PostgreSQL主从同步配置,切换步骤

零 说明

本文档详细记录如何一步一步在两台Linux服务器上,搭建PostgreSQL数据库主从模式的高可用数据库环境,以及如何手工切换主备库,及其注意事项。

为了实现数据库对于应用层的透明,配合浮动IP来实现,即使数据库执行主备切换,应用层依然通过浮动IP来访问数据库。即:后端数据库执行主备切换前后,应用系统无需做任何更改。

搭建环境选择在CentOS 7+PostgreSQL 9.4.10上。

基本思路和流程大致如下:

1 主库真实IP为172.16.11.148,配置浮动IP为172.16.11.121,主机名为pgprimary,在其上安装并初始化PostgreSQL数据库;

2 备库真实IP为172.16.11.149,配置浮动IP为172.16.11.121,主机名为pgstandby,在其上只安装PostgreSQL数据库软件;

3 通常情况下,浮动IP 172.16.11.121运行在主库上,当主库出故障时,手工执行数据库的主备切换,备库成为新主库,然后新主库上启用浮动IP;

4 修复并重配原主库,使其成为新主库的备库;

一 环境

两台数据库服务器的主机名、文件系统、OS level、网络配置分别如下:

1主库:

[root@pgprimary ~]# hostname
pgprimary
[root@pgprimary ~]# df -Th
文件系统                类型      容量  已用  可用 已用% 挂载点
/dev/mapper/centos-root xfs        47G  1.1G   46G    3% /
devtmpfs                devtmpfs  908M     0  908M    0% /dev
tmpfs                   tmpfs     920M     0  920M    0% /dev/shm
tmpfs                   tmpfs     920M  8.8M  911M    1% /run
tmpfs                   tmpfs     920M     0  920M    0% /sys/fs/cgroup
/dev/sda1               xfs      1014M  142M  873M   14% /boot
tmpfs                   tmpfs     184M     0  184M    0% /run/user/0
[root@pgprimary ~]# cat /etc/redhat-release 
CentOS Linux release 7.5.1804 (Core) 
[root@pgprimary ~]# uname -rm
3.10.0-862.el7.x86_64 x86_64
[root@pgprimary ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.148  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::58b2:50cd:9d7c:8b23  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)
        RX packets 21338  bytes 4481980 (4.2 MiB)
        RX errors 0  dropped 2378  overruns 0  frame 0
        TX packets 270  bytes 31662 (30.9 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgprimary ~]#

2 备库:

[root@pgstandby ~]# hostname
pgstandby
[root@pgstandby ~]# df -Th
文件系统                类型      容量  已用  可用 已用% 挂载点
/dev/mapper/centos-root xfs        47G  1.1G   46G    3% /
devtmpfs                devtmpfs  908M     0  908M    0% /dev
tmpfs                   tmpfs     920M     0  920M    0% /dev/shm
tmpfs                   tmpfs     920M  8.8M  911M    1% /run
tmpfs                   tmpfs     920M     0  920M    0% /sys/fs/cgroup
/dev/sda1               xfs      1014M  142M  873M   14% /boot
tmpfs                   tmpfs     184M     0  184M    0% /run/user/0
[root@pgstandby ~]# cat /etc/redhat-release 
CentOS Linux release 7.5.1804 (Core) 
[root@pgstandby ~]# uname -rm
3.10.0-862.el7.x86_64 x86_64
[root@pgstandby ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.149  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::486a:f6d4:7976:4c9a  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)
        RX packets 22927  bytes 4822730 (4.5 MiB)
        RX errors 0  dropped 2560  overruns 0  frame 0
        TX packets 262  bytes 28988 (28.3 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgstandby ~]#

二 配置浮动IP

步骤3中的监控并启动浮动IP是可选项配置,不适用于本文的数据库主备切换。数据库主备切换还是推荐手工在主备库上启停浮动IP。

1 主库配置浮动IP:

a 拷贝网络配置文件,并修改:

[root@pgprimary ~]# cp /etc/sysconfig/network-scripts/ifcfg-ens160 /etc/sysconfig/network-scripts/ifcfg-ens160:1
[root@pgprimary ~]#

b 修改DEVICE为ens160:1,添加NM_CONTROLLED设置为no,修改IPADDR为浮动IP地址172.16.11.121,并且去掉网关信息,修改完如下:

[root@pgprimary ~]# cat /etc/sysconfig/network-scripts/ifcfg-ens160:1
TYPE=Ethernet
PROXY_METHOD=none
BROWSER_ONLY=no
BOOTPROTO=static
DEFROUTE=yes
IPV4_FAILURE_FATAL=no
IPV6INIT=yes
IPV6_AUTOCONF=yes
IPV6_DEFROUTE=yes
IPV6_FAILURE_FATAL=no
IPV6_ADDR_GEN_MODE=stable-privacy
NAME=ens160
UUID=be70afb7-a578-4dc0-b694-af530586a76e
DEVICE=ens160:1
ONBOOT=yes
IPADDR=172.16.11.121
NETMASK=255.255.255.0
NM_CONTROLLED=no
[root@pgprimary ~]#

c 启用ens160:1网络设备,并测试:

[root@pgprimary ~]# ifup ens160:1
[root@pgprimary ~]# ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
       valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
       valid_lft forever preferred_lft forever
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:50:56:80:64:27 brd ff:ff:ff:ff:ff:ff
    inet 172.16.11.148/24 brd 172.16.11.255 scope global noprefixroute ens160
       valid_lft forever preferred_lft forever
    inet 172.16.11.121/24 brd 172.16.11.255 scope global secondary ens160:1
       valid_lft forever preferred_lft forever
    inet6 fe80::58b2:50cd:9d7c:8b23/64 scope link noprefixroute 
       valid_lft forever preferred_lft forever
[root@pgprimary ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.148  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::58b2:50cd:9d7c:8b23  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)
        RX packets 172902  bytes 59418679 (56.6 MiB)
        RX errors 0  dropped 15626  overruns 0  frame 0
        TX packets 15516  bytes 1443358 (1.3 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ens160:1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.121  netmask 255.255.255.0  broadcast 172.16.11.255
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgprimary ~]#

d 客户端去ping floating IP

$ ping 172.16.11.121
PING 172.16.11.121 (172.16.11.121): 56 data bytes
64 bytes from 172.16.11.121: icmp_seq=0 ttl=63 time=0.653 ms
64 bytes from 172.16.11.121: icmp_seq=1 ttl=63 time=0.766 ms
^C
--- 172.16.11.121 ping statistics ---
2 packets transmitted, 2 packets received, 0.0% packet loss
round-trip min/avg/max/stddev = 0.653/0.710/0.766/0.057 ms

asher at MacBook-Air-3 in ~
$

2 备库配置浮动IP

具体操作步骤,同上述主库配置浮动IP a,b,c步骤。

测试的时候,需要先在主库停止网络设备ens160:1,然后再在备库启动网络设备ens160:1。

主库停ens160:1

[root@pgprimary ~]# ifdown ens160:1
[root@pgprimary ~]#

备库启动ens160:1

[root@pgstandby ~]# ifup ens160:1
[root@pgstandby ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.149  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::486a:f6d4:7976:4c9a  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)
        RX packets 182284  bytes 61612908 (58.7 MiB)
        RX errors 0  dropped 16880  overruns 0  frame 0
        TX packets 15365  bytes 1381833 (1.3 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ens160:1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.121  netmask 255.255.255.0  broadcast 172.16.11.255
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 0  bytes 0 (0.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 0  bytes 0 (0.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgstandby ~]#

至此,完成了对主备库的浮动IP配置。

3 监控并启停浮动IP【可选项】

a 主库监控并启动浮动IP

监控脚本为:

[root@pgprimary ~]# cat monitor_floating_ip.sh 
#!/bin/bash
#This script is used for monitoring the floating IP(172.16.11.121),if it is unreachable,
#then ifup the ens160:1 device on this postgres PRIMARY server.
#Created by 黄伟, whuang@onlyou.com,20181010.

FLOATING_IP=172.16.11.121

c=$(/usr/bin/ping -$FLOATING_IP -c 1|grep Unreachable|wc -l)
if [ $c -gt 0 ]
then 
   /usr/sbin/ifup ens160:1
fi
[root@pgprimary ~]#

crontab定时任务为:

[root@pgprimary ~]# crontab -l
* * * * * /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 10; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 20; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 30; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 40; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 50; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
[root@pgprimary ~]#

说明,脚本用于ping浮动IP,如果ping不通,则在主库上启动浮动IP。定时任务表示每隔10秒调用一次监控浮动IP脚本。

b 备库监控并启动浮动IP

监控脚本为:

[root@pgstandby ~]# cat monitor_floating_ip.sh 
#!/bin/bash
#This script is used for monitoring the floating IP(172.16.11.121),if it is unreachable,and also the PRIMARY
#IP(172.16.11.148) unreachable,then ifup the ens160:1 device on this postgres STANDBY server.
#Created by 黄伟, whuang@onlyou.com,20181010.

PRIMARY_IP=172.16.11.148
FLOATING_IP=172.16.11.121
c1=$(/usr/bin/ping $PRIMARY_IP -c 1|grep Unreachable|wc -l)
c2=$(/usr/bin/ping $FLOATING_IP -c 1|grep Unreachable|wc -l)
c3=$(/usr/sbin/ip addr|grep ens160:1)
if [ $c1 -gt 0 -a $c2 -gt 0 ]
then
   /usr/sbin/ifup ens160:1
elif [ $c1 -eq 0 -a $c2 -eq 0 -a "$c3" ]
then
   /usr/sbin/ifdown ens160:1
fi
[root@pgstandby ~]#

crontab定时任务为:

[root@pgstandby ~]# crontab -l
* * * * * /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 10; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 20; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 30; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 40; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
* * * * * sleep 50; /root/monitor_floating_ip.sh>>/root/monitor_floating_ip.log
[root@pgstandby ~]#

说明,脚本用于ping浮动IP,如果ping不通,且同时ping不通主库实际IP地址,那么意味着主库服务器已经宕机或者整个网络异常,则在备库上启动浮动IP。如果ping主、备库IP正常,且备库上浮动IP正常,则意味着主库服务器恢复正常或网络恢复正常,则在备库上停止浮动IP,把浮动IP交给主库托管。定时任务表示每隔10秒调用一次监控浮动IP脚本。

三 主库安装、配置数据库

1 创建用户、路径

[root@pgprimary ~]# groupadd postgres
[root@pgprimary ~]# useradd -g postgres postgres
[root@pgprimary ~]# passwd postgres
更改用户 postgres 的密码 。
新的 密码:
无效的密码: 密码包含用户名在某些地方
重新输入新的 密码:
passwd:所有的身份验证令牌已经成功更新。
[root@pgprimary ~]# mkdir -p /postgres/9.4.10
[root@pgprimary ~]# chown -R postgres:postgres /postgres/
[root@pgprimary ~]#

2 操作系统安装软件包gcc,zlib,readline,tar

[root@pgprimary ~]# yum install gcc*
具体略
[root@pgprimary ~]# yum install -y zlib*
具体略
[root@pgprimary ~]# yum install -y readline*
具体略
[root@pgprimary ~]# yum install -y tar*
具体略

3 下载软件,postgres用户执行:

[postgres@pgprimary ~]$ wget https://ftp.postgresql.org/pub/source/v9.4.10/postgresql-9.4.10.tar.gz
--2018-09-30 17:19:51--  https://ftp.postgresql.org/pub/source/v9.4.10/postgresql-9.4.10.tar.gz
正在解析主机 ftp.postgresql.org (ftp.postgresql.org)... 204.145.124.244, 217.196.149.55, 174.143.35.246, ...
正在连接 ftp.postgresql.org (ftp.postgresql.org)|204.145.124.244|:443... 已连接。
已发出 HTTP 请求,正在等待回应... 200 OK
长度:23333040 (22M) [application/x-gzip]
正在保存至: “postgresql-9.4.10.tar.gz”

100%[====================================================================================================================================================================================>] 23,333,040  1.32MB/s 用时 20s    

2018-09-30 17:20:14 (1.10 MB/s) - 已保存 “postgresql-9.4.10.tar.gz” [23333040/23333040])

[postgres@pgprimary ~]$ ll
总用量 22788
-rw-rw-r--. 1 postgres postgres 23333040 10月 25 2016 postgresql-9.4.10.tar.gz
[postgres@pgprimary ~]$

4 解压软件,configure,postgres用户执行:

[postgres@pgprimary ~]$ tar -zxvf postgresql-9.4.10.tar.gz 
...
postgresql-9.4.10/configure.in
postgresql-9.4.10/INSTALL
[postgres@pgprimary ~]$ cd postgresql-9.4.10 && ll
总用量 620
-rw-r--r--.  1 postgres postgres    385 10月 25 2016 aclocal.m4
drwxrwxr-x.  2 postgres postgres   4096 10月 25 2016 config
-rwxr-xr-x.  1 postgres postgres 448579 10月 25 2016 configure
-rw-r--r--.  1 postgres postgres  69119 10月 25 2016 configure.in
drwxrwxr-x. 59 postgres postgres   4096 10月 25 2016 contrib
-rw-r--r--.  1 postgres postgres   1192 10月 25 2016 COPYRIGHT
drwxrwxr-x.  3 postgres postgres    107 10月 25 2016 doc
-rw-r--r--.  1 postgres postgres   3620 10月 25 2016 GNUmakefile.in
-rw-r--r--.  1 postgres postgres    283 10月 25 2016 HISTORY
-rw-r--r--.  1 postgres postgres  75420 10月 25 2016 INSTALL
-rw-r--r--.  1 postgres postgres   1489 10月 25 2016 Makefile
-rw-r--r--.  1 postgres postgres   1209 10月 25 2016 README
drwxrwxr-x. 15 postgres postgres   4096 10月 25 2016 src
[postgres@pgprimary postgresql-9.4.10]$ ./configure --prefix=/postgres/9.4.10/
...
config.status: linking src/makefiles/Makefile.linux to src/Makefile.port
[postgres@pgprimary postgresql-9.4.10]$

5 gmake world,postgres用户执行:

[postgres@pgprimary postgresql-9.4.10]$ gmake world
..
gmake[1]: 离开目录“/home/postgres/postgresql-9.4.10/contrib”
PostgreSQL, contrib, and documentation successfully made. Ready to install.
[postgres@pgprimary postgresql-9.4.10]$

6 gmake install-world,postgres用户执行:

[postgres@pgprimary postgresql-9.4.10]$ gmake install-world
...
PostgreSQL, contrib, and documentation installation complete.
[postgres@pgprimary postgresql-9.4.10]$

7 初始化数据库,postgres用户执行:

[postgres@pgprimary ~]$ /postgres/9.4.10/bin/initdb -D /postgres/9.4.10/data
The files belonging to this database system will be owned by user "postgres".
This user must also own the server process.

The database cluster will be initialized with locale "zh_CN.UTF-8".
The default database encoding has accordingly been set to "UTF8".
initdb: could not find suitable text search configuration for locale "zh_CN.UTF-8"
The default text search configuration will be set to "simple".

Data page checksums are disabled.

creating directory /postgres/9.4.10/data ... ok
creating subdirectories ... ok
selecting default max_connections ... 100
selecting default shared_buffers ... 128MB
selecting dynamic shared memory implementation ... posix
creating configuration files ... ok
creating template1 database in /postgres/9.4.10/data/base/1 ... ok
initializing pg_authid ... ok
initializing dependencies ... ok
creating system views ... ok
loading system objects' descriptions ... ok
creating collations ... ok
creating conversions ... ok
creating dictionaries ... ok
setting privileges on built-in objects ... ok
creating information schema ... ok
loading PL/pgSQL server-side language ... ok
vacuuming database template1 ... ok
copying template1 to template0 ... ok
copying template1 to postgres ... ok
syncing data to disk ... ok

WARNING: enabling "trust" authentication for local connections
You can change this by editing pg_hba.conf or using the option -A, or
--auth-local and --auth-host, the next time you run initdb.

Success. You can now start the database server using:

    /postgres/9.4.10/bin/postgres -D /postgres/9.4.10/data
or
    /postgres/9.4.10/bin/pg_ctl -D /postgres/9.4.10/data -l logfile start

[postgres@pgprimary ~]$

8 修改postgres用户环境变量

[postgres@pgprimary ~]$ cat ~/.bash_profile 
# .bash_profile

# Get the aliases and functions
if [ -f ~/.bashrc ]; then
        . ~/.bashrc
fi

# User specific environment and startup programs

PATH=$PATH:$HOME/.local/bin:$HOME/bin:/postgres/9.4.10/bin
PGDATA=/postgres/9.4.10/data
export PATH
export PGDATA
[postgres@pgprimary ~]$

postgres用户的/home/.bash_profile修改之后的信息如下:

9 修改数据库参数

修改/postgres/9.4.10/data/postgresql.conf文件,使其:

listen_addresses = '*'
port = 5432

其它参数暂时不动。
修改/postgres/9.4.10/data/pg_hba.conf,在其中添加下述;

host    all             all             0.0.0.0/0               md5

其它保持不变。

10 启动数据库

[postgres@pgprimary ~]$ pg_ctl start
server starting
[postgres@pgprimary ~]$ LOG:  database system was shut down at 2018-10-10 09:32:14 CST
LOG:  MultiXact member wraparound protections are now enabled
LOG:  database system is ready to accept connections
LOG:  autovacuum launcher started

[postgres@pgprimary ~]$

11 测试数据库

a 备库停用浮动IP

[root@pgstandby ~]# ifdown ens160:1
[root@pgstandby ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.149  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::486a:f6d4:7976:4c9a  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)
        RX packets 14738337  bytes 3468772296 (3.2 GiB)
        RX errors 0  dropped 1833331  overruns 0  frame 0
        TX packets 149019  bytes 13182295 (12.5 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 1  bytes 104 (104.0 B)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 1  bytes 104 (104.0 B)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgstandby ~]#

b 主库启用浮动IP

[root@pgprimary ~]# ifup  ens160:1
[root@pgprimary ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.148  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::58b2:50cd:9d7c:8b23  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)
        RX packets 14551224  bytes 3431186752 (3.1 GiB)
        RX errors 0  dropped 1817255  overruns 0  frame 0
        TX packets 108201  bytes 9678459 (9.2 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ens160:1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.121  netmask 255.255.255.0  broadcast 172.16.11.255
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 12  bytes 1089 (1.0 KiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 12  bytes 1089 (1.0 KiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgprimary ~]#

c 连接主库原来IP

MacBook-Air-3:~ postgres$ psql -h 172.16.11.148 -p 5432 -d postgres -U postgres
Password for user postgres: 
psql (9.5.2, server 9.4.10)
Type "help" for help.

postgres=# \c
psql (9.5.2, server 9.4.10)
You are now connected to database "postgres" as user "postgres".
postgres=#

d 连接主库浮动IP

MacBook-Air-3:~ postgres$ psql -h 172.16.11.121 -p 5432 -d postgres -U postgres
Password for user postgres: 
psql (9.5.2, server 9.4.10)
Type "help" for help.

postgres=# \c
psql (9.5.2, server 9.4.10)
You are now connected to database "postgres" as user "postgres".
postgres=#

从上,可以看到,当主库启用浮动IP之后,客户端既可以通过原IP访问主库,也可以通过浮动IP访问主库。

四 备库安装数据库软件

具体步骤参照步骤三的1-6和步骤8。其它不需要做更多的配置,需要注意的是,该备库只需要安装数据库软件即可,并不需要初始化数据库。

五 配置主备库同步

1 主库添加同步用户

[root@pgprimary ~]# su - postgres
上一次登录:三 10月 10 10:08:01 CST 2018pts/0 上
[postgres@pgprimary ~]$ psql
psql (9.4.10)
Type "help" for help.

postgres=# CREATE ROLE replica login replication encrypted password 'replica';
CREATE ROLE
postgres=# \q
[postgres@pgprimary ~]$

2 主库修改配置文件pg_hba.conf和postgresql.conf

修改/postgres/9.4.10/data/pg_hba.conf,添加下述行:

host    replication     replica     172.16.11.149/32           md5

修改/postgres/9.4.10/data/postgresql.conf,修改下述行:

wal_level = hot_standby  # 这个是设置主为wal的主机
max_wal_senders = 4 # 这个设置了可以最多有几个流复制连接,差不多有几个从,就设置几个
wal_keep_segments = 256 # 设置流复制保留的最多的xlog数目
wal_sender_timeout = 60s # 设置流复制主机发送数据的超时时间
max_connections = 100 # 这个设置要注意下,主库的该参数要根据实际情况设置。从库的max_connections必须要大于主库的该参数

3 主库重启

[postgres@pgprimary ~]$ pg_ctl restart -m fast
waiting for server to shut down...LOG:  received fast shutdown request
.LOG:  aborting any active transactions
LOG:  autovacuum launcher shutting down
FATAL:  terminating connection due to administrator command
LOG:  shutting down
LOG:  database system is shut down
 done
server stopped
server starting
[postgres@pgprimary ~]$ LOG:  database system was shut down at 2018-10-10 16:35:19 CST
LOG:  MultiXact member wraparound protections are now enabled
LOG:  database system is ready to accept connections
LOG:  autovacuum launcher started

[postgres@pgprimary ~]$

4 备库,执行主库的基础备份

[root@pgstandby ~]# su - postgres
上一次登录:三 10月 10 09:43:15 CST 2018pts/0 上
[postgres@pgstandby ~]$ ll /postgres/9.4.10/
总用量 16
drwxrwxr-x. 2 postgres postgres 4096 9月  30 17:30 bin
drwxrwxr-x. 6 postgres postgres 4096 9月  30 17:30 include
drwxrwxr-x. 4 postgres postgres 4096 9月  30 17:30 lib
drwxrwxr-x. 8 postgres postgres 4096 9月  30 17:30 share
[postgres@pgstandby ~]$ which pg_basebackup 
/postgres/9.4.10/bin/pg_basebackup
[postgres@pgstandby ~]$ pg_basebackup -F p --progress -D /postgres/9.4.10/data -h 172.16.11.121 -p 5432 -U replica --password -X stream
Password: 
21122/21122 kB (100%), 1/1 tablespace
[postgres@pgstandby ~]$ ll /postgres/9.4.10/
总用量 20
drwxrwxr-x.  2 postgres postgres 4096 9月  30 17:30 bin
drwx------. 18 postgres postgres 4096 10月 10 16:38 data
drwxrwxr-x.  6 postgres postgres 4096 9月  30 17:30 include
drwxrwxr-x.  4 postgres postgres 4096 9月  30 17:30 lib
drwxrwxr-x.  8 postgres postgres 4096 9月  30 17:30 share
[postgres@pgstandby ~]$ du -sh /postgres/9.4.10/data/
37M     /postgres/9.4.10/data/
[postgres@pgstandby ~]$

可以看到,备库已经生成/postgres/9.4.10/data/。

5 备库,配置recovery.conf文件

从安装文件里cp过来。然后编辑,更改为下面3个参数,其它保持不变。

[postgres@pgstandby ~]$ cp /home/postgres/postgresql-9.4.10/src/backend/access/transam/recovery.conf.sample /postgres/9.4.10/data/recovery.conf
[postgres@pgstandby ~]$ vi /postgres/9.4.10/data/recovery.conf 
...
standby_mode = on
primary_conninfo = 'host=172.16.11.148 port=5432 user=replica password=replica'         # e.g. 'host=localhost port=5432'
recovery_target_timeline = 'latest'
...

6 备库,修改配置文件postgresql.conf

修改下列4个参数即可:

[postgres@pgstandby ~]$ vi /postgres/9.4.10/data/postgresql.conf 
...
hot_standby = on  # 说明这台机器不仅仅是用于数据归档,也用于数据查询
max_standby_streaming_delay = 30s # 数据流备份的最大延迟时间
wal_receiver_status_interval = 1s  # 多久向主报告一次从的状态,当然从每次数据复制都会向主报告状态,这里只是设置最长的间隔时间
hot_standby_feedback = on # 如果有错误的数据复制,是否向主进行反馈
...

7 备库,启动

[postgres@pgstandby ~]$ pg_ctl status
pg_ctl: no server running
[postgres@pgstandby ~]$ pg_ctl start
server starting
[postgres@pgstandby ~]$ LOG:  database system was interrupted; last known up at 2018-10-10 16:38:47 CST
LOG:  entering standby mode
LOG:  redo starts at 0/2000028
LOG:  consistent recovery state reached at 0/2000128
LOG:  database system is ready to accept read only connections
LOG:  started streaming WAL from primary at 0/3000000 on timeline 1

[postgres@pgstandby ~]$ pg_ctl status
pg_ctl: server is running (PID: 28027)
/postgres/9.4.10/bin/postgres
[postgres@pgstandby ~]$ ps -ef|grep postgres
root     27400 25585  0 16:36 pts/0    00:00:00 su - postgres
postgres 27401 27400  0 16:36 pts/0    00:00:00 -bash
postgres 28027     1  0 16:48 pts/0    00:00:00 /postgres/9.4.10/bin/postgres
postgres 28028 28027  0 16:48 ?        00:00:00 postgres: startup process   recovering 000000010000000000000003
postgres 28029 28027  0 16:48 ?        00:00:00 postgres: checkpointer process  
postgres 28030 28027  0 16:48 ?        00:00:00 postgres: writer process   
postgres 28031 28027  0 16:48 ?        00:00:00 postgres: stats collector process  
postgres 28032 28027  0 16:48 ?        00:00:00 postgres: wal receiver process   streaming 0/3000210
postgres 28073 27401  0 16:49 pts/0    00:00:00 ps -ef
postgres 28074 27401  0 16:49 pts/0    00:00:00 grep --color=auto postgres
[postgres@pgstandby ~]$

从上,可以看到,备库的receiver进程,在接收主库的日志信息。

8 验证主备库同步

主库执行:

[postgres@pgprimary ~]$ ps -ef|grep postgres
root     20280 17302  0 16:30 pts/0    00:00:00 su - postgres
postgres 20281 20280  0 16:30 pts/0    00:00:00 -bash
postgres 20677     1  0 16:35 pts/0    00:00:00 /postgres/9.4.10/bin/postgres
postgres 20679 20677  0 16:35 ?        00:00:00 postgres: checkpointer process  
postgres 20680 20677  0 16:35 ?        00:00:00 postgres: writer process   
postgres 20681 20677  0 16:35 ?        00:00:00 postgres: wal writer process  
postgres 20682 20677  0 16:35 ?        00:00:00 postgres: autovacuum launcher process  
postgres 20683 20677  0 16:35 ?        00:00:00 postgres: stats collector process  
postgres 21656 20677  0 16:48 ?        00:00:00 postgres: wal sender process replica 172.16.11.149(51868) streaming 0/3000210
postgres 21939 20281  0 16:52 pts/0    00:00:00 ps -ef
postgres 21940 20281  0 16:52 pts/0    00:00:00 grep --color=auto postgres
[postgres@pgprimary ~]$ psql
psql (9.4.10)
Type "help" for help.

postgres=# \x
Expanded display is on.
postgres=# select * from pg_stat_replication;
-[ RECORD 1 ]----+------------------------------
pid              | 21656
usesysid         | 16385
usename          | replica
application_name | walreceiver
client_addr      | 172.16.11.149
client_hostname  | 
client_port      | 51868
backend_start    | 2018-10-10 16:48:44.010195+08
backend_xmin     | 1893
state            | streaming
sent_location    | 0/3000210
write_location   | 0/3000210
flush_location   | 0/3000210
replay_location  | 0/3000210
sync_priority    | 0
sync_state       | async

postgres=#

从上,看到主库有sender进程,发送同步信息给备库。同时,登录主库也可以查看到同步信息。

 

六 主备库切换

这里,模拟主库出现故障,先做一次switchover测试,即先切换主备库角色,主备互换。

然后,重建原主库或者直接将原主库当作新主库的备库。

最后,再执行一次switchover,主备互换,即恢复到之前的最初状态,主库还是原主库,备库还是原备库。

1 执行第一次主备切换

当原主库库服务器或者数据库故障时,我们可以先停止主库,然后启动备库。

a 停止主库

[postgres@pgprimary data]$ pg_ctl stop -m fast -l /home/postgres/pg.log 
waiting for server to shut down.... done
server stopped
[postgres@pgprimary data]$

b 启动备库

 [postgres@pgstandby data]$ pwd
/postgres/9.4.10/data
[postgres@pgstandby data]$ ll 
总用量 72
-rw-------. 1 postgres postgres   206 10月 10 16:38 backup_label.old
drwx------. 5 postgres postgres    41 10月 10 16:38 base
drwx------. 2 postgres postgres  4096 10月 26 15:57 global
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_clog
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_dynshmem
-rw-------. 1 postgres postgres  4605 10月 16 16:35 pg_hba.conf
-rw-------. 1 postgres postgres  1636 10月 10 16:38 pg_ident.conf
drwx------. 4 postgres postgres    39 10月 10 16:38 pg_logical
drwx------. 4 postgres postgres    36 10月 10 16:38 pg_multixact
drwx------. 2 postgres postgres    18 10月 26 15:57 pg_notify
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_replslot
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_serial
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_snapshots
drwx------. 2 postgres postgres     6 10月 26 15:57 pg_stat
drwx------. 2 postgres postgres     6 10月 26 09:29 pg_stat_tmp
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_subtrans
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_tblspc
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_twophase
-rw-------. 1 postgres postgres     4 10月 10 16:38 PG_VERSION
drwx------. 3 postgres postgres  4096 10月 26 15:57 pg_xlog
-rw-------. 1 postgres postgres    88 10月 10 16:38 postgresql.auto.conf
-rw-------. 1 postgres postgres 20810 10月 10 16:45 postgresql.conf
-rw-------. 1 postgres postgres    30 10月 26 15:57 postmaster.opts
-rw-------. 1 postgres postgres    71 10月 26 15:57 postmaster.pid
-rw-r--r--. 1 postgres postgres  5641 10月 16 16:54 recovery.conf
[postgres@pgstandby data]$ pg_ctl promote -l /home/postgres/pg.log 
server promoting
[postgres@pgstandby data]$ ll
总用量 72
-rw-------. 1 postgres postgres   206 10月 10 16:38 backup_label.old
drwx------. 5 postgres postgres    41 10月 10 16:38 base
drwx------. 2 postgres postgres  4096 10月 26 15:57 global
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_clog
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_dynshmem
-rw-------. 1 postgres postgres  4605 10月 16 16:35 pg_hba.conf
-rw-------. 1 postgres postgres  1636 10月 10 16:38 pg_ident.conf
drwx------. 4 postgres postgres    39 10月 10 16:38 pg_logical
drwx------. 4 postgres postgres    36 10月 10 16:38 pg_multixact
drwx------. 2 postgres postgres    18 10月 26 15:57 pg_notify
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_replslot
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_serial
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_snapshots
drwx------. 2 postgres postgres     6 10月 26 15:57 pg_stat
drwx------. 2 postgres postgres    42 10月 26 16:10 pg_stat_tmp
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_subtrans
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_tblspc
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_twophase
-rw-------. 1 postgres postgres     4 10月 10 16:38 PG_VERSION
drwx------. 3 postgres postgres  4096 10月 26 16:10 pg_xlog
-rw-------. 1 postgres postgres    88 10月 10 16:38 postgresql.auto.conf
-rw-------. 1 postgres postgres 20810 10月 10 16:45 postgresql.conf
-rw-------. 1 postgres postgres    30 10月 26 15:57 postmaster.opts
-rw-------. 1 postgres postgres    71 10月 26 15:57 postmaster.pid
-rw-r--r--. 1 postgres postgres  5641 10月 16 16:54 recovery.done
[postgres@pgstandby data]$

这里,需要注意的是,启动备库的命令是pg_ctl promote,之前备库是出在接受主库同步的状态,需要通过这个命令来将备库提升到primary状态。同时,备库切换为主库之后,/postgres/9.4.10/data/recovery.conf文件自动变为/postgres/9.4.10/data/recovery.done。我个人猜想,主要是为了下次重启该数据库时,直接进入“主库模式”,如果还是recovery.conf的话,意味着该库需要执行recovery操作。

c 备库启动浮动IP

先在主库上停止浮动IP

 [root@pgprimary ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.148  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::58b2:50cd:9d7c:8b23  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)
        RX packets 37310308  bytes 8976173370 (8.3 GiB)
        RX errors 0  dropped 4868601  overruns 0  frame 0
        TX packets 1426780  bytes 127364686 (121.4 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ens160:1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.121  netmask 255.255.255.0  broadcast 172.16.11.255
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 134634  bytes 25999628 (24.7 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 134634  bytes 25999628 (24.7 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

您在 /var/spool/mail/root 中有邮件
[root@pgprimary ~]# ifdown ens160:1
[root@pgprimary ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.148  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::58b2:50cd:9d7c:8b23  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:64:27  txqueuelen 1000  (Ethernet)
        RX packets 37310444  bytes 8976218410 (8.3 GiB)
        RX errors 0  dropped 4868628  overruns 0  frame 0
        TX packets 1426794  bytes 127368186 (121.4 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 134634  bytes 25999628 (24.7 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 134634  bytes 25999628 (24.7 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgprimary ~]#

然后,备库启动浮动IP

 [root@pgstandby ~]# ifup ens160:1
您在 /var/spool/mail/root 中有邮件
[root@pgstandby ~]# ifconfig 
ens160: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.149  netmask 255.255.255.0  broadcast 172.16.11.255
        inet6 fe80::486a:f6d4:7976:4c9a  prefixlen 64  scopeid 0x20
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)
        RX packets 37445369  bytes 8975382297 (8.3 GiB)
        RX errors 0  dropped 4884823  overruns 0  frame 0
        TX packets 1509890  bytes 162194380 (154.6 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

ens160:1: flags=4163<UP,BROADCAST,RUNNING,MULTICAST>  mtu 1500
        inet 172.16.11.121  netmask 255.255.255.0  broadcast 172.16.11.255
        ether 00:50:56:80:5f:53  txqueuelen 1000  (Ethernet)

lo: flags=73<UP,LOOPBACK,RUNNING>  mtu 65536
        inet 127.0.0.1  netmask 255.0.0.0
        inet6 ::1  prefixlen 128  scopeid 0x10
        loop  txqueuelen 1000  (Local Loopback)
        RX packets 195774  bytes 35184330 (33.5 MiB)
        RX errors 0  dropped 0  overruns 0  frame 0
        TX packets 195774  bytes 35184330 (33.5 MiB)
        TX errors 0  dropped 0 overruns 0  carrier 0  collisions 0

[root@pgstandby ~]#

d 测试新主库是否可用

MacBook-Air-3:~ postgres$ psql -h 172.16.11.121 -p 5432 -d postgres -U postgres
Password for user postgres: 
psql (9.5.2, server 9.4.10)
Type "help" for help.

postgres=# select * from t;
 id 
----
  1
(1 row)

postgres=# delete from t;
DELETE 1
postgres=# select * from t;
 id 
----
(0 rows)

postgres=# 

客户端通过浮动IP访问数据库,并删除t表的数据。说明,新主库可用正常使用。同时,记录下,这里是从新主库里删除的数据,接下来,我们重建新备库之后,看看这条被删除的数据,是否也从备库里查询不到?

2 重建主库

这里,为了快速验证和测试主备互换的流程,只是假设主库出现故障,所以,在这里并不需要重建原主库。真实情况下,如果主库因出现故障而执行切换的话,就需要重建。重建的流程和步骤可以参考初始搭建备库的流程来完成。

当然,这里需要将原主库配置为新备库,同样需要参照搭建备库流程五 配置主备库同步中的5,6,7三个步骤。

a 新备库创建配置recovery.conf文件

[postgres@pgprimary ~]$ cp /home/postgres/postgresql-9.4.10/src/backend/access/transam/recovery.conf.sample /postgres/9.4.10/data/recovery.conf
[postgres@pgprimary ~]$ vi /postgres/9.4.10/data/recovery.conf 
...
standby_mode = on
primary_conninfo = 'host=172.16.11.149 port=5432 user=replica password=replica'         # e.g. 'host=localhost port=5432'
recovery_target_timeline = 'latest'
...

需要注意的是,这里主库的信息,需要指向新主库的IP,即原备库的IP 172.16.11.149。

b 新备库修改配置文件postgresql.conf

修改下列4个参数即可:

[postgres@pgprimary ~]$ vi /postgres/9.4.10/data/postgresql.conf 
...
hot_standby = on  # 说明这台机器不仅仅是用于数据归档,也用于数据查询
max_standby_streaming_delay = 30s # 数据流备份的最大延迟时间
wal_receiver_status_interval = 1s  # 多久向主报告一次从的状态,当然从每次数据复制都会向主报告状态,这里只是设置最长的间隔时间
hot_standby_feedback = on # 如果有错误的数据复制,是否向主进行反馈
...

c 启动新备库

[postgres@pgprimary data]$ pg_ctl start -l /home/postgres/pg.log 
server starting
[postgres@pgprimary data]$ ps -ef|grep postgres
root     13481 13458  0 09:22 pts/1    00:00:00 su - postgres
postgres 13482 13481  0 09:22 pts/1    00:00:00 -bash
postgres 14425     1  1 16:55 pts/1    00:00:00 /postgres/9.4.10/bin/postgres
postgres 14426 14425  0 16:55 ?        00:00:00 postgres: startup process   recovering 000000060000000000000005
postgres 14427 14425  0 16:55 ?        00:00:00 postgres: checkpointer process  
postgres 14428 14425  0 16:55 ?        00:00:00 postgres: writer process   
postgres 14429 14425  0 16:55 ?        00:00:00 postgres: stats collector process  
postgres 14430 14425  1 16:55 ?        00:00:00 postgres: wal receiver process   streaming 0/508DDD0
postgres 14431 13482  0 16:55 pts/1    00:00:00 ps -ef
postgres 14432 13482  0 16:55 pts/1    00:00:00 grep --color=auto postgres
[postgres@pgprimary data]$

启动之后,发现新备库上开始有wal receiver process 。

同时,登录新备库,查看t表的数据有没有同步:

[postgres@pgprimary data]$ psql
psql (9.4.10)
Type "help" for help.

postgres=# select * from t;
 id 
----
(0 rows)

postgres=#

显然,之前从新主库里删除的那条数据,在这个新备库里也看不到了。说明,新备库是OK的了。

3 再次执行主备切换

a 停掉新主库(原备库)

 [postgres@pgstandby data]$ ps -ef|grep postgres
postgres 16230     1  0 15:57 pts/1    00:00:00 /postgres/9.4.10/bin/postgres
postgres 16232 16230  0 15:57 ?        00:00:00 postgres: checkpointer process  
postgres 16233 16230  0 15:57 ?        00:00:00 postgres: writer process   
postgres 16234 16230  0 15:57 ?        00:00:00 postgres: stats collector process  
postgres 16998 16230  0 16:10 ?        00:00:00 postgres: wal writer process  
postgres 16999 16230  0 16:10 ?        00:00:00 postgres: autovacuum launcher process  
postgres 17546 16230  0 16:20 ?        00:00:00 postgres: postgres postgres 172.16.135.72(54471) idle
postgres 19262 16230  0 16:55 ?        00:00:00 postgres: wal sender process replica 172.16.11.148(53298) streaming 0/508DEA8
postgres 19520 29607  0 17:00 pts/1    00:00:00 ps -ef
postgres 19521 29607  0 17:00 pts/1    00:00:00 grep --color=auto postgres
root     29606 29605  0 09:23 pts/1    00:00:00 su - postgres
postgres 29607 29606  0 09:23 pts/1    00:00:00 -bash
[postgres@pgstandby data]$ pg_ctl stop -m fast -l /home/postgres/pg.log 
waiting for server to shut down.... done
server stopped
[postgres@pgstandby data]$ ll
总用量 68
-rw-------. 1 postgres postgres   206 10月 10 16:38 backup_label.old
drwx------. 5 postgres postgres    41 10月 10 16:38 base
drwx------. 2 postgres postgres  4096 10月 26 15:57 global
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_clog
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_dynshmem
-rw-------. 1 postgres postgres  4605 10月 16 16:35 pg_hba.conf
-rw-------. 1 postgres postgres  1636 10月 10 16:38 pg_ident.conf
drwx------. 4 postgres postgres    39 10月 10 16:38 pg_logical
drwx------. 4 postgres postgres    36 10月 10 16:38 pg_multixact
drwx------. 2 postgres postgres    18 10月 26 15:57 pg_notify
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_replslot
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_serial
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_snapshots
drwx------. 2 postgres postgres    63 10月 26 17:00 pg_stat
drwx------. 2 postgres postgres     6 10月 26 17:00 pg_stat_tmp
drwx------. 2 postgres postgres    18 10月 10 16:38 pg_subtrans
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_tblspc
drwx------. 2 postgres postgres     6 10月 10 16:38 pg_twophase
-rw-------. 1 postgres postgres     4 10月 10 16:38 PG_VERSION
drwx------. 3 postgres postgres  4096 10月 26 16:10 pg_xlog
-rw-------. 1 postgres postgres    88 10月 10 16:38 postgresql.auto.conf
-rw-------. 1 postgres postgres 20810 10月 10 16:45 postgresql.conf
-rw-------. 1 postgres postgres    30 10月 26 15:57 postmaster.opts
-rw-r--r--. 1 postgres postgres  5641 10月 16 16:54 recovery.done
[postgres@pgstandby data]$

需要注意的是,这里的新主库停止之后,系统同样自动的将/postgres/9.4.10/data/recovery.conf文件变为/postgres/9.4.10/data/recovery.done。在我们重新启动这个备库的时候,需要把recovery.done改名为recovery.conf,使其在重新启动时,自己担任的是“备库角色”,自动同主库进行同步。

b 启动新备库(原主库)

[postgres@pgprimary data]$ ps -ef|grep postgres
root     13481 13458  0 09:22 pts/1    00:00:00 su - postgres
postgres 13482 13481  0 09:22 pts/1    00:00:00 -bash
postgres 14425     1  0 16:55 pts/1    00:00:00 /postgres/9.4.10/bin/postgres
postgres 14426 14425  0 16:55 ?        00:00:00 postgres: startup process   recovering 000000060000000000000005
postgres 14427 14425  0 16:55 ?        00:00:00 postgres: checkpointer process  
postgres 14428 14425  0 16:55 ?        00:00:00 postgres: writer process   
postgres 14429 14425  0 16:55 ?        00:00:00 postgres: stats collector process  
postgres 14894 13482  0 17:01 pts/1    00:00:00 ps -ef
postgres 14895 13482  0 17:01 pts/1    00:00:00 grep --color=auto postgres
[postgres@pgprimary data]$ pg_ctl promote -l /home/postgres/
server promoting
[postgres@pgprimary data]$ ps -ef|grep postgres
root     13481 13458  0 09:22 pts/1    00:00:00 su - postgres
postgres 13482 13481  0 09:22 pts/1    00:00:00 -bash
postgres 14425     1  0 16:55 pts/1    00:00:00 /postgres/9.4.10/bin/postgres
postgres 14427 14425  0 16:55 ?        00:00:00 postgres: checkpointer process  
postgres 14428 14425  0 16:55 ?        00:00:00 postgres: writer process   
postgres 14429 14425  0 16:55 ?        00:00:00 postgres: stats collector process  
postgres 14916 14425  0 17:01 ?        00:00:00 postgres: wal writer process  
postgres 14917 14425  0 17:01 ?        00:00:00 postgres: autovacuum launcher process  
postgres 14918 13482  0 17:01 pts/1    00:00:00 ps -ef
postgres 14919 13482  0 17:01 pts/1    00:00:00 grep --color=auto postgres
[postgres@pgprimary data]$

此时,这个库又恢复到最初的状态,成为主库。

c 启动原备库

在启动之前,修改recovery.done改名为recovery.conf。

 [postgres@pgstandby data]$ pwd
/postgres/9.4.10/data
[postgres@pgstandby data]$ mv recovery.done recovery.conf
[postgres@pgstandby data]$

启动备库:

 [postgres@pgstandby data]$ pg_ctl start -l /home/postgres/pg.log 
server starting
[postgres@pgstandby data]$ ps -ef|grep postgres
postgres 19871     1  0 17:07 pts/1    00:00:00 /postgres/9.4.10/bin/postgres
postgres 19872 19871  0 17:07 ?        00:00:00 postgres: startup process   recovering 000000070000000000000005
postgres 19873 19871  0 17:07 ?        00:00:00 postgres: checkpointer process  
postgres 19874 19871  0 17:07 ?        00:00:00 postgres: writer process   
postgres 19875 19871  0 17:07 ?        00:00:00 postgres: stats collector process  
postgres 19876 19871  0 17:07 ?        00:00:00 postgres: wal receiver process   streaming 0/508E108
postgres 19881 29607  0 17:07 pts/1    00:00:00 ps -ef
postgres 19882 29607  0 17:07 pts/1    00:00:00 grep --color=auto postgres
root     29606 29605  0 09:23 pts/1    00:00:00 su - postgres
postgres 29607 29606  0 09:23 pts/1    00:00:00 -bash
[postgres@pgstandby data]$

启动之后,看到有wal receiver process。说明,此时备库依然是备库的角色了,并开始同主库进行同步了。

至此,完成了对于主备库的相互切换,并且验证了数据同步,以及注意事项。

七 小结

本文档对于在CentOS7+PostgreSQL9.4.10的数据库环境,利用浮动IP来构建一套数据库的主备同步环境,给出详细搭建和配置的流程。通过自动监控启停浮动IP来实现网络的畅通。正常情况下,主库提供读写操作,备库提供只读操作。客户端和应用程序应该通过浮动IP来访问数据库,而不应该使用真实IP访问数据库。

 

发表评论

邮箱地址不会被公开。 必填项已用*标注