Redhat LINUX 9.3 + PG 16.1 搭建主备流复制

郑全

1.环境：

1）IP:
主：192.168.133.151 server1
备：192.168.133.152 server2


2）操作系统版本：
[postgres@server2 data]$ cat /etc/redhat-release
Red Hat Enterprise Linux release 9.3 (Plow)



3）数据库版本：

postgres@[local]:5432/postgres-11653#=select version();
                                                    version                                                     
----------------------------------------------------------------------------------------------------------------
PostgreSQL 16.1 on x86_64-pc-linux-gnu, compiled by gcc (GCC) 11.4.1 20230605 (Red Hat 11.4.1-2.1.0.1), 64-bit
(1 row)



2.创建主备流复制

2.1 在主备两台机器修改 /etc/hosts 文件

vim /etc/hosts

192.168.133.151 server1
192.168.133.152 server2

2.2 在主库设置
1）建立同步用户

postgres=# create role rep1 login replication encrypted password 'rep1';

2) 配置复制认证配置

vim $PGDATA/pg_hba.conf

host    replication     rep1            server2                 md5


3) 修改$PGDATA/postgres.conf

listen_addresses = '*'
port=5432
wal_level = replica
max_wal_senders=10
archive_mode = on
archive_timeout=1800

archive_command = 'cp %p /home/postgres/arch/%f'
restore_command = 'cp /home/postgres/arch/%f %p'
recovery_target_timeline = 'latest'
full_page_writes = on
wal_log_hints = on

logging_collector = on
log_directory = 'pg_log'
log_filename = 'postgresql-%Y-%m-%d.log'

--下面参数设置大一些，否则，后面主备切换，可能出现找不到日志文件的问题
min_wal_size = 800MB
wal_keep_size = 1024

4).重启主库，让参数生效

pg_ctl restart

2.3 备库设置

1）初始化，直接从主库备份就行

pg_basebackup -h server1 -p 5432 -U repl -X stream -R -Fp -P -D /pgdata/data

-h：指定要连接的服务器的主机名或IP地址。
-U：指定连接使用的用户名。
-Fp：指定备份的格式。在这种情况下，备份以纯文本格式存储。
-P：指定在备份过程中显示进度信息。
-X stream：指定备份的流式传输方法。
-R：指定包括所有必需的WAL文件以进行一致性备份。
-D：指定备份存储的目录。

2) 修改参数，加上下面参数

hot_standby = on #在备份的同时允许查询，默认值
max_standby_streaming_delay = 30s #可选，流复制最大延迟
wal_receiver_status_interval = 10s #可选，从向主报告状态的最大间隔时间
hot_standby_feedback = on #可选，查询冲突时向主反馈


3）配置$PGDATA/data/pg_hba.conf
host    replication     rep1            server1                 md5 #这里修改主机名为主机的

4) 重启让配置生效
pg_ctl start


3.验证

3.1 查看主备进程

1）主库端

[postgres@server1 data]$ ps -ef|grep wal
postgres    3003    2998  0 21:57 ?        00:00:00 postgres: walwriter
postgres    3063    2998  0 22:08 ?        00:00:00 postgres: walsender rep1 192.168.133.152(49190) streaming 0/260021C8
postgres    3080    1518  0 22:13 pts/0    00:00:00 grep --color=auto wal
[postgres@server1 data]$

2）备库端

[postgres@server2 data]$ ps -ef|grep wal
postgres   11355   11345  0 22:08 ?        00:00:00 postgres: walreceiver streaming 0/260021C8
postgres   11373    9351  0 22:14 pts/0    00:00:00 grep --color=auto wal
[postgres@server2 data]$

3.2 主库切换日志

postgres@[local]:5432/postgres-3100#=select pg_switch_wal();
pg_switch_wal
---------------
0/260021E0
(1 row)

postgres@[local]:5432/postgres-3100#=\! ls -ltr /pgdata/data/pg_wal
总用量 458768
-rw------- 1 postgres postgres       41 12月 15 23:16 00000002.history
-rw------- 1 postgres postgres 16777216 12月 17 20:09 00000002000000000000000D
-rw------- 1 postgres postgres 16777216 12月 17 20:22 00000002000000000000000E
-rw------- 1 postgres postgres 16777216 12月 17 20:22 00000002000000000000000F
-rw------- 1 postgres postgres 16777216 12月 17 20:25 000000020000000000000010
-rw------- 1 postgres postgres 16777216 12月 17 20:28 000000020000000000000011
-rw------- 1 postgres postgres 16777216 12月 17 20:28 000000020000000000000012.partial
-rw------- 1 postgres postgres       84 12月 17 20:29 00000003.history
-rw------- 1 postgres postgres 16777216 12月 17 20:34 000000030000000000000012
-rw------- 1 postgres postgres 16777216 12月 17 21:04 000000030000000000000013
-rw------- 1 postgres postgres 16777216 12月 17 21:26 000000030000000000000014
-rw------- 1 postgres postgres 16777216 12月 17 21:29 000000030000000000000015
-rw------- 1 postgres postgres 16777216 12月 17 21:29 000000030000000000000016
-rw------- 1 postgres postgres 16777216 12月 17 21:30 000000030000000000000017
-rw------- 1 postgres postgres 16777216 12月 17 21:30 000000030000000000000018
-rw------- 1 postgres postgres 16777216 12月 17 21:31 000000030000000000000019
-rw------- 1 postgres postgres 16777216 12月 17 21:31 00000003000000000000001A
-rw------- 1 postgres postgres 16777216 12月 17 21:33 00000003000000000000001B
-rw------- 1 postgres postgres 16777216 12月 17 21:33 00000003000000000000001C
-rw------- 1 postgres postgres 16777216 12月 17 21:47 00000003000000000000001D
-rw------- 1 postgres postgres 16777216 12月 17 21:47 00000003000000000000001E
-rw------- 1 postgres postgres 16777216 12月 17 21:50 00000003000000000000001F
-rw------- 1 postgres postgres 16777216 12月 17 21:50 000000030000000000000020
-rw------- 1 postgres postgres 16777216 12月 17 21:53 000000030000000000000021
-rw------- 1 postgres postgres 16777216 12月 17 21:53 000000030000000000000022
-rw------- 1 postgres postgres 16777216 12月 17 21:57 000000030000000000000023
-rw------- 1 postgres postgres 16777216 12月 17 21:58 000000030000000000000024
-rw------- 1 postgres postgres 16777216 12月 17 21:58 000000030000000000000025
-rw------- 1 postgres postgres      341 12月 17 21:58 000000030000000000000025.00000028.backup
-rw------- 1 postgres postgres 16777216 12月 17 22:17 000000030000000000000026
drwx------ 2 postgres postgres     4096 12月 17 22:17 archive_status
-rw------- 1 postgres postgres 16777216 12月 17 22:17 000000030000000000000027
postgres@[local]:5432/postgres-3100#=

--备库

[postgres@server2 data]$ ls -ltr /pgdata/data/pg_wal
总用量 49156
-rw------- 1 postgres postgres       84 12月 17 21:58 00000003.history
-rw------- 1 postgres postgres 16777216 12月 17 21:58 000000030000000000000025
drwx------ 2 postgres postgres       72 12月 17 22:17 archive_status
-rw------- 1 postgres postgres 16777216 12月 17 22:17 000000030000000000000026
-rw------- 1 postgres postgres 16777216 12月 17 22:17 000000030000000000000027

3.3 查看当前备库状态

[postgres@server2 data]$ psql
psql (16.1)
Type "help" for help.

postgres@server2:5432/postgres-11485#=select pg_is_in_recovery();
pg_is_in_recovery
-------------------
t
(1 row)


postgres@server1:5432/postgres-3210#=select pg_is_in_recovery();
pg_is_in_recovery
-------------------
f
(1 row)


t ：true，意味着处于 recovery 状态
f ：false，意味着处于正常服务状态


3.4 主库查询

postgres@server1:5432/postgres-3210#=select pid, usename,client_addr,state,sync_state,reply_time from pg_stat_replication;
pid  | usename |   client_addr   |   state   | sync_state |          reply_time           
------+---------+-----------------+-----------+------------+-------------------------------
3188 | rep1    | 192.168.133.152 | streaming | async      | 2023-12-17 22:31:28.775596+08
(1 row)

postgres@server1:5432/postgres-3210#=select pid, usename,client_addr,state,sync_state,reply_time from pg_stat_replication;
-[ RECORD 1 ]------------------------------
pid         | 3188
usename     | rep1
client_addr | 192.168.133.152
state       | streaming
sync_state  | async
reply_time  | 2023-12-17 22:31:38.761116+08

sync_state表示同步模式
sent_lsn表示发送日志的起点
reply_time表示应用日志的起点

postgres@server1:5432/postgres-3210#=select * from pg_stat_replication;
-[ RECORD 1 ]----+------------------------------
pid              | 3188
usesysid         | 16409
usename          | rep1
application_name | walreceiver
client_addr      | 192.168.133.152
client_hostname  | server2
client_port      | 59278
backend_start    | 2023-12-17 22:25:07.174465+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/28000C88
write_lsn        | 0/28000C88
flush_lsn        | 0/28000C88
replay_lsn       | 0/28000C88
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 0
sync_state       | async
reply_time       | 2023-12-17 22:33:18.809739+08


3.5 备库情况

postgres@server2:5432/postgres-11485#=select * from pg_stat_wal_receiver;-[ RECORD 1 ]---------+------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
pid                   | 11386
status                | streaming
receive_start_lsn     | 0/28000000
receive_start_tli     | 3
written_lsn           | 0/28000C88
flushed_lsn           | 0/28000C88
received_tli          | 3
last_msg_send_time    | 2023-12-17 22:35:08.850593+08
last_msg_receipt_time | 2023-12-17 22:35:08.85238+08
latest_end_lsn        | 0/28000C88
latest_end_time       | 2023-12-17 22:30:08.735565+08
slot_name             |
sender_host           | server1
sender_port           | 5432
conninfo              | user=rep1 password=******** channel_binding=disable dbname=replication host=server1 port=5432 client_encoding=GBK fallback_application_name=walreceiver sslmode=disable sslcompression=0 sslcertmode=disable sslsni=1 ssl_min_protocol_version=TLSv1.2 gssencmode=disable krbsrvname=postgres gssdelegation=0 target_session_attrs=any load_balance_hosts=disable



--备库日志

023-12-17 22:25:07.215 CST [11386] LOG:  started streaming WAL from primary at 0/28000000 on timeline 3
2023-12-17 22:25:08.324 CST [11347] LOG:  restartpoint starting: time
2023-12-17 22:25:08.331 CST [11347] LOG:  restartpoint complete: wrote 0 buffers (0.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.001 s, sync=0.001 s, total=0.007 s; sync files=0, longest=0.000 s, average=0.000 s; distance=16383 kB, estimate=16389 kB; lsn=0/28000028, redo lsn=0/28000028
2023-12-17 22:25:08.331 CST [11347] LOG:  recovery restart point at 0/28000028
2023-12-17 22:30:08.225 CST [11347] LOG:  restartpoint starting: time
2023-12-17 22:30:08.332 CST [11347] LOG:  restartpoint complete: wrote 2 buffers (0.0%); 0 WAL file(s) added, 0 removed, 0 recycled; write=0.103 s, sync=0.001 s, total=0.107 s; sync files=2, longest=0.001 s, average=0.001 s; distance=2 kB, estimate=14751 kB; lsn=0/28000BD8, redo lsn=0/28000BA0
2023-12-17 22:30:08.332 CST [11347] LOG:  recovery restart point at 0/28000BA0
2023-12-17 22:30:08.332 CST [11347] DETAIL:  Last completed transaction was at log time 2023-12-17 22:27:29.986132+08.


3.6 测试数据同步

1）主库创建表：

postgres@server1:5432/postgres-3210#=create table emp(emp int);
CREATE TABLE
postgres@server1:5432/postgres-3210#=insert into emp values(100);
INSERT 0 1
postgres@server1:5432/postgres-3210#=select * from emp;
emp
-----
100
(1 row)

2） 备库验证

[postgres@server2 pg_log]$ psql -h server2
Password for user postgres:
psql (16.1)
Type "help" for help.

postgres@server2:5432/postgres-11497#=\d
             List of relations
Schema |      Name      | Type  |  Owner   
--------+----------------+-------+----------
public | emp            | table | postgres
public | pg_buffercache | view  | postgres
(2 rows)

postgres@server2:5432/postgres-11497#=select * from emp;
emp
-----
100
(1 row)

通过以上可以看出，主库新增数据已经同步到备库

4.主备切换

4.1 停止主库

[postgres@server1 data]$ pg_ctl stop
waiting for server to shut down.... done
server stopped
[postgres@server1 data]$
[postgres@server1 data]$
[postgres@server1 data]$ pg_ctl status
pg_ctl: no server running

4.2 提升备库为主

pg_ctl promote

postgres@server2 pg_log]$ pg_ctl promote
waiting for server to promote.... done
server promoted
[postgres@server2 pg_log]$ pg_ctl status
pg_ctl: server is running (PID: 11345)
/u01/app/postgres/bin/postgres
[postgres@server2 pg_log]$ pg_controldata |grep stat
Database cluster state:               in production
[postgres@server2 pg_log]$

[postgres@server2 data]$ ls -ltr standby*
ls: 无法访问 'stand*': 没有那个文件或目录
[postgres@server2 data]$

[postgres@server2 data]$ ps -ef|grep wal
postgres   11569   11345  0 22:44 ?        00:00:00 postgres: walwriter
postgres   11592    9351  0 22:46 pts/0    00:00:00 grep --color=auto wal
[postgres@server2 data]$

4.3 在新备库上(原主库)创建一个 standby.signal文件

[postgres@server1 data]$ touch standby.signal
[postgres@server1 data]$
[postgres@server1 data]$ ls -ltr standby.signal
-rw-r--r-- 1 postgres postgres 0 12月 17 22:49 standby.signal


4.4 在新备库的 postgresql.auto.conf 文件中添加如下内容：

primary_conninfo = 'user=rep1 password=rep1 passfile=''/home/postgres/.pgpass'' channel_binding=disable host=server2 port=5432 client_encoding=GBK sslmode=disable sslcompression=0 sslcertmode=disable sslsni=1 ssl_min_protocol_version=TLSv1.2 gssencmode=disable krbsrvname=postgres gssdelegation=0 target_session_attrs=any load_balance_hosts=disable'



4.5 启动新备库
[postgres@server1 data]$ pg_ctl restart
waiting for server to shut down.... done
server stopped
waiting for server to start....2023-12-17 22:53:08.352 CST [3340] LOG:  redirecting log output to logging collector process
2023-12-17 22:53:08.352 CST [3340] HINT:  Future log output will appear in directory "pg_log".
done
server started

4.6 验证
新备库
[postgres@server1 data]$ pg_controldata |grep state
Database cluster state:               in archive recovery

新主库
[postgres@server2 data]$ pg_controldata |grep stat
Database cluster state:               in production


[postgres@server2 data]$ psql
psql (16.1)
Type "help" for help.

postgres@[local]:5432/postgres-11653#=\x
Expanded display is on.
postgres@[local]:5432/postgres-11653#=select pid, usename,client_addr,state,sync_state,reply_time from pg_stat_replication;
-[ RECORD 1 ]------------------------------
pid         | 11644
usename     | rep1
client_addr | 192.168.133.151
state       | streaming
sync_state  | async
reply_time  | 2023-12-17 22:55:48.476883+08

postgres@[local]:5432/postgres-11653#=select * from pg_stat_replication;
-[ RECORD 1 ]----+------------------------------
pid              | 11644
usesysid         | 16409
usename          | rep1
application_name | walreceiver
client_addr      | 192.168.133.151
client_hostname  | server1
client_port      | 47196
backend_start    | 2023-12-17 22:53:08.395347+08
backend_xmin     |
state            | streaming
sent_lsn         | 0/290001B8
write_lsn        | 0/290001B8
flush_lsn        | 0/290001B8
replay_lsn       | 0/290001B8
write_lag        |
flush_lag        |
replay_lag       |
sync_priority    | 0
sync_state       | async
reply_time       | 2023-12-17 22:56:18.489045+08


到此，切换完成。