概念描述
背景信息
当数据库崩溃或希望回退到数据库之前的某一状态时，MogDB的即时恢复功能（Point-In-Time Recovery，简称PITR）可以支持恢复到备份归档数据之后的任意时间点。
说明:
PITR仅支持恢复到物理备份数据之后的某一时间点。
仅主节点可以进行PITR恢复，备机需要进行全量build达成与主机数据同步。
前提条件
基于经过物理备份的全量数据文件。
基于已归档的WAL日志文件。
PITR恢复流程
将物理备份的文件替换目标数据库目录。
删除数据库目录下pg_xlog/中的所有文件。
将归档的WAL日志文件复制到pg_xlog文件中（此步骤可以省略，通过配置recovery.conf恢复命令文件中的restore_command项替代）。
在数据库目录下创建恢复命令文件recovery.conf，指定数据库恢复的程度。
启动数据库。
连接数据库，查看是否恢复到希望预期的状态。
若已经恢复到预期状态，通过pg_xlog_replay_resume()指令使主节点对外提供服务。
恢复目标设置(四选一)
recovery_target_name = ‘restore_point_1’ ## 还原到一个使用pg_create_restore_point()创建的还原点
recovery_target_time = ‘2020-01-01 12:00:00’ ## 还原到一个指定时间戳
recovery_target_xid = ‘3000’ ## 还原到一个事务ID
recovery_target_lsn = ‘0/0FFFFFF’ ## 还原到日志的指定LSN点
recovery_target_inclusive = true ## 声明是否在指定恢复目标之后停止(true) 或 之前停止(false),不支持 recovery_target_name 配置#注意：如果不配置任何恢复目标 或 配置目标不存在，则默认恢复到最新的WAL日志点。
测试验证
1、环境准备(gs_baasebackup 备份)
[omm@db1 db1]$ gs_basebackup -U em_ljc -W -h 192.168.3.24 -p 26000 -D /home/omm/gs_pitr
Password:
INFO: The starting position of the xlog copy of the full build is: 4/EA000028. The slot minimum LSN is: 0/0.
[2022-10-29 14:18:24]:begin build tablespace list
[2022-10-29 14:18:24]:finish build tablespace list
[2022-10-29 14:18:24]:begin get xlog by xlogstream
[2022-10-29 14:18:24]: check identify system success
[2022-10-29 14:18:24]: send START_REPLICATION 4/EA000000 success
[2022-10-29 14:18:24]: keepalive message is received
[2022-10-29 14:18:24]: keepalive message is received
[2022-10-29 14:18:27]: keepalive message is received
[2022-10-29 14:18:30]: keepalive message is received
[2022-10-29 14:18:30]: keepalive message is received
[2022-10-29 14:18:33]: keepalive message is received
[2022-10-29 14:18:36]: keepalive message is received
[2022-10-29 14:18:36]: keepalive message is received
[2022-10-29 14:18:39]: keepalive message is received
[2022-10-29 14:18:42]: keepalive message is received
[2022-10-29 14:18:42]: keepalive message is received
[2022-10-29 14:18:45]: keepalive message is received
[2022-10-29 14:18:47]: keepalive message is received
[2022-10-29 14:18:50]: keepalive message is received
[2022-10-29 14:18:53]: keepalive message is received
[2022-10-29 14:18:56]: keepalive message is received
[2022-10-29 14:18:59]: keepalive message is received
[2022-10-29 14:18:59]: keepalive message is received
[2022-10-29 14:19:02]: keepalive message is received
[2022-10-29 14:19:05]: keepalive message is received
[2022-10-29 14:19:05]: keepalive message is received
[2022-10-29 14:19:08]: keepalive message is received
[2022-10-29 14:19:11]: keepalive message is received
[2022-10-29 14:19:11]: keepalive message is received
[2022-10-29 14:19:14]: keepalive message is received
[2022-10-29 14:19:17]: keepalive message is received
[2022-10-29 14:19:17]: keepalive message is received
[2022-10-29 14:19:20]: keepalive message is received
[2022-10-29 14:19:23]: keepalive message is received
[2022-10-29 14:19:23]: keepalive message is received
[2022-10-29 14:19:29]:gs_basebackup: base backup successfully
[omm@db1 db1]$
2、创建测试数据（还原点recovery_target_name ）
miao=> create table t1 (id int,tm timestamp,LSN varchar(20));
insert into t1 values(1,now(),'Started');
select * from t1;
select * from pg_switch_xlog();
CREATE TABLE
miao=> insert into t1 values(1,now(),'Started');
INSERT 0 1
miao=> select * from t1;
id | tm | lsn
----+----------------------------+---------
1 | 2022-10-29 14:22:52.273009 | Started
(1 row)
miao=> select * from pg_switch_xlog();
pg_switch_xlog
----------------
4/EC0029F8
(1 row)
miao=>
-- 创建一个还原点 restore_point_1
miao=> select pg_create_restore_point('restore_point_1');
pg_create_restore_point
-------------------------
4/ED00EF40
(1 row)
3、第2次插入数据（时间recovery_target_time ）
miao=> insert into t1 values(2,now(),'First Insert');
INSERT 0 1
miao=> select * from t1;
id | tm | lsn
----+----------------------------+--------------
1 | 2022-10-29 14:22:52.273009 | Started
2 | 2022-10-29 14:26:10.55628 | First Insert
(2 rows)
miao=> select pg_switch_xlog();
select now();
pg_switch_xlog
----------------
4/ED017848
(1 row)
miao=> select now();
now
-------------------------------
2022-10-29 14:26:10.754567+08
(1 row)
4、第3次插入数据（LSN recovery_target_lsn ）
miao=> insert into t1 values(3,now(),'Second Insert');
INSERT 0 1
miao=> select * from t1;
id | tm | lsn
----+----------------------------+---------------
1 | 2022-10-29 14:22:52.273009 | Started
2 | 2022-10-29 14:26:10.55628 | First Insert
3 | 2022-10-29 14:31:13.220708 | Second Insert
(3 rows)
miao=> select pg_switch_xlog();
select * from pg_current_xlog_location();
pg_switch_xlog
----------------
4/EE0230C0
(1 row)
miao=>
miao=> select * from pg_current_xlog_location();
pg_current_xlog_location
--------------------------
4/EF000148
(1 row)
5、模拟故障删除目录
[omm@db1 gs_pitr]$ cp /archivelog/* /home/omm/gs_pitr/archivelog/
[omm@db1 gs_pitr]$ gs_om -t stop
Stopping cluster.
=========================================
Successfully stopped cluster.
=========================================
End stop cluster.
[omm@db1 gs_pitr]$ rm -fr /mogdb/data/db1/
6、第一阶段恢复（按照还原点进行恢复）
[omm@db1 data]$ mkdir /mogdb/data/db1/
[omm@db1 db1]$ cp -fr /home/omm/gs_pitr/* /mogdb/data/db1/
[omm@db1 db1]$ vi recovery.conf
restore_command = 'cp /mogdb/data/db1/archivelog/%f %p'
recovery_target_name = 'restore_point_1' ## 恢复到指定的还原点restore_point_1,此时后面2条数据
recovery_target_inclusive = true
[omm@db1 db1]$ gs_om -t start
Starting cluster.
=========================================
[SUCCESS] db1
2022-10-29 15:22:47.061 635cd4c6.1 [unknown] 140295007823424 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: could not create any HA TCP/IP sockets
2022-10-29 15:22:47.064 635cd4c6.1 [unknown] 140295007823424 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: Failed to initialize the memory protect for g_instance.attr.attr_storage.cstore_buffers (1024 Mbytes) or shared memory (1967 Mbytes) is larger.
=========================================
Successfully started.
[omm@db1 db1]$ gsql -d miao -p 26000 -U dbmt -r
Password for user dbmt:
gsql ((MogDB 2.1.1 build b5f25b20) compiled at 2022-03-21 14:42:30 commit 0 last mr )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.
miao=> \d
List of relations
Schema | Name | Type | Owner | Storage
--------+--------------------------------+-------+-------+------------------------------------------------------
dbmt | b1921101_bak | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | cc_chx | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | cd_material | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | customer_me_degree_wbinfo | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | customer_me_degree_wbinfo_test | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | erp_i_contract_class | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | scm_warning_reauidt_detail | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | sm_filemanagerconfig | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | t_o | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | t_o_t | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | t_p_t | table | dbmt | {orientation=row,compression=no,storage_type=USTORE}
dbmt | temp1 | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tiji_update | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_mis_code_zq | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_mul_stock_detail_20200614 | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_output_bill_20190821 | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_project_party | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_unit_vol | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | tmp_user_rep | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | ur_s_role | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | w_work_cooperate_gjback0721 | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
dbmt | yang_tempate3 | table | dbmt | {orientation=row,compression=no,storage_type=ustore}
public | t1 | table | omm | {orientation=row,compression=no,storage_type=USTORE}
(105 rows)
miao=> select * from t1;
id | tm | lsn
----+----------------------------+---------
1 | 2022-10-29 14:22:52.273009 | Started
(1 row)
miao=>
7、 第二阶段恢复(按照时间进行还原)
[omm@db1 db1]$ vi recovery.conf
restore_command = 'cp /mogdb/data/db1/archivelog/%f %p'
recovery_target_time = '2022-10-29 14:26:44'
recovery_target_inclusive = true
~
"recovery.conf" [New] 3L, 137C written
[omm@db1 db1]$ gs_om -t start
Starting cluster.
=========================================
[SUCCESS] db1
2022-10-29 15:22:47.061 635cd4c6.1 [unknown] 140295007823424 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: could not create any HA TCP/IP sockets
2022-10-29 15:22:47.064 635cd4c6.1 [unknown] 140295007823424 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: Failed to initialize the memory protect for g_instance.attr.attr_storage.cstore_buffers (1024 Mbytes) or shared memory (1967 Mbytes) is larger.
=========================================
Successfully started.
[omm@db1 db1]$ gsql -d miao -U dbmt -p 26999 -r
failed to connect /opt/mogdb/tools/omm_mppdb:26999.
[omm@db1 db1]$ gsql -d miao -U dbmt -p 26000 -r
Password for user dbmt:
gsql ((MogDB 2.1.1 build b5f25b20) compiled at 2022-03-21 14:42:30 commit 0 last mr )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.
miao=> select * from t1;
id | tm | lsn
----+----------------------------+--------------
1 | 2022-10-29 14:22:52.273009 | Started
2 | 2022-10-29 14:26:10.55628 | First Insert
(2 rows)
8、 第三阶段恢复(按照lsn进行恢复)
restore_command = 'cp /mogdb/data/db1/archivelog/%f %p'
recovery_target_lsn = '4/EE0230C0'
recovery_target_inclusive = true
[omm@db1 archivelog]$ gs_om -t stop
Stopping cluster.
[omm@db1 db1]$ gs_om -t start
Starting cluster.
=========================================
[SUCCESS] db1
2022-10-29 16:08:54.842 635cdf96.1 [unknown] 140627704972864 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: could not create any HA TCP/IP sockets
2022-10-29 16:08:54.887 635cdf96.1 [unknown] 140627704972864 [unknown] 0 dn_6001 01000 0 [BACKEND] WARNING: Failed to initialize the memory protect for g_instance.attr.attr_storage.cstore_buffers (1024 Mbytes) or shared memory (1967 Mbytes) is larger.
=========================================
Successfully started.
[omm@db1 db1]$ gsql -d miao -U dbmt -p 26000
Password for user dbmt:
gsql ((MogDB 2.1.1 build b5f25b20) compiled at 2022-03-21 14:42:30 commit 0 last mr )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.
miao=> select 8 from t1;
?column?
----------
8
8
8
(3 rows)
miao=> select * from t1;
id | tm | lsn
----+----------------------------+---------------
1 | 2022-10-29 14:22:52.273009 | Started
2 | 2022-10-29 14:26:10.55628 | First Insert
3 | 2022-10-29 14:31:13.220708 | Second Insert
(3 rows)
miao=>
9、 手动结束PITR状态
miao=> select pg_is_in_recovery();
pg_is_in_recovery
-------------------
t
(1 row)
miao=> select pg_xlog_replay_resume();
pg_xlog_replay_resume
-----------------------
(1 row)
[omm@db1 db1]$ gsql -d miao -U dbmt -p 26000 -r
Password for user dbmt:
gsql ((MogDB 2.1.1 build b5f25b20) compiled at 2022-03-21 14:42:30 commit 0 last mr )
Non-SSL connection (SSL connection is recommended when requiring high-security)
Type "help" for help.
miao=> select pg_is_in_recovery();
pg_is_in_recovery
-------------------
f
(1 row)
miao=>
「喜欢这篇文章，您的关注和赞赏是给作者最好的鼓励」
关注作者
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文章的来源（墨天轮），文章链接，文章作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
