PostgreSQL的一主两从集群搭建部署 (两同步)

一、实验环境

虚拟机名	IP	身份	简称
keep-postgres12-node1	192.168.122.87	主节点	node1
keep-postgres12-node2	192.168.122.89	备节点	node2
keep-postgres12-node3	192.168.122.90	备节点	node3

二、安装数据库

源码包方式（主）

1、创建用户

[root@keep-postgres12-node1 ~]# groupadd postgres
[root@keep-postgres12-node1 ~]# useradd -g postgres postgres -m -s /bin/bash
[root@keep-postgres12-node1 ~]# echo "Database@123" | passwd --stdin postgres

2、修改服务器内核信息

[root@keep-postgres12-node1 ~]# echo '# 最大共享内存段大小 (默认值68719476736)
kernel.shmmax = 68719476736 
# 可以使用的共享内存的总量 (默认值4294967296)
kernel.shmall = 4294967296 
# 整个系统共享内存段的最大数目
kernel.shmmni = 4096 
# 每个信号对象集的最大信号对象数
kernel.sem = 50100 64128000 50100 1280 
# 文件句柄的最大数量
fs.file-max = 7672460 
# 应用程序可使用的IPv4端口范围
net.ipv4.ip_local_port_range = 9000 65000
# 套接字接收缓冲区大小的缺省值
net.core.rmem_default = 1048576 
# 套接字发送缓冲区大小的缺省值
net.core.wmem_default = 262144 
# 套接字发送缓冲区大小的最大值
net.core.wmem_max = 1048576' >> /etc/sysctl.conf

# 使参数生效
[root@keep-postgres12-node1 ~]# sysctl -p

3、安装依赖包

[root@keep-postgres12-node1 ~]# yum install gcc gcc-c++ zlib-devel readline-devel perl-ExtUtils-Embed pam-devel openssl openssl-devel cmake libxslt-devel libxml2-devel openldap-devel python-devel tcl tcl-devel bison flex xmlto -y

4、创建数据库目录

[root@keep-postgres12-node1 ~]# mkdir /data
[root@keep-postgres12-node1 ~]# mkdir /data/postgres12.2
[root@keep-postgres12-node1 ~]# chown -R postgres: /data

5、创建环境变量

[root@keep-postgres12-node1 ~]# su - postgres
[postgres@keep-postgres12-node1 ~]$ echo 'export PGPORT=5432
export PG_HOME=/data/postgres12.2
export PATH=$PG_HOME/bin:$PATH
export PGDATA=$PG_HOME/data
export LD_LIBRARY_PATH=$PG_HOME/lib
export LANG=en_US.utf8' >> ~/.bash_profile && source  ~/.bash_profile

6、下载源码包，并解压

链接1：https://ftp.postgresql.org/pub/source

链接2：https://www.postgresql.org/ftp/source/

# 安装postgresql-12.2版本
# 下载postgresql-12.2源码包
[postgres@keep-postgres12-node1 ~]$ cd /data
[postgres@keep-postgres12-node1 data]$ wget https://ftp.postgresql.org/pub/source/v12.2/postgresql-12.2.tar.bz2

# 解压源码包
[postgres@keep-postgres12-node1 data]$ tar -xf postgresql-12.2.tar.bz2

7、编译源码包

[postgres@keep-postgres12-node1 postgresql-12.2]$ ./configure --prefix=/data/postgres12.2 --with-pgport=5432 --with-openssl --with-perl \
--with-tcl --with-python --with-pam --without-ldap --with-libxml --with-libxslt \
--enable-thread-safety --with-wal-blocksize=16 --with-blocksize=8

# 包括第三方插件全部编译，包含文档和所有的contirb
[postgres@keep-postgres12-node1 postgresql-12.2]$ gmake world

8、安装数据库

[postgres@keep-postgres12-node1 postgresql-12.2]$ make && make install

9、配置wal日志目录

[postgres@keep-postgres12-node1 postgresql-12.2]$ # 用于保存wal日志
[postgres@keep-postgres12-node1 postgresql-12.2]$ mkdir $PG_HOME/arch

10、初始化数据库

[postgres@keep-postgres12-node1 postgresql-12.2]$ # 创建目录
[postgres@keep-postgres12-node1 postgresql-12.2]$ mkdir $PG_HOME/data
[postgres@keep-postgres12-node1 postgresql-12.2]$ # 初始化数据库集簇
[postgres@keep-postgres12-node1 postgresql-12.2]$ # --data-checksums 主从复制时需要
[postgres@keep-postgres12-node1 postgresql-12.2]$ initdb -D $PGDATA -W --data-checksums

11、启动数据库

[postgres@keep-postgres12-node1 postgresql-12.2]$ pg_ctl -D $PGDATA start

12、创建数据库用户和数据库

[postgres@keep-postgres12-node1 data]$ psql -d postgres -p 5432
postgres=# create database testdb;
postgres=# create user keep with Superuser password  'keep';

13、配置远程连接

# 配置连接

[postgres@keep-postgres12-node1 data]$ cd $PGDATA
[postgres@keep-postgres12-node1 data]$ sed -i "s/#listen_addresses = 'localhost'/listen_addresses = '*'/g" postgresql.conf
[postgres@keep-postgres12-node1 data]$ echo 'host all all 0.0.0.0/0 md5' >> pg_hba.conf

# 重启数据库

[postgres@keep-postgres12-node1 data]$ pg_ctl stop
[postgres@keep-postgres12-node1 data]$ pg_ctl -D $PGDATA start

# 验证远程连接

[postgres@keep-postgres12-node1 data]$ psql -d testdb -U keep -p 5432 -h 192.168.140.96

14、配置数据库日志

[postgres@keep-postgres12-node1 data]$ echo "# Add settings for extensions here
log_destination='csvlog'
logging_collector=on
log_directory='pg_log'
log_filename='postgresql-%Y-%m-%d.log'
log_truncate_on_rotation=off
log_rotation_age=1d
log_rotation_size=0
log_error_verbosity=verbose" >> $PGDATA/postgresql.conf

源码包方式（备 node2/node3）

1、创建用户

[root@keep-postgres12-node2 ~]# groupadd postgres
[root@keep-postgres12-node2 ~]# useradd -g postgres postgres -m -s /bin/bash
[root@keep-postgres12-node2 ~]# echo "Database@123" | passwd --stdin postgres

2、修改服务器内核信息

[root@keep-postgres12-node2 ~]# echo '# 最大共享内存段大小 (默认值68719476736)
kernel.shmmax = 68719476736 
# 可以使用的共享内存的总量 (默认值4294967296)
kernel.shmall = 4294967296 
# 整个系统共享内存段的最大数目
kernel.shmmni = 4096 
# 每个信号对象集的最大信号对象数
kernel.sem = 50100 64128000 50100 1280 
# 文件句柄的最大数量
fs.file-max = 7672460 
# 应用程序可使用的IPv4端口范围
net.ipv4.ip_local_port_range = 9000 65000
# 套接字接收缓冲区大小的缺省值
net.core.rmem_default = 1048576 
# 套接字发送缓冲区大小的缺省值
net.core.wmem_default = 262144 
# 套接字发送缓冲区大小的最大值
net.core.wmem_max = 1048576' >> /etc/sysctl.conf

# 使参数生效
[root@keep-postgres12-node2 ~]# sysctl -p

3、安装依赖包

[root@keep-postgres12-node2 ~]# yum install gcc gcc-c++ zlib-devel readline-devel perl-ExtUtils-Embed pam-devel openssl openssl-devel cmake libxslt-devel libxml2-devel openldap-devel python-devel tcl tcl-devel bison flex xmlto -y

4、创建数据库目录

[root@keep-postgres12-node2 ~]# mkdir /data
[root@keep-postgres12-node2 ~]# mkdir /data/postgres12.2
[root@keep-postgres12-node2 ~]# chown -R postgres: /data

5、创建环境变量

[root@keep-postgres12-node2 ~]# su - postgres
[postgres@keep-postgres12-node2 ~]$ echo 'export PGPORT=5432
export PG_HOME=/data/postgres12.2
export PATH=$PG_HOME/bin:$PATH
export PGDATA=$PG_HOME/data
export LD_LIBRARY_PATH=$PG_HOME/lib
export LANG=en_US.utf8' >> ~/.bash_profile && source  ~/.bash_profile

6、下载源码包，并解压

链接1：https://ftp.postgresql.org/pub/source

链接2：https://www.postgresql.org/ftp/source/

# 安装postgresql-12.2版本
# 下载postgresql-12.2源码包
[postgres@keep-postgres12-node2 ~]$ cd /data
[postgres@keep-postgres12-node2 data]$ wget https://ftp.postgresql.org/pub/source/v12.2/postgresql-12.2.tar.bz2

# 解压源码包
[postgres@keep-postgres12-node2 data]$ tar -xf postgresql-12.2.tar.bz2

7、编译源码包

[postgres@keep-postgres12-node2 postgresql-12.2]$ ./configure --prefix=/data/postgres12.2 --with-pgport=5432 --with-openssl --with-perl \
--with-tcl --with-python --with-pam --without-ldap --with-libxml --with-libxslt \
--enable-thread-safety --with-wal-blocksize=16 --with-blocksize=8

# 包括第三方插件全部编译，包含文档和所有的contirb
[postgres@keep-postgres12-node2 postgresql-12.2]$ gmake world

8、安装数据库

[postgres@keep-postgres12-node2 postgresql-12.2]$ make && make install

9、配置wal日志目录

[postgres@keep-postgres12-node1 postgresql-12.2]$ # 用于保存wal日志
[postgres@keep-postgres12-node1 postgresql-12.2]$ mkdir $PG_HOME/arch

配置互信

1、配置主机名

# keep-postgres12-node1
[root@keep-postgres12-node1 ~]# echo '192.168.122.87 keep-postgres12-node1
192.168.122.88 keep-postgres12-node2
192.168.122.90 keep-postgres12-node3' >> /etc/hosts

# keep-postgres12-node2
[root@keep-postgres12-node2 ~]# echo '192.168.122.87 keep-postgres12-node1
192.168.122.88 keep-postgres12-node2
192.168.122.90 keep-postgres12-node3' >> /etc/hosts

# keep-postgres12-node3
[root@keep-postgres12-node3 ~]# echo '192.168.122.87 keep-postgres12-node1
192.168.122.88 keep-postgres12-node2
192.168.122.90 keep-postgres12-node3' >> /etc/hosts

2、配置postgres用户互信

# keep-postgres12-node1

[postgres@keep-postgres12-node1 ~]$ ssh-keygen
[postgres@keep-postgres12-node1 ~]$ ssh-copy-id keep-postgres12-node1
[postgres@keep-postgres12-node1 ~]$ ssh-copy-id keep-postgres12-node2
[postgres@keep-postgres12-node1 ~]$ ssh-copy-id keep-postgres12-node3

# 验证
[postgres@keep-postgres12-node1 ~]$ ssh keep-postgres12-node1
[postgres@keep-postgres12-node1 ~]$ ssh keep-postgres12-node2
[postgres@keep-postgres12-node1 ~]$ ssh keep-postgres12-node3

# keep-postgres12-node2

[postgres@keep-postgres12-node2 ~]$ ssh-keygen
[postgres@keep-postgres12-node2 ~]$ ssh-copy-id keep-postgres12-node1
[postgres@keep-postgres12-node2 ~]$ ssh-copy-id keep-postgres12-node2
[postgres@keep-postgres12-node2 ~]$ ssh-copy-id keep-postgres12-node3

# 验证
[postgres@keep-postgres12-node2 ~]$ ssh keep-postgres12-node1
[postgres@keep-postgres12-node2 ~]$ ssh keep-postgres12-node2
[postgres@keep-postgres12-node2 ~]$ ssh keep-postgres12-node3

# keep-postgres12-node3

[postgres@keep-postgres12-node3 ~]$ ssh-keygen
[postgres@keep-postgres12-node3 ~]$ ssh-copy-id keep-postgres12-node1
[postgres@keep-postgres12-node3 ~]$ ssh-copy-id keep-postgres12-node2
[postgres@keep-postgres12-node3 ~]$ ssh-copy-id keep-postgres12-node3

# 验证
[postgres@keep-postgres12-node3 ~]$ ssh keep-postgres12-node1
[postgres@keep-postgres12-node3 ~]$ ssh keep-postgres12-node2
[postgres@keep-postgres12-node3 ~]$ ssh keep-postgres12-node3

三、修改主节点配置

1、创建同步用户

# 数据库需要在线状态
[postgres@keep-postgres12-node1 ~]$ psql -c "create role replrole login replication encrypted password 'ReplRole@123';"

2、配置$PGDATA/pg_hba.conf

[postgres@keep-postgres12-node1 ~]$ echo 'host  replication  replrole  keep-postgres12-node1  trust
host  replication  replrole  keep-postgres12-node2  trust
host  replication  replrole  keep-postgres12-node3  trust
' >> $PGDATA/pg_hba.conf

3、修改数据库参数

[postgres@keep-postgres12-node1 ~]$ echo "wal_level = replica
max_wal_senders=20
wal_keep_segments =64
# 开启归档
archive_mode = on
archive_command = 'cp %p /data/postgres12.2/arch/%f'
restore_command = 'cp /data/postgres12.2/arch/%f %p'
recovery_target_timeline = 'latest'
# full_page_writes是控制是否开启全页写入
full_page_writes = on
# 将每个磁盘页的全部内容写入到WAL
wal_log_hints = on 
# 配置同步,此时pg2的优先级比pg3的要高
# 一个节点的sync_state为sync，剩下的是potential
synchronous_standby_names = 'standby_pg2,standby_pg3' 
# 默认值为 on 
# 可以设置为remote_write，对主库性能有利
# 开启同步模式，此刻需要强同步，数据无法入库
synchronous_commit = on 
" >> $PGDATA/postgresql.conf

4、重启数据库

[postgres@keep-postgres12-node1 ~]$ pg_ctl restart

5、synchronous_standby_names参数详解

# 该参数指定流复制中需要同步复制的服务器列表，需要配置standby服务器的名字
# 情况一：（无顺序要求）
# 只有这三台备机会被选为同步备机，即其中一个会被选为活动的同步备机，其余两台将处于备用状态。
synchronous_standby_names = 'standby_pg1,standby_pg2,standby_pg3'

# 设置WAL日志强同步至N个节点中的某M个节点
# 情况二：（无顺序要求）
# 只有这三台备机会被选为同步备机，即其中一个会被选为活动的同步备机，其余两台将处于备用状态。
synchronous_standby_names = 'ANY 1 (standby_pg1,standby_pg2,standby_pg3)'

# 情况三：（无顺序要求）
# 只有这三台备机会被选为同步备机，即其中两个会被选为活动的同步备机，其余一台将处于备用状态。
synchronous_standby_names = 'ANY 2 (standby_pg1,standby_pg2,standby_pg3)'

# 情况四：（按照指定节点的顺序）
# 从括号中列出的备机中按顺序选择前两台作为同步备机，选择不了两台节点时，则会无法操作。
synchronous_standby_names = 'FIRST 2 (standby_pg1,standby_pg2,standby_pg3)'

四、修改备节点配置

node2 配置

1、进行数据恢复

[postgres@keep-postgres12-node2 ~]$ pg_basebackup -h keep-postgres12-node1 -p 5432 -U replrole -R -F p -P -D $PG_HOME/data

2、修改standby.signal配置

[postgres@keep-postgres12-node2 data]$ cd $PGDATA
[postgres@keep-postgres12-node2 data]$ cat standby.signal
primary_conninfo = 'host=keep-postgres12-node1 port=5432 user=replrole password=ReplRole@123 application_name=standby_pg2 options=''-c wal_sender_timeout=5000'''
restore_command = 'cp /data/postgres12.2/arch/%f %p'
archive_cleanup_command = 'pg_archivecleanup /data/postgres12.2/arch %r'
standby_mode = on

3、修改postgresql.auto.conf配置

# 注意：强同步需要注意在postgresql.auto.conf中加入application_name=standby_pg2

[postgres@keep-postgres12-node2 data]$ cat postgresql.auto.conf
primary_conninfo = 'user=replrole passfile=''/home/postgres/.pgpass'' host=''keep-postgres12-node1'' application_name=standby_pg2 port=5432 sslmode=prefer sslcompression=0 gssencmode=disable krbsrvname=postgres target_session_attrs=any'

node3 配置

1、进行数据恢复

[postgres@keep-postgres12-node3 ~]$ pg_basebackup -h keep-postgres12-node1 -p 5432 -U replrole -R -F p -P -D $PG_HOME/data

2、修改standby.signal配置

[postgres@keep-postgres12-node3 data]$ cd $PGDATA
[postgres@keep-postgres12-node3 data]$ cat standby.signal
primary_conninfo = 'host=keep-postgres12-node1 port=5432 user=replrole password=ReplRole@123 application_name=standby_pg3 options=''-c wal_sender_timeout=5000'''
restore_command = 'cp /data/postgres12.2/arch/%f %p'
archive_cleanup_command = 'pg_archivecleanup /data/postgres12.2/arch %r'
standby_mode = on

3、修改postgresql.auto.conf配置

# 注意：强同步需要注意在postgresql.auto.conf中加入application_name=standby_pg3

[postgres@keep-postgres12-node3 data]$ cat postgresql.auto.conf
primary_conninfo = 'user=replrole passfile=''/home/postgres/.pgpass'' host=''keep-postgres12-node1'' application_name=standby_pg3 port=5432 sslmode=prefer sslcompression=0 gssencmode=disable krbsrvname=postgres target_session_attrs=any'

五、启动数据同步

# 注意：主库修改synchronous_standby_names = ‘standby_pg2’ 后，需等待备库接收主库发送的WAL日志流并写入WAL文件，之后才向客户端返回成功。

1、主库检查点

[postgres@keep-postgres12-node1 ~]$ pg_controldata
pg_control version number:            1201
Catalog version number:               201909212
Database system identifier:           7435168474920575893
Database cluster state:               `in production`
pg_control last modified:             Tue 10 Dec 2024 04:23:13 PM CST
Latest checkpoint location:           0/2E000060
Latest checkpoint's REDO location:    0/2E000028
Latest checkpoint's REDO WAL file:    00000007000000000000002E
Latest checkpoint's TimeLineID:       7
Latest checkpoint's PrevTimeLineID:   7
Latest checkpoint's full_page_writes: on
Latest checkpoint's NextXID:          0:640
Latest checkpoint's NextOID:          24913
Latest checkpoint's NextMultiXactId:  1
Latest checkpoint's NextMultiOffset:  0
Latest checkpoint's oldestXID:        480
Latest checkpoint's oldestXID's DB:   1
Latest checkpoint's oldestActiveXID:  640
Latest checkpoint's oldestMultiXid:   1
Latest checkpoint's oldestMulti's DB: 1
Latest checkpoint's oldestCommitTsXid:0
Latest checkpoint's newestCommitTsXid:0
Time of latest checkpoint:            Tue 10 Dec 2024 04:23:13 PM CST
Fake LSN counter for unlogged rels:   0/3E8
Minimum recovery ending location:     0/0
Min recovery ending loc's timeline:   0
Backup start location:                0/0
Backup end location:                  0/0
End-of-backup record required:        no
wal_level setting:                    replica
wal_log_hints setting:                on
max_connections setting:              100
max_worker_processes setting:         8
max_wal_senders setting:              20
max_prepared_xacts setting:           0
max_locks_per_xact setting:           64
track_commit_timestamp setting:       off
Maximum data alignment:               8
Database block size:                  8192
Blocks per segment of large relation: 131072
WAL block size:                       16384
Bytes per WAL segment:                16777216
Maximum length of identifiers:        64
Maximum columns in an index:          32
Maximum size of a TOAST chunk:        1996
Size of a large-object chunk:         2048
Date/time type storage:               64-bit integers
Float4 argument passing:              by value
Float8 argument passing:              by value
Data page checksum version:           1
Mock authentication nonce:            2fd610c9c82cf604c47448c0e7c842aa9ac4944e8fea828fcc0d2425b6f3a6e4

2、启动备库

node2

[postgres@keep-postgres12-node2 data]$ pg_ctl start

node3

[postgres@keep-postgres12-node3 data]$ pg_ctl start

3、查看备库日志

node2

[postgres@keep-postgres12-node2 pg_log]$ tail -f postgresql-2024-12-10.csv

2024-12-10 16:20:45.667 CST,,,24030,,6757f9dd.5dde,1,,2024-12-10 16:20:45 CST,,0,LOG,00000,"ending log output to stderr",,"Future log output will go to log destination ""csvlog"".",,,,,,"PostmasterMain, postmaster.c:1300",""
2024-12-10 16:20:45.672 CST,,,24032,,6757f9dd.5de0,1,,2024-12-10 16:20:45 CST,,0,LOG,00000,"database system was interrupted; last known up at 2024-12-10 16:13:44 CST",,,,,,,,"StartupXLOG, xlog.c:6269",""
2024-12-10 16:20:45.894 CST,,,24032,,6757f9dd.5de0,2,,2024-12-10 16:20:45 CST,,0,LOG,00000,"entering standby mode",,,,,,,,"StartupXLOG, xlog.c:6324",""
2024-12-10 16:20:45.906 CST,,,24032,,6757f9dd.5de0,3,,2024-12-10 16:20:45 CST,1/0,0,LOG,00000,"redo starts at 0/2B000060",,,,,,,,"StartupXLOG, xlog.c:7037",""
2024-12-10 16:20:45.907 CST,,,24032,,6757f9dd.5de0,4,,2024-12-10 16:20:45 CST,1/0,0,LOG,00000,"consistent recovery state reached at 0/2B000138",,,,,,,,"CheckRecoveryConsistency, xlog.c:7880",""
2024-12-10 16:20:45.908 CST,,,24030,,6757f9dd.5dde,2,,2024-12-10 16:20:45 CST,,0,LOG,00000,"database system is ready to accept read only connections",,,,,,,,"sigusr1_handler, postmaster.c:5153",""
2024-12-10 16:20:45.923 CST,,,24046,,6757f9dd.5dee,1,,2024-12-10 16:20:45 CST,,0,LOG,00000,"started streaming WAL from primary at 0/2C000000 on timeline 7",,,,,,,,"WalReceiverMain, walreceiver.c:371",""

node3

[postgres@keep-postgres12-node3 pg_log]$ tail -f postgresql-2024-12-10.csv

2024-12-10 16:24:45.339 CST,,,25094,,6757facd.6206,1,,2024-12-10 16:24:45 CST,,0,LOG,00000,"ending log output to stderr",,"Future log output will go to log destination ""csvlog"".",,,,,,"PostmasterMain, postmaster.c:1300",""
2024-12-10 16:24:45.344 CST,,,25096,,6757facd.6208,1,,2024-12-10 16:24:45 CST,,0,LOG,00000,"database system was interrupted; last known up at 2024-12-10 16:23:13 CST",,,,,,,,"StartupXLOG, xlog.c:6269",""
2024-12-10 16:24:45.637 CST,,,25096,,6757facd.6208,2,,2024-12-10 16:24:45 CST,,0,LOG,00000,"entering standby mode",,,,,,,,"StartupXLOG, xlog.c:6324",""
2024-12-10 16:24:45.649 CST,,,25096,,6757facd.6208,3,,2024-12-10 16:24:45 CST,1/0,0,LOG,00000,"redo starts at 0/2E000028",,,,,,,,"StartupXLOG, xlog.c:7037",""
2024-12-10 16:24:45.651 CST,,,25096,,6757facd.6208,4,,2024-12-10 16:24:45 CST,1/0,0,LOG,00000,"consistent recovery state reached at 0/2E000100",,,,,,,,"CheckRecoveryConsistency, xlog.c:7880",""
2024-12-10 16:24:45.652 CST,,,25094,,6757facd.6206,2,,2024-12-10 16:24:45 CST,,0,LOG,00000,"database system is ready to accept read only connections",,,,,,,,"sigusr1_handler, postmaster.c:5153",""
2024-12-10 16:24:45.672 CST,,,25104,,6757facd.6210,1,,2024-12-10 16:24:45 CST,,0,LOG,00000,"started streaming WAL from primary at 0/2F000000 on timeline 7",,,,,,,,"WalReceiverMain, walreceiver.c:371",""

4、检查数据库进程

# keep-postgres12-node1
[postgres@keep-postgres12-node1 ~]$ ps -ef | grep postgres
postgres 26423     1  0 16:17 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 26424 26423  0 16:17 ?        00:00:00 postgres: logger
postgres 26426 26423  0 16:17 ?        00:00:00 postgres: checkpointer
postgres 26427 26423  0 16:17 ?        00:00:00 postgres: background writer
postgres 26428 26423  0 16:17 ?        00:00:00 postgres: walwriter
postgres 26429 26423  0 16:17 ?        00:00:00 postgres: autovacuum launcher
postgres 26430 26423  0 16:17 ?        00:00:00 postgres: archiver   last was 00000007000000000000002E.00000028.backup
postgres 26431 26423  0 16:17 ?        00:00:00 postgres: stats collector
postgres 26432 26423  0 16:17 ?        00:00:00 postgres: logical replication launcher
postgres 27747 26423  0 16:20 ?        00:00:00` postgres: walsender replrole 192.168.122.89(54134) streaming 0/2F000148`
postgres 28509 26423  0 16:24 ?        00:00:00 `postgres: walsender replrole 192.168.122.90(62200) streaming 0/2F000148`

# keep-postgres12-node2
[postgres@keep-postgres12-node2 data]$ ps -ef | grep postgres
postgres 24030     1  0 16:20 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 24031 24030  0 16:20 ?        00:00:00 postgres: logger
postgres 24032 24030  0 16:20 ?        00:00:00 postgres: startup   recovering 00000007000000000000002F
postgres 24042 24030  0 16:20 ?        00:00:00 postgres: checkpointer
postgres 24043 24030  0 16:20 ?        00:00:00 postgres: background writer
postgres 24045 24030  0 16:20 ?        00:00:00 postgres: stats collector
postgres 24046 24030  0 16:20 ?        00:00:00 `postgres: walreceiver   streaming 0/2F000148`

# keep-postgres12-node3
[postgres@keep-postgres12-node3 data]$ ps -ef | grep postgres
postgres 25094     1  0 16:24 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 25095 25094  0 16:24 ?        00:00:00 postgres: logger
postgres 25096 25094  0 16:24 ?        00:00:00 postgres: startup   recovering 00000007000000000000002F
postgres 25100 25094  0 16:24 ?        00:00:00 postgres: checkpointer
postgres 25101 25094  0 16:24 ?        00:00:00 postgres: background writer
postgres 25103 25094  0 16:24 ?        00:00:00 postgres: stats collector
postgres 25104 25094  0 16:24 ?        00:00:00 `postgres: walreceiver   streaming 0/2F000148`

六、验证数据同步

1、主库检查同步状态

一节点sync_state 为sync，一节点为potential

[postgres@keep-postgres12-node1 data]$ psql
postgres=# \x
Expanded display is on.
postgres=#  SELECT * FROM pg_stat_replication;
-[ RECORD 1 ]----+------------------------------
pid              | 27747
usesysid         | 16642
usename          | replrole
application_name | standby_pg2
client_addr      | 192.168.122.89
client_hostname  | keep-postgres12-node2
client_port      | 54134
backend_start    | 2024-12-10 16:20:45.918235+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/2F000148
write_lsn        | 0/2F000148
flush_lsn        | 0/2F000148
replay_lsn       | 0/2F000148
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 1
sync_state       | sync
reply_time       | 2024-12-10 16:36:26.891383+08
-[ RECORD 2 ]----+------------------------------
pid              | 29698
usesysid         | 16642
usename          | replrole
application_name | standby_pg3
client_addr      | 192.168.122.90
client_hostname  | keep-postgres12-node3
client_port      | 62202
backend_start    | 2024-12-10 16:31:09.071909+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/2F000148
write_lsn        | 0/2F000148
flush_lsn        | 0/2F000148
replay_lsn       | 0/2F000148
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 2
sync_state       | potential
reply_time       | 2024-12-10 16:36:29.798595+08

2、测试数据同步

-- 主库执行
drop table if exists employees_multi_sync CASCADE;
CREATE TABLE employees_multi_sync (
    id SERIAL PRIMARY KEY,
    name VARCHAR(100),
    position VARCHAR(100),
    department VARCHAR(100),
    hire_date DATE
);

INSERT INTO employees_multi_sync (name, position, department, hire_date)
SELECT 
    'Employee ' || (generate_series(1, 200))::text,
    'Position ' || (generate_series(1, 200))::text,
    'Department ' || (random() *(200-1)+1)::text,
    '2010-01-01'::date + (generate_series(1, 200) * interval '1 day');

-- 备库查询数据
select count(1) from employees_multi_sync ;
-- 计算两边的md5值
SELECT md5(
	string_agg(
    id::text || '-' ||
    name || '-' ||
    position || '-' ||
    department || '-' ||
    TO_CHAR(hire_date, 'YYYY-MM-DD'),',')
) AS row_md5
FROM employees_multi_sync;

七、主备切换

1、停主库

[postgres@keep-postgres12-node1 ~]$ pg_ctl stop -m fast

2、主从切换

备库升主

# 切换之后，$PGDATA下原有的 standby.signal 文件不存在了
[postgres@keep-postgres12-node2 ~]$ pg_ctl promote

查看状态

[postgres@keep-postgres12-node2 ~]$ pg_controldata | grep -i cluster
Database cluster state:               `in production`

3、检查原备库的配置（现主）

检查配置文件postgresql.auto.conf

[postgres@keep-postgres12-node2 data]$ cat postgresql.auto.conf
# Do not edit this file manually!
# It will be overwritten by the ALTER SYSTEM command.
# 检查是否进行注释，配置还是当作备库，自相矛盾
# primary_conninfo = 'user=replrole passfile=''/home/postgres/.pgpass'' host=''keep-postgres12-node1'' application_name=standby_pg2 port=5432 sslmode=prefer sslcompression=0 gssencmode=disable krbsrvname=postgres target_session_attrs=any'

检查配置文件postgresql.conf

# 修改synchronous_standby_names 
[postgres@keep-postgres12-node2 data]$ vim postgresql.conf
synchronous_standby_names = 'standby_pg1,standby_pg3'

检查状态

# 此处进程应为 walwriter，表示还是备机状态。若修改了postgresql.auto.conf，需重启数据库
[postgres@keep-postgres12-node2 data]$ ps -ef | grep postgres
postgres 25288     1  0 Dec02 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 25289 25288  0 Dec02 ?        00:00:00 postgres: logger
postgres 25294 25288  0 Dec02 ?        00:00:02 postgres: checkpointer
postgres 25295 25288  0 Dec02 ?        00:00:02 postgres: background writer
postgres 25296 25288  0 Dec02 ?        00:00:00 postgres: stats collector
postgres 29058 25288  0 16:11 ?        00:00:00 postgres: `walwriter`
postgres 29059 25288  0 16:11 ?        00:00:00 postgres: autovacuum launcher
postgres 29060 25288  0 16:11 ?        00:00:00 postgres: archiver   last was 000000030000000000000020.partial
postgres 29061 25288  0 16:11 ?        00:00:00 postgres: logical replication launcher

[postgres@keep-postgres12-node2 data]$ pg_ctl restart

# 重启数据库后，备机状态为未连接，不会启动walsender
[postgres@keep-postgres12-node2 data]$ ps -ef | grep postgres
postgres 32261     1  0 16:20 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 32262 32261  0 16:20 ?        00:00:00 postgres: logger
postgres 32264 32261  0 16:20 ?        00:00:00 postgres: checkpointer
postgres 32265 32261  0 16:20 ?        00:00:00 postgres: background writer
postgres 32266 32261  0 16:20 ?        00:00:00 postgres: walwriter
postgres 32267 32261  0 16:20 ?        00:00:00 postgres: autovacuum launcher
postgres 32268 32261  0 16:20 ?        00:00:00 postgres: archiver
postgres 32269 32261  0 16:20 ?        00:00:00 postgres: stats collector
postgres 32270 32261  0 16:20 ?        00:00:00 postgres: logical replication launcher

4、备库配置standby.signal文件

node1

# standby.signal文件需要新建
[postgres@keep-postgres12-node1 data]$ cat standby.signal
# 添加以下内容
primary_conninfo = 'host=keep-postgres12-node2 port=5432 user=replrole password=ReplRole@123 application_name=standby_pg1 options=''-c wal_sender_timeout=5000'''
restore_command = 'cp /data/postgres12.2/arch/%f %p'
archive_cleanup_command = 'pg_archivecleanup /data/postgres12.2/arch %r'
standby_mode = on

node3

# standby.signal文件需要新建
[postgres@keep-postgres12-node1 data]$ cat standby.signal
# 修改host为keep-postgres12-node2
primary_conninfo = 'host=keep-postgres12-node2 port=5432 user=replrole password=ReplRole@123 application_name=standby_pg3 options=''-c wal_sender_timeout=5000'''
restore_command = 'cp /data/postgres12.2/arch/%f %p'
archive_cleanup_command = 'pg_archivecleanup /data/postgres12.2/arch %r'
standby_mode = on

5、备库修改配置文件postgresql.auto.conf

node1

[postgres@keep-postgres12-node1 data]$ cat postgresql.auto.conf
# 添加以下内容
primary_conninfo = 'user=replrole passfile=''/home/postgres/.pgpass'' host=''keep-postgres12-node2'' application_name=standby_pg2 port=5432 sslmode=prefer sslcompression=0 gssencmode=disable krbsrvname=postgres target_session_attrs=any'

node3

[postgres@keep-postgres12-node1 data]$ cat postgresql.auto.conf
# 修改host为keep-postgres12-node2
primary_conninfo = 'user=replrole passfile=''/home/postgres/.pgpass'' host=''keep-postgres12-node2'' application_name=standby_pg3 port=5432 sslmode=prefer sslcompression=0 gssencmode=disable krbsrvname=postgres target_session_attrs=any'

6、启动备库

node1

[postgres@keep-postgres12-node1 data]$ pg_ctl restart

node3

[postgres@keep-postgres12-node3 data]$ pg_ctl restart

7、检查数据库进程

# 备库 node1
[postgres@keep-postgres12-node1 ~]$ ps -ef | grep postgres
postgres   440     1  0 16:52 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres   441   440  0 16:52 ?        00:00:00 postgres: logger
postgres   442   440  0 16:52 ?        00:00:00 postgres: startup   recovering 000000080000000000000032
postgres   446   440  0 16:52 ?        00:00:00 postgres: checkpointer
postgres   447   440  0 16:52 ?        00:00:00 postgres: background writer
postgres   448   440  0 16:52 ?        00:00:00 postgres: stats collector
postgres   449   440  0 16:52 ?        00:00:00 `postgres: walreceiver   streaming 0/320001C0`

# 主库
[postgres@keep-postgres12-node2 ~]$ ps -ef | grep postgres
postgres 25853     1  0 16:47 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 25854 25853  0 16:47 ?        00:00:00 postgres: logger
postgres 25856 25853  0 16:47 ?        00:00:00 postgres: checkpointer
postgres 25857 25853  0 16:47 ?        00:00:00 postgres: background writer
postgres 25858 25853  0 16:47 ?        00:00:00 postgres: walwriter
postgres 25859 25853  0 16:47 ?        00:00:00 postgres: autovacuum launcher
postgres 25860 25853  0 16:47 ?        00:00:00 postgres: archiver
postgres 25861 25853  0 16:47 ?        00:00:00 postgres: stats collector
postgres 25862 25853  0 16:47 ?        00:00:00 postgres: logical replication launcher
postgres 26311 25853  0 16:52 ?        00:00:00 `postgres: walsender replrole 192.168.122.90(16994) streaming 0/320001C0`
postgres 26312 25853  0 16:52 ?        00:00:00 `postgres: walsender replrole 192.168.122.87(18939) streaming 0/320001C0`

# 备库 node3
[postgres@keep-postgres12-node3 ~]$ ps -ef | grep postgres
postgres 31097     1  0 16:52 ?        00:00:00 /data/postgres12.2/bin/postgres
postgres 31098 31097  0 16:52 ?        00:00:00 postgres: logger
postgres 31099 31097  0 16:52 ?        00:00:00 postgres: startup   recovering 000000080000000000000032
postgres 31103 31097  0 16:52 ?        00:00:00 postgres: checkpointer
postgres 31104 31097  0 16:52 ?        00:00:00 postgres: background writer
postgres 31105 31097  0 16:52 ?        00:00:00 postgres: stats collector
postgres 31106 31097  0 16:52 ?        00:00:00 `postgres: walreceiver   streaming 0/320001C0`

8、检查同步状态

[postgres@keep-postgres12-node2 ~]$ psql
postgres=# \x
Expanded display is on.
postgres=# SELECT * FROM pg_stat_replication;
-[ RECORD 1 ]----+------------------------------
pid              | 26311
usesysid         | 16642
usename          | replrole
application_name | standby_pg3
client_addr      | 192.168.122.90
client_hostname  | keep-postgres12-node3
client_port      | 16994
backend_start    | 2024-12-10 16:52:56.465729+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/32036768
write_lsn        | 0/32036768
flush_lsn        | 0/32036768
replay_lsn       | 0/32036768
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 2
sync_state       | potential
reply_time       | 2024-12-10 16:56:27.053456+08
-[ RECORD 2 ]----+------------------------------
pid              | 26312
usesysid         | 16642
usename          | replrole
application_name | standby_pg1
client_addr      | 192.168.122.87
client_hostname  | keep-postgres12-node1
client_port      | 18939
backend_start    | 2024-12-10 16:52:58.753138+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/32036768
write_lsn        | 0/32036768
flush_lsn        | 0/32036768
replay_lsn       | 0/32036768
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 1
sync_state       | sync
reply_time       | 2024-12-10 16:56:27.047169+08

9、验证是否同步

-- 主库执行
drop table if exists employees_multi_sync1 CASCADE;
CREATE TABLE employees_multi_sync1 (
    id SERIAL PRIMARY KEY,
    name VARCHAR(100),
    position VARCHAR(100),
    department VARCHAR(100),
    hire_date DATE
);

INSERT INTO employees_multi_sync1 (name, position, department, hire_date)
SELECT 
    'Employee ' || (generate_series(1, 200))::text,
    'Position ' || (generate_series(1, 200))::text,
    'Department ' || (random() *(200-1)+1)::text,
    '2010-01-01'::date + (generate_series(1, 200) * interval '1 day');

-- 备库查询数据
select count(1) from employees_multi_sync1 ;
-- 计算两边的md5值
SELECT md5(
	string_agg(
    id::text || '-' ||
    name || '-' ||
    position || '-' ||
    department || '-' ||
    TO_CHAR(hire_date, 'YYYY-MM-DD'),',')
) AS row_md5
FROM employees_multi_sync1;

八、pg_rewind 工具

当主备集群中的备库意外崩溃，且经过长时间，归档日志又被删除了，需要把这段时间的增量数据同步回来，那么就可以用到pg_rewind 工具进行同步。pg_rewind 使一个PostgreSQL 数据目录与另一个数据目录一致。

[postgres@keep-postgres12-node1 ~]$ pg_rewind --target-pgdata $PGDATA --source-server='host=keep-postgres12-node2 port=5432 user=keep password=keep dbname=postgres' -P

pg_rewind帮助命令

[postgres@keep-postgres12-node1 ~]$ pg_rewind --help
pg_rewind resynchronizes a PostgreSQL cluster with another copy of the cluster.

Usage:
  pg_rewind [OPTION]...

Options:
  -D, --target-pgdata=DIRECTORY  existing data directory to modify
      --source-pgdata=DIRECTORY  source data directory to synchronize with
      --source-server=CONNSTR    source server to synchronize with
  -n, --dry-run                  stop before modifying anything
  -N, --no-sync                  do not wait for changes to be written
                                 safely to disk
  -P, --progress                 write progress messages
      --debug                    write a lot of debug messages
  -V, --version                  output version information, then exit
  -?, --help                     show this help, then exit