使用pg_basebackup搭建PostgreSQL流复制环境

环境：
    OS:

[ha@node0 ~]$ uname -a
Linux node0 2.6.32-358.el6.x86_64 #1 SMP Fri Feb 22 00:31:26 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux

    内存：1G
    CPU：1核
    数据库:

postgres=# select version();
                                                    version                     
                               
--------------------------------------------------------------------------------
-------------------------------
 PostgreSQL 9.4.5 on x86_64-unknown-linux-gnu, compiled by gcc (GCC) 4.4.7 20120
313 (Red Hat 4.4.7-16), 64-bit
(1 row)

postgres=#

    .bash_profile

[ha@node0 ~]$ cat .bash_profile 
# .bash_profile

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

# User specific environment and startup programs

export PGHOME=/home/ha/pgdb/
export PGDATA=/home/ha/pgdb/data/
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$PGHOME/lib/
export PATH=$PGHOME/bin:$PATH:$HOME/bin

    搭建过程：
    1.主库配置
        a.postgresql.conf：

listen_addresses = '*'
wal_level = hot_standby
archive_mode = off
max_wal_senders = 3
wal_keep_segments = 16

        这里要说明的是参数archive_mode，我在实验过程中有一次发现参数archive_mode和archive_command在不设置的情况下也可以完成流复制，所以我对此进行了调研，发现francs已经对此进行了解释，这里引用francs的结论：

在搭建流复制环境时，并不必须设置 archive_mode 参数为 on ，很多资料在介绍搭建流复制环境时设置这个参数为 on ，可能是出于开启 WAL 归档更安全的原因，因为在主库宕机并且较长时间不能恢复时，从库依然可以读取归档目录的 WAL，从而保证不丢数据; 另一方面，如果主库设置了较大的 wal_keep_segments ，也可以不用开启archive_mode，因为主库保留了足够的 WAL，从而大大降低了因从库所需要的 WAL 被主库覆盖而需要从归档去取 WAL 的情况。所以从这方面说，archive_mode 参数的设置与搭建流复制并没有直接关系。 提示： 对于比较繁忙的库，在搭建流复制从库时，建议主库设置较大的 wal_keep_segments 参数。

        b.pg_hba.conf

host    all             all             192.168.238.0/24        md5
host    replication     rep             192.168.238.0/24        trust

        这里说明的是，网上有其他对replication设置为md5，并建立.pgpass文件，这种方法也是可行的。

        c.新建用户rep

create user rep replication encrypted password 'rep';

    2.使用pg_basebackup进行从库备份

[ha@localhost pgdb]$ pg_basebackup -D $PGDATA -F p -X stream -v -P -h 192.168.238.130 -U rep
transaction log start point: 0/2000028 on timeline 1
pg_basebackup: starting background WAL receiver
20945/20945 kB (100%), 1/1 tablespace                                         
transaction log end point: 0/20000F0
pg_basebackup: waiting for background process to finish streaming ...
pg_basebackup: base backup completed
[ha@localhost pgdb]$ ll
total 20
drwxrwxr-x.  2 ha ha 4096 Dec 26 00:54 bin
drwx------. 18 ha ha 4096 Dec 26 12:02 data
drwxrwxr-x.  4 ha ha 4096 Dec 26 00:54 include
drwxrwxr-x.  4 ha ha 4096 Dec 26 00:54 lib
drwxrwxr-x.  4 ha ha 4096 Dec 26 00:54 share
[ha@localhost pgdb]$ cd data/
[ha@localhost data]$ ls
backup_label  pg_hba.conf    pg_replslot   pg_subtrans  postgresql.auto.conf
base          pg_ident.conf  pg_serial     pg_tblspc    postgresql.conf
global        pg_logical     pg_snapshots  pg_twophase
pg_clog       pg_multixact   pg_stat       PG_VERSION
pg_dynshmem   pg_notify      pg_stat_tmp   pg_xlog

    这里需要对目录进行一下说明：data目录可以通过pg_basebackup自动创建，但是其他表空间，需要手动创建并赋权，这里需要注意的是，权限为700。

    3.从库配置
        a.postgresql.conf

hot_standby = on

        b.recovery.conf

standby_mode = on
primary_conninfo = 'host=192.168.238.130 port=5432 user=rep'
trigger_file = '/home/ha/pgdb/pg.trigger.file'

    4.备库启动前，主库进程

[ha@localhost pgdb]$ ps -ef | grep post
root       2124      1  0 04:47 ?        00:00:00 /usr/libexec/postfix/master
postfix    2147   2124  0 04:47 ?        00:00:00 qmgr -l -t fifo -u
postfix   10385   2124  0 11:27 ?        00:00:00 pickup -l -t fifo -u
ha        10691      1  0 12:33 pts/3    00:00:00 /home/ha/pgdb/bin/postgres_ora
ha        10693  10691  0 12:33 ?        00:00:00 postgres: checkpointer process   
ha        10694  10691  0 12:33 ?        00:00:00 postgres: writer process      
ha        10695  10691  0 12:33 ?        00:00:00 postgres: wal writer process   
ha        10696  10691  0 12:33 ?        00:00:00 postgres: autovacuum launcher process   
ha        10697  10691  0 12:33 ?        00:00:00 postgres: stats collector process   
ha        10717   4087  0 12:37 pts/3    00:00:00 grep post

    5.备库启动后
        a.主库进程

[ha@localhost pgdb]$ ps -ef | grep post
root       2124      1  0 04:47 ?        00:00:00 /usr/libexec/postfix/master
postfix    2147   2124  0 04:47 ?        00:00:00 qmgr -l -t fifo -u
postfix   10385   2124  0 11:27 ?        00:00:00 pickup -l -t fifo -u
ha        10691      1  0 12:33 pts/3    00:00:00 /home/ha/pgdb/bin/postgres_ora
ha        10693  10691  0 12:33 ?        00:00:00 postgres: checkpointer process   
ha        10694  10691  0 12:33 ?        00:00:00 postgres: writer process      
ha        10695  10691  0 12:33 ?        00:00:00 postgres: wal writer process   
ha        10696  10691  0 12:33 ?        00:00:00 postgres: autovacuum launcher process   
ha        10697  10691  0 12:33 ?        00:00:00 postgres: stats collector process   
ha        10718  10691  0 12:37 ?        00:00:00 postgres: wal sender process rep 192.168.238.131(59195) streaming 0/3000060
ha        10720   4087  0 12:37 pts/3    00:00:00 grep post

        b.备库进程

[ha@localhost data]$ ps -ef | grep post
root       2086      1  0 04:48 ?        00:00:00 /usr/libexec/postfix/master
postfix    2108   2086  0 04:48 ?        00:00:00 qmgr -l -t fifo -u
postfix    9657   2086  0 11:27 ?        00:00:00 pickup -l -t fifo -u
ha         9782      1  0 12:36 pts/2    00:00:00 /home/ha/pgdb/bin/postgres
ha         9783   9782  0 12:36 ?        00:00:00 postgres: startup process   recovering 000000010000000000000003
ha         9784   9782  0 12:36 ?        00:00:00 postgres: checkpointer process   
ha         9785   9782  0 12:36 ?        00:00:00 postgres: writer process   
ha         9786   9782  0 12:36 ?        00:00:00 postgres: stats collector process   
ha         9787   9782  0 12:36 ?        00:00:00 postgres: wal receiver process   streaming 0/3000060
ha         9792   3744  0 12:37 pts/2    00:00:00 grep post

    6.实验效果
        a.主库

[ha@localhost pgdb]$ psql postgres
psql (9.4.5)
Type "help" for help.

postgres=# create table test(id int);
CREATE TABLE
postgres=# insert into test values (1),(2);
INSERT 0 2
postgres=# select * from test ;
 id 
----
  1
  2
(2 rows)

postgres=#

        b.备库

[ha@localhost data]$ psql postgres
psql (9.4.5)
Type "help" for help.

postgres=# select * from test ;
 id 
----
  1
  2
(2 rows)