基于streaming replication搭建的PostgreSQL HA环境,Old Standby节点升级为New Master节点后,时间线会切换为新的时间线,比如从n变为n + 1.而Old Master节点的时间线仍然为原来的时间线,比如仍为n,通过使用pg_rewind工具,可使原来其实”完好”的Old Master成为New Standby节点.
图解
如下图所示:
pg_rewind图解
在执行主备切换后,New Master节点的时间线切换为n + 1,通过pg_rewind可使Old Master在分叉点开始与New Master同步,成为New Standby节点.

实测
New Master
切换后为New Master后,执行以下SQL:


testdb=# create table t_new(id int,flag varchar(40));
CREATE TABLE
testdb=# insert into t_new select c,'flag'||c from generate_series(1,1000000) as c;
INSERT 0 1000000
testdb=#

Old Master
执行pg_rewind前,重启主库,执行以下SQL


testdb=# create table t_fork(id int,flag varchar(40));
CREATE TABLE
testdb=# insert into t_fork select c,'flag'||c from generate_series(1,1000000) as c;
INSERT 0 1000000
testdb=#

执行pg_rewind


[xdb@localhost testdb]$ cp /data/archivelog/* ./pg_wal
[xdb@localhost testdb]$ pg_rewind --target-pgdata=$PGDATA --source-server="host=192.168.26.25 port=5432 dbname=testdb" --progress
connected to server
servers diverged at WAL location 0/B41F12B8 on timeline 23
rewinding from last common checkpoint at 0/AFCF99E0 on timeline 23
reading source file list
reading target file list
reading WAL in target
need to copy 360 MB (total source directory size is 501 MB)
369312/369312 kB (100%) copied
creating backup label and updating control file
syncing target data directory
Done!

配置recovery.conf文件


[xdb@localhost testdb]$ mv recovery.done recovery.conf
[xdb@localhost testdb]$ vim recovery.conf 
[xdb@localhost testdb]$ cat recovery.conf 
standby_mode = 'on'
primary_conninfo = 'user=replicator password=replicator host=192.168.26.26 port=5432 sslmode=prefer sslcompression=0 krbsrvname=postgres target_session_attrs=any'
restore_command = 'cp /data/archivelog/%f %p'

重启数据库


[xdb@localhost testdb]$ pg_ctl start
waiting for server to start....2019-03-28 12:39:31.918 CST [1961] LOG:  listening on IPv4 address "0.0.0.0", port 5432
2019-03-28 12:39:31.918 CST [1961] LOG:  listening on IPv6 address "::", port 5432
2019-03-28 12:39:31.920 CST [1961] LOG:  listening on Unix socket "/tmp/.s.PGSQL.5432"
2019-03-28 12:39:31.970 CST [1961] LOG:  redirecting log output to logging collector process
2019-03-28 12:39:31.970 CST [1961] HINT:  Future log output will appear in directory "pg_log".
... done
server started

与New Master数据同步了,而在原TL上的t_fork数据表消失了.


testdb=# select count(*) from t_new;
  count  
---------
 1000000
(1 row)
testdb=# select count(*) from t_old;
  count  
---------
 1000000
(1 row)
testdb=# select count(*) from t_fork;
ERROR:  relation "t_fork" does not exist
LINE 1: select count(*) from t_fork;
                             ^
testdb=#

PostgreSQL DBA(33) - HA#2(pg_rewind切换图解)