录目标机器

· replication.enable-placement-rules:设置这个 PD 参数来确保 TiFlash 正常运行

· host:设置为本部署主机的 IP

· deploy_dir 和 data_dir:软件目录和数据目录,这里我加了一级目录/tidb

配置模板如下,大家可修改 IP 为自己机器 IP,也可自定义端口和节点个数。

vi topo.yaml

# # Global variables are applied to all deployments and used as the default value of
# # the deployments if a specific deployment value is missing.
global:
 user: "tidb"
 ssh_port: 22
 deploy_dir: "/app/tidb-deploy"
 data_dir: "/app/tidb-data"

# # Monitored variables are applied to all the machines.
monitored:
 node_exporter_port: 9100
 blackbox_exporter_port: 9115

server_configs:
 tidb:
   instance.tidb_slow_log_threshold: 300
 tikv:
   readpool.storage.use-unified-pool: false
   readpool.coprocessor.use-unified-pool: true
 pd:
   replication.enable-placement-rules: true
   replication.location-labels: ["host"]
 tiflash:
   logger.level: "info"

pd_servers:
 - host: 192.168.75.88

tidb_servers:
 - host: 192.168.75.88

tikv_servers:
 - host: 192.168.75.88
   port: 20160
   status_port: 20180
   config:
     server.labels: { host: "logic-host-1" }

 - host: 192.168.75.88
   port: 20161
   status_port: 20181
   config:
     server.labels: { host: "logic-host-2" }

 - host: 192.168.75.88
   port: 20162
   status_port: 20182
   config:
     server.labels: { host: "logic-host-3" }

tiflash_servers:
 - host: 192.168.75.88

monitoring_servers:
 - host: 192.168.75.88

grafana_servers:
 - host: 192.168.75.88



7)执行集群部署命令

tiup cluster deploy <cluster-name> <version> ./topo.yaml --user root -p

tiup cluster deploy JiekeXu 8.0.0 ./topo.yaml --user root -p

·  参数 <cluster-name> 表示设置集群名称

·  参数 <version> 表示设置集群版本,例如 v8.0.0。可以通过 tiup list tidb 命令来查看当前支持部署的 TiDB 版本

·  参数 -p 表示在连接目标机器时使用密码登录

TiDB 8.0 新特性尝鲜_https_07

TiDB 8.0 新特性尝鲜_tiflash_08

[root@JiekeXu ~]# tiup cluster deploy JiekeXu 8.0.0 ./topo.yaml --user root -p
Input SSH password: 

+ Detect CPU Arch Name
  - Detecting node 192.168.75.88 Arch info ... Done

+ Detect CPU OS Name
  - Detecting node 192.168.75.88 OS info ... Done
Please confirm your topology:
Cluster type:    tidb
Cluster name:    JiekeXu
Cluster version: v8.0.0
Role        Host            Ports                            OS/Arch       Directories
----        ----            -----                            -------       -----------
pd          192.168.75.88  2379/2380                        linux/x86_64  /tidb/tidb-deploy/pd-2379,/tidb/tidb-data/pd-2379
tikv        192.168.75.88  20160/20180                      linux/x86_64  /tidb/tidb-deploy/tikv-20160,/tidb/tidb-data/tikv-20160
tikv        192.168.75.88  20161/20181                      linux/x86_64  /tidb/tidb-deploy/tikv-20161,/tidb/tidb-data/tikv-20161
tikv        192.168.75.88  20162/20182                      linux/x86_64  /tidb/tidb-deploy/tikv-20162,/tidb/tidb-data/tikv-20162
tidb        192.168.75.88  4000/10080                       linux/x86_64  /tidb/tidb-deploy/tidb-4000
tiflash     192.168.75.88  9000/8123/3930/20170/20292/8234  linux/x86_64  /tidb/tidb-deploy/tiflash-9000,/tidb/tidb-data/tiflash-9000
prometheus  192.168.75.88  9090/12020                       linux/x86_64  /tidb/tidb-deploy/prometheus-9090,/tidb/tidb-data/prometheus-9090
grafana     192.168.75.88  3000                             linux/x86_64  /tidb/tidb-deploy/grafana-3000
Attention:
    1. If the topology is not what you expected, check your yaml file.
    2. Please confirm there is no port/directory conflicts in same host.
Do you want to continue? [y/N]: (default=N) y
+ Generate SSH keys ... Done
+ Download TiDB components
  - Download pd:v8.0.0 (linux/amd64) ... Done
  - Download tikv:v8.0.0 (linux/amd64) ... Done
  - Download tidb:v8.0.0 (linux/amd64) ... Done
  - Download tiflash:v8.0.0 (linux/amd64) ... Done
  - Download prometheus:v8.0.0 (linux/amd64) ... Done
  - Download grafana:v8.0.0 (linux/amd64) ... Done
  - Download node_exporter: (linux/amd64) ... Done
  - Download blackbox_exporter: (linux/amd64) ... Done
+ Initialize target host environments
  - Prepare 192.168.75.88:22 ... Done
+ Deploy TiDB instance
  - Copy pd -> 192.168.75.88 ... Done
  - Copy tikv -> 192.168.75.88 ... Done
  - Copy tikv -> 192.168.75.88 ... Done
  - Copy tikv -> 192.168.75.88 ... Done
  - Copy tidb -> 192.168.75.88 ... Done
  - Copy tiflash -> 192.168.75.88 ... Done
  - Copy prometheus -> 192.168.75.88 ... Done
  - Copy grafana -> 192.168.75.88 ... Done
  - Deploy node_exporter -> 192.168.75.88 ... Done
  - Deploy blackbox_exporter -> 192.168.75.88 ... Done
+ Copy certificate to remote host
+ Init instance configs
  - Generate config pd -> 192.168.75.88:2379 ... Done
  - Generate config tikv -> 192.168.75.88:20160 ... Done
  - Generate config tikv -> 192.168.75.88:20161 ... Done
  - Generate config tikv -> 192.168.75.88:20162 ... Done
  - Generate config tidb -> 192.168.75.88:4000 ... Done
  - Generate config tiflash -> 192.168.75.88:9000 ... Done
  - Generate config prometheus -> 192.168.75.88:9090 ... Done
  - Generate config grafana -> 192.168.75.88:3000 ... Done
+ Init monitor configs
  - Generate config node_exporter -> 192.168.75.88 ... Done
  - Generate config blackbox_exporter -> 192.168.75.88 ... Done
Enabling component pd
        Enabling instance 192.168.75.88:2379
        Enable instance 192.168.75.88:2379 success
Enabling component tikv
        Enabling instance 192.168.75.88:20162
        Enabling instance 192.168.75.88:20161
        Enabling instance 192.168.75.88:20160
        Enable instance 192.168.75.88:20162 success
        Enable instance 192.168.75.88:20161 success
        Enable instance 192.168.75.88:20160 success
Enabling component tidb
        Enabling instance 192.168.75.88:4000
        Enable instance 192.168.75.88:4000 success
Enabling component tiflash
        Enabling instance 192.168.75.88:9000
        Enable instance 192.168.75.88:9000 success
Enabling component prometheus
        Enabling instance 192.168.75.88:9090
        Enable instance 192.168.75.88:9090 success
Enabling component grafana
        Enabling instance 192.168.75.88:3000
        Enable instance 192.168.75.88:3000 success
Enabling component node_exporter
        Enabling instance 192.168.75.88
        Enable 192.168.75.88 success
Enabling component blackbox_exporter
        Enabling instance 192.168.75.88
        Enable 192.168.75.88 success
Cluster `JiekeXu` deployed successfully, you can start it with command: `tiup cluster start JiekeXu --init`
[root@JiekeXu ~]#



8)启动集群

tiup cluster start JiekeXu --init

[root@JiekeXu ~]# tiup cluster start JiekeXu --init
Starting cluster JiekeXu...
+ [ Serial ] - SSHKeySet: privateKey=/root/.tiup/storage/cluster/clusters/JiekeXu/ssh/id_rsa, publicKey=/root/.tiup/storage/cluster/clusters/JiekeXu/ssh/id_rsa.pub
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [Parallel] - UserSSH: user=tidb, host=192.168.75.88
+ [ Serial ] - StartCluster
Starting component pd
        Starting instance 192.168.75.88:2379
        Start instance 192.168.75.88:2379 success
Starting component tikv
        Starting instance 192.168.75.88:20162
        Starting instance 192.168.75.88:20160
        Starting instance 192.168.75.88:20161
        Start instance 192.168.75.88:20160 success
        Start instance 192.168.75.88:20162 success
        Start instance 192.168.75.88:20161 success
Starting component tidb
        Starting instance 192.168.75.88:4000
        Start instance 192.168.75.88:4000 success
Starting component tiflash
        Starting instance 192.168.75.88:9000
        Start instance 192.168.75.88:9000 success
Starting component prometheus
        Starting instance 192.168.75.88:9090
        Start instance 192.168.75.88:9090 success
Starting component grafana
        Starting instance 192.168.75.88:3000
        Start instance 192.168.75.88:3000 success
Starting component node_exporter
        Starting instance 192.168.75.88
        Start 192.168.75.88 success
Starting component blackbox_exporter
        Starting instance 192.168.75.88
        Start 192.168.75.88 success
+ [ Serial ] - UpdateTopology: cluster=JiekeXu
Started cluster `JiekeXu` successfully
The root password of TiDB database has been changed.
The new password is: '+^Z_021Ywm7K@93naU'.
Copy and record it to somewhere safe, it is only displayed once, and will not be stored.
The generated password can NOT be get and shown again.
[root@JiekeXu ~]#

TiDB 8.0 新特性尝鲜_z_09

** 注意:1、上图中黄色密码需要记住,'+^Z_021Ywm7K@93naU' 这是登录数据库的初始密码,也是访问 TiDB 的 Dashboard 的密码;2、如果你的主机还有其他用途,安装了其他数据库,需要避免端口占用情况。我这里安装了其他数据库占用了 2380 端口,导致启动 TiKV 失败了。**

TiDB 8.0 新特性尝鲜_tidb_10

启动集群操作会按 PD -> TiKV -> Pump -> TiDB -> TiFlash -> Drainer -> TiCDC -> Prometheus -> Grafana -> Alertmanager 的顺序启动整个 TiDB 集群所有组件;该命令支持通过 -R 和 -N 参数来只启动部分组件。

例如,下列命令只启动 PD 组件:

tiup cluster start ${cluster-name} -R pd

下列命令只启动 10.10.3.4 和 10.10.3.5 这两台机器上的 PD 组件:

tiup cluster start ${cluster-name} -N 10.10.3.4:2379,10.10.3.5:2379

关闭集群操作会按 Alertmanager -> Grafana -> Prometheus -> TiCDC -> Drainer -> TiFlash -> TiDB -> Pump -> TiKV -> PD 的顺序关闭整个 TiDB 集群所有组件(同时也会关闭监控组件):

tiup cluster stop ${cluster-name}

stop 命令也支持通过 -R 和 -N 参数来只停止部分组件。



9) 查看集群状态

tiup cluster display JiekeXu

[root@JiekeXu ~]# tiup cluster display JiekeXu
Cluster type:       tidb
Cluster name:       JiekeXu
Cluster version:    v8.0.0
Deploy user:        tidb
SSH type:           builtin
Dashboard URL:      http://192.168.75.88:2379/dashboard
Grafana URL:        http://192.168.75.88:3000
ID                   Role        Host           Ports                            OS/Arch       Status   Data Dir                        Deploy Dir
--                   ----        ----           -----                            -------       ------   --------                        ----------
192.168.75.88:3000   grafana     192.168.75.88  3000                             linux/x86_64  Up       -                               /app/tidb-deploy/grafana-3000
192.168.75.88:2379   pd          192.168.75.88  2379/2380                        linux/x86_64  Up|L|UI  /app/tidb-data/pd-2379          /app/tidb-deploy/pd-2379
192.168.75.88:9090   prometheus  192.168.75.88  9090/12020                       linux/x86_64  Up       /app/tidb-data/prometheus-9090  /app/tidb-deploy/prometheus-9090
192.168.75.88:4000   tidb        192.168.75.88  4000/10080                       linux/x86_64  Up       -                               /app/tidb-deploy/tidb-4000
192.168.75.88:9000   tiflash     192.168.75.88  9000/8123/3930/20170/20292/8234  linux/x86_64  Up       /app/tidb-data/tiflash-9000     /app/tidb-deploy/tiflash-9000
192.168.75.88:20160  tikv        192.168.75.88  20160/20180                      linux/x86_64  Up       /app/tidb-data/tikv-20160       /app/tidb-deploy/tikv-20160
192.168.75.88:20161  tikv        192.168.75.88  20161/20181                      linux/x86_64  Up       /app/tidb-data/tikv-20161       /app/tidb-deploy/tikv-20161
192.168.75.88:20162  tikv        192.168.75.88  20162/20182                      linux/x86_64  Up       /app/tidb-data/tikv-20162       /app/tidb-deploy/tikv-20162
Total nodes: 8

TiDB 8.0 新特性尝鲜_z_11



10)访问 TiDB 集群

执行以下命令确认当前已经部署的集群列表

[root@JiekeXu ~]# tiup cluster list
Name     User  Version  Path                                          PrivateKey
----     ----  -------  ----                                          ----------
JiekeXu  tidb  v8.0.0   /root/.tiup/storage/cluster/clusters/JiekeXu  /root/.tiup/storage/cluster/clusters/JiekeXu/ssh/id_rsa

·  访问 TiDB 的 Grafana 监控

通过 http://192.168.75.88:3000 访问集群 Grafana 监控页面,默认用户名和密码均为 admin。首次登录之后强制修改密码,我这里将其修改为 TiDB8.0

TiDB 8.0 新特性尝鲜_tikv_12

 ·  访问 TiDB 的 Dashboard

通过 http://192.168.75.88:2379/dashboard 访问集群 TiDB Dashboard 监控页面,默认用户名为 root,密码为上面启动时打印在屏幕上的密码 +^Z_021Ywm7K@93naU

The new password is: '+^Z_021Ywm7K@93naU'.

TiDB 8.0 新特性尝鲜_tidb_13

 

由于 TiDB 兼容 MySQL,所以可以通过 MySQL 客户端来访问。这里先安装一个 MySQL 命令行客户端工具访问 TiDB。

[root@JiekeXu log]# yum install mysql -y
Loaded plugins: fastestmirror, langpacks
Loading mirror speeds from cached hostfile
 * base: mirrors.huaweicloud.com
 * epel: mirrors.bfsu.edu.cn
 * extras: mirrors.huaweicloud.com
 * updates: mirrors.huaweicloud.com
file:///mnt/centos/repodata/repomd.xml: [Errno 14] curl#37 - "Couldn't open file /mnt/centos/repodata/repomd.xml"
Trying other mirror.
Resolving Dependencies
--> Running transaction check
---> Package mariadb.x86_64 1:5.5.68-1.el7 will be installed
--> Processing Dependency: mariadb-libs(x86-64) = 1:5.5.68-1.el7 for package: 1:mariadb-5.5.68-1.el7.x86_64
--> Running transaction check
---> Package mariadb-libs.x86_64 1:5.5.60-1.el7_5 will be updated
---> Package mariadb-libs.x86_64 1:5.5.68-1.el7 will be an update
--> Finished Dependency Resolution
 
Dependencies Resolved
 
=============================================================================================================================
 Package                         Arch                      Version                             Repository               Size
=============================================================================================================================
Installing:
 mariadb                         x86_64                    1:5.5.68-1.el7                      base                    8.8 M
Updating for dependencies:
 mariadb-libs                    x86_64                    1:5.5.68-1.el7                      base                    760 k
 
Transaction Summary
=============================================================================================================================
Install  1 Package
Upgrade             ( 1 Dependent package)
 
Total size: 9.5 M
Total download size: 8.8 M
Downloading packages:
mariadb-5.5.68-1.el7.x86_64.rpm                                                                       | 8.8 MB  00:00:00     
Running transaction check
Running transaction test
Transaction test succeeded
Running transaction
Warning: RPMDB altered outside of yum.
  Updating   : 1:mariadb-libs-5.5.68-1.el7.x86_64                                                                        1/3
  Installing : 1:mariadb-5.5.68-1.el7.x86_64                                                                             2/3
  Cleanup    : 1:mariadb-libs-5.5.60-1.el7_5.x86_64                                                                      3/3
  Verifying  : 1:mariadb-libs-5.5.68-1.el7.x86_64                                                                        1/3
  Verifying  : 1:mariadb-5.5.68-1.el7.x86_64                                                                             2/3
  Verifying  : 1:mariadb-libs-5.5.60-1.el7_5.x86_64                                                                      3/3
 
Installed:
  mariadb.x86_64 1:5.5.68-1.el7                                                                                              
 
Dependency Updated:
  mariadb-libs.x86_64 1:5.5.68-1.el7                                                                                         
 
Complete!
[root@JiekeXu log]#

 

TiDB 8.0 新特性尝鲜_tikv_14

访问 TiDB 数据库,密码为上面第九步屏幕上打印出来的复杂密码:

mysql -h 192.168.75.88 -P 4000 -u root -p

mysql -h 192.168.75.88 -P 4000 -u root -p+^Z_021Ywm7K@93naU

TiDB 8.0 新特性尝鲜_tikv_15

MySQL [(none)]> create database JiekeXu;
Query OK, 0 rows affected (0.24 sec)
 
MySQL [(none)]> show databases;
+--------------------+
| Database           |
+--------------------+
| INFORMATION_SCHEMA |
| JiekeXu            |
| METRICS_SCHEMA     |
| PERFORMANCE_SCHEMA |
| mysql              |
| sys                |
| test               |
+--------------------+
7 rows in set (0.00 sec)
 
MySQL [(none)]> use JiekeXu;
Database changed
MySQL [JiekeXu]>

当然我们也可以使用 navicat Permium 16 去连接查询。

 

TiDB 8.0 新特性尝鲜_tidb_16

 接下来就可以体验V8.0 的新特性了,比如观测索引使用情况,在会话级将不可见索引 (Invisible Indexes) 调整为可见,建表时使用更多的表达式设置列的默认值等等新特性。

TiDB 8.0 新特性尝鲜_tiflash_17

默认情况下,不可见索引 (Invisible Indexes) 不会被查询优化器使用:

CREATE TABLE t1 (c1 INT, c2 INT, UNIQUE(c2));
INSERT INTO T1 VALUES(1,1),(2,2),(3,3);
CREATE UNIQUE INDEX c1 ON t1 (c1) INVISIBLE;
select * from t1 where c1=1;
explain select * from t1 where c1=1;

从 TiDB v8.0.0 开始,你可以通过修改系统变量 tidb_opt_use_invisible_indexes,默认值为 OFF,当修改变量为 ON 时,对该会话中的查询,优化器可以选择不可见索引进行查询优化。

TiDB 8.0 新特性尝鲜_tiflash_18

MySQL [INFORMATION_SCHEMA]> use JiekeXu;
Database changed
MySQL [JiekeXu]> CREATE TABLE JiekeXu (blog BLOB DEFAULT (RAND()));
Query OK, 0 rows affected (0.18 sec)
 
MySQL [JiekeXu]>

这样,我们就可以去体验测试新特性了,由于时间关系,今天就先体验到这里吧,剩下的新特性后面有机会的话在慢慢发掘。如果觉得这个环境太占你的磁盘空间的话,那么我们也可以清理日志,甚至清理数据。



清除集群数据

此操作会关闭所有服务,并清空其数据目录或/和日志目录,并且无法恢复,需要谨慎操作

清空集群所有服务的数据,但保留日志:

tiup cluster clean ${cluster-name} --data

清空集群所有服务的日志,但保留数据:

tiup cluster clean ${cluster-name} --log

清空集群所有服务的数据和日志:

tiup cluster clean ${cluster-name} --all

清空 Prometheus 以外的所有服务的日志和数据:

tiup cluster clean ${cluster-name} --all --ignore-role prometheus

清空节点 172.16.13.11:9000 以外的所有服务的日志和数据:

tiup cluster clean ${cluster-name} --all --ignore-node 172.16.13.11:9000

清空部署在 172.16.13.12 以外的所有服务的日志和数据:

tiup cluster clean ${cluster-name} --all --ignore-node 172.16.13.12



销毁集群

销毁集群操作会关闭服务,清空数据目录和部署目录,并且无法恢复,需要谨慎操作

tiup cluster destroy ${cluster-name}



附:2380端口占用处理办法

启动 TiKV 报错

Starting component pd
        Starting instance 192.168.75.88:2379
        Start instance 192.168.75.88:2379 success
Starting component tikv
        Starting instance 192.168.75.88:20162
        Starting instance 192.168.75.88:20160
        Starting instance 192.168.75.88:20161

Error: failed to start tikv: failed to start: 192.168.75.88 tikv-20161.service, please check the instance's log(/app/tidb-deploy/tikv-20161/log) for more detail.: timed out waiting for port 20161 to be started after 2m0s

Verbose debug logs has been written to /root/.tiup/logs/tiup-cluster-debug-2024-04-06-23-44-47.log.

** 查看 TiKV 日志**

/app/tidb-deploy/tikv-20161/log,发现 Connection refused 192.168.75.88:2379,这是 PD 的端口地址,那么我们去看 PD 的启动日志吧。

[2024/04/06 23:50:49.888 +08:00] [INFO] [<unknown>] ["subchannel 0x7f5f62c56c00 {address=ipv4:192.168.75.88:2379, args=grpc.c
lient_channel_factory=0x7f5f62c978a8, grpc.default_authority=192.168.75.88:2379, grpc.initial_reconnect_backoff_ms=1000, grpc
.internal.subchannel_pool=0x7f5f62c38e50, grpc.keepalive_time_ms=10000, grpc.keepalive_timeout_ms=3000, grpc.max_receive_mess
age_length=-1, grpc.max_reconnect_backoff_ms=5000, grpc.max_send_message_length=-1, grpc.primary_user_agent=grpc-rust/0.10.4,
 grpc.resource_quota=0x7f5f62cbb0e0, grpc.server_uri=dns:///192.168.75.88:2379}: connect failed: {\"created\":\"@1712418649.8
88694501\",\"description\":\"Failed to connect to remote host: Connection refused\",\"errno\":111,\"file\":\"/workspace/.carg
o/registry/src/index.crates.io-6f17d22bba15001f/grpcio-sys-0.10.3+1.44.0-patched/grpc/src/core/lib/iomgr/tcp_client_posix.cc\
",\"file_line\":200,\"os_error\":\"Connection refused\",\"syscall\":\"connect\",\"target_address\":\"ipv4:192.168.75.88:2379\
"}"] [thread_id=13]
[2024/04/06 23:50:49.888 +08:00] [INFO] [<unknown>] ["subchannel 0x7f5f62c56c00 {address=ipv4:192.168.75.88:2379, args=grpc.c
lient_channel_factory=0x7f5f62c978a8, grpc.default_authority=192.168.75.88:2379, grpc.initial_reconnect_backoff_ms=1000, grpc
.internal.subchannel_pool=0x7f5f62c38e50, grpc.keepalive_time_ms=10000, grpc.keepalive_timeout_ms=3000, grpc.max_receive_mess
age_length=-1, grpc.max_reconnect_backoff_ms=5000, grpc.max_send_message_length=-1, grpc.primary_user_agent=grpc-rust/0.10.4,
 grpc.resource_quota=0x7f5f62cbb0e0, grpc.server_uri=dns:///192.168.75.88:2379}: Retry in 999 milliseconds"] [thread_id=13]
[2024/04/06 23:50:49.888 +08:00] [INFO] [util.rs:601] ["PD failed to respond"] [err="Grpc(RpcFailure(RpcStatus { code: 14-UNA
VAILABLE, message: \"failed to connect to all addresses\", details: [] }))"] [endpoints=192.168.75.88:2379] [thread_id=1]
[2024/04/06 23:50:49.888 +08:00] [WARN] [client.rs:169] ["validate PD endpoints failed"] [err="Other(\"[components/pd_client/
src/util.rs:634]: PD cluster failed to respond\")"] [thread_id=1]
[2024/04/06 23:50:50.190 +08:00] [INFO] [util.rs:639] ["connecting to PD endpoint"] [endpoints=192.168.75.88:2379] [thread_id
=1]
[2024/04/06 23:50:50.190 +08:00] [INFO] [<unknown>] ["subchannel 0x7f5f62c72800 {address=ipv4:192.168.75.88:2379, args=grpc.c
lient_channel_factory=0x7f5f62c978a8, grpc.default_authority=192.168.75.88:2379, grpc.initial_reconnect_backoff_ms=1000, grpc
.internal.subchannel_pool=0x7f5f62c38e50, grpc.keepalive_time_ms=10000, grpc.keepalive_timeout_ms=3000, grpc.max_receive_mess
age_length=-1, grpc.max_reconnect_backoff_ms=5000, grpc.max_send_message_length=-1, grpc.primary_user_agent=grpc-rust/0.10.4,
 grpc.resource_quota=0x7f5f62cbb0e0, grpc.server_uri=dns:///192.168.75.88:2379}: connect failed: {\"created\":\"@1712418650.1
90326725\",\"description\":\"Failed to connect to remote host: Connection refused\",\"errno\":111,\"file\":\"/workspace/.carg
o/registry/src/index.crates.io-6f17d22bba15001f/grpcio-sys-0.10.3+1.44.0-patched/grpc/src/core/lib/iomgr/tcp_client_posix.cc\
",\"file_line\":200,\"os_error\":\"Connection refused\",\"syscall\":\"connect\",\"target_address\":\"ipv4:192.168.75.88:2379\
"}"] [thread_id=13]
[2024/04/06 23:50:50.190 +08:00] [INFO] [<unknown>] ["subchannel 0x7f5f62c72800 {address=ipv4:192.168.75.88:2379, args=grpc.c
lient_channel_factory=0x7f5f62c978a8, grpc.default_authority=192.168.75.88:2379, grpc.initial_reconnect_backoff_ms=1000, grpc
.internal.subchannel_pool=0x7f5f62c38e50, grpc.keepalive_time_ms=10000, grpc.keepalive_timeout_ms=3000, grpc.max_receive_mess
age_length=-1, grpc.max_reconnect_backoff_ms=5000, grpc.max_send_message_length=-1, grpc.primary_user_agent=grpc-rust/0.10.4,
 grpc.resource_quota=0x7f5f62cbb0e0, grpc.server_uri=dns:///192.168.75.88:2379}: Retry in 1000 milliseconds"] [thread_id=13]
[2024/04/06 23:50:50.190 +08:00] [INFO] [util.rs:601] ["PD failed to respond"] [err="Grpc(RpcFailure(RpcStatus { code: 14-UNA
VAILABLE, message: \"failed to connect to all addresses\", details: [] }))"] [endpoints=192.168.75.88:2379] [thread_id=1]
[root@JiekeXu log]#

** PD  日志**

[root@JiekeXu log]# pwd
/app/tidb-deploy/tidb-4000/log
[root@JiekeXu log]# tail -300 pd.log
[2024/04/06 23:52:10.609 +08:00] [INFO] [registry.go:92] ["restful API service registered successfully"] [prefix=pd-192.168.75.88-2379] [service-name=ResourceManager]
[2024/04/06 23:52:10.609 +08:00] [WARN] [config.go:620] ["Running http and grpc server on single port. This is not recommended for production."]
[2024/04/06 23:52:10.609 +08:00] [INFO] [etcd.go:119] ["configuring peer listeners"] [listen-peer-urls="[http://0.0.0.0:2380]"]
[2024/04/06 23:52:10.609 +08:00] [ERROR] [etcd.go:537] ["creating peer listener failed"] [error="listen tcp 0.0.0.0:2380: bind: address already in use"]
[2024/04/06 23:52:10.609 +08:00] [INFO] [etcd.go:373] ["closing etcd server"] [name=pd-192.168.75.88-2379] [data-dir=/app/tidb-data/pd-2379] [advertise-peer-urls="[http://192.168.75.88:2380]"] [advertise-client-urls="[http://192.168.75.88:2379]"]
[2024/04/06 23:52:10.609 +08:00] [INFO] [etcd.go:377] ["closed etcd server"] [name=pd-192.168.75.88-2379] [data-dir=/app/tidb-data/pd-2379] [advertise-peer-urls="[http://192.168.75.88:2380]"] [advertise-client-urls="[http://192.168.75.88:2379]"]
[2024/04/06 23:52:10.609 +08:00] [INFO] [systimemon.go:30] ["start system time monitor"]
[2024/04/06 23:52:10.609 +08:00] [FATAL] [main.go:280] ["run server failed"] [error="[PD:etcd:ErrStartEtcd]listen tcp 0.0.0.0:2380: bind: address already in use: listen tcp 0.0.0.0:2380: bind: address already in use"] [stack="main.start\n\t/workspace/source/pd/cmd/pd-server/main.go:280\nmain.createServerWrapper\n\t/workspace/source/pd/cmd/pd-server/main.go:191\ngithub.com/spf13/cobra.(*Command).execute\n\t/root/go/pkg/mod/github.com/spf13/cobra@v1.0.0/command.go:846\ngithub.com/spf13/cobra.(*Command).ExecuteC\n\t/root/go/pkg/mod/github.com/spf13/cobra@v1.0.0/command.go:950\ngithub.com/spf13/cobra.(*Command).Execute\n\t/root/go/pkg/mod/github.com/spf13/cobra@v1.0.0/command.go:887\nmain.main\n\t/workspace/source/pd/cmd/pd-server/main.go:71\nruntime.main\n\t/usr/local/go/src/runtime/proc.go:267"]

查找并杀掉占用进程

[root@JiekeXu log]# isof -i:2380
bash: isof: command not found...
[root@JiekeXu log]# lsof -i:2380
COMMAND   PID USER   FD   TYPE   DEVICE SIZE/OFF NODE NAME
etcd    54500 etcd    5u  IPv4 27152002      0t0  TCP 192.168.32.31:2380 (LISTEN)
etcd    54500 etcd    8u  IPv4 40957112      0t0  TCP JiekeXu:59596->192.168.32.32:2380 (SYN_SENT)
etcd    54500 etcd   15u  IPv4 40957110      0t0  TCP JiekeXu:49312->192.168.32.33:2380 (SYN_SENT)
etcd    54500 etcd   16u  IPv4 40957111      0t0  TCP JiekeXu:59594->192.168.32.32:2380 (SYN_SENT)
[root@JiekeXu log]#
[root@JiekeXu log]# kill 54500
[root@JiekeXu log]# lsof -i:2380

全文完,希望可以帮到正在阅读的你,如果觉得有帮助,可以分享给你身边的朋友,同事,你关心谁就分享给谁,一起学习共同进步~~~

❤️ 欢迎关注我的公众号【JiekeXu DBA之路】,一起学习新知识!————————————————————————————

公众号:JiekeXu DBA之路

墨天轮:https://www.modb.pro/u/4347

ITPUB:https://blog.itpub.net/69968215

腾讯云:https://cloud.tencent.com/developer/user/5645107

————————————————————————————

TiDB 8.0 新特性尝鲜_tidb_19