openGauss/MogDB的mog_xlogdump工具类似于PostgreSQL的pg_xlogdump/pg_waldump,可以解析xlog日志,获取xlog里的相关记录。可以通过MogDB的官网下载对应的版本使用,
https://www.mogdb.io/downloads/mogdb
一、 创建表并增加主键(模拟数据)
omm1@ubuntu-linux-22-04-desktop:~$ gsql -r gsql ((openGauss 3.0.5 build b54d05de) compiled at 2023-09-23 09:23:38 commit 0 last mr ) Non-SSL connection (SSL connection is recommended when requiring high-security) Type "help" for help. openGauss=# create table test_tab1(id int, name varchar(20),insert_time timestamp not null default now()); CREATE TABLE openGauss=# d test_tab1 ? Table "public.test_tab1" Column | Type | Modifiers -------------+-----------------------------+------------------------ id | integer | name | character varying(20) | insert_time | timestamp without time zone | not null default now() openGauss=# show wal_level ; wal_level ------------- hot_standby (1 row) openGauss=# alter table test_tab1 add primary key(id); NOTICE: ALTER TABLE / ADD PRIMARY KEY will create implicit index "test_tab1_pkey" for table "test_tab1" ALTER TABLE openGauss=# d test_tab1 ? Table "public.test_tab1" Column | Type | Modifiers -------------+-----------------------------+------------------------ id | integer | not null name | character varying(20) | insert_time | timestamp without time zone | not null default now() Indexes: "test_tab1_pkey" PRIMARY KEY, btree (id) TABLESPACE pg_default
二、插入数据(模拟数据)
openGauss=# insert into test_tab1 values(1,'ysl'); INSERT 0 1 //查看数据的xmin,xmax,通过xmin可以看到这条数据属于事务13263 openGauss=# select xmin,xmax,* from test_tab1; xmin | xmax | id | name | insert_time -------+------+----+------+---------------------------- 13263 | 0 | 1 | ysl | 2023-12-15 14:49:35.975264 (1 row) //查看表的filenode openGauss=# select pg_relation_filenode('test_tab1'); pg_relation_filenode ---------------------- ? 16384 (1 row) //看到当前使用的wal文件对应的xlog_location openGauss=# select pg_current_xlog_location(); pg_current_xlog_location ------------------------- 0/22B8508 (1 row) //可以用如下直接转成使用的xlog文件 openGauss=# select pg_xlogfile_name(pg_current_xlog_location()); pg_xlogfile_name -------------------------- 000000010000000000000002 (1 row)
三、使用mog_xlogdump解析xlog文件
//指定事物id,表应的filenode
mog_xlogdump -x 13263 -o 16384 -R int,varchar,timestamp 000000010000000000000002 'insert','tuple':{'int':'1','varchar':'ysl','timestamp':'2023-12-15 14:49:35.975264'}mog_xlogdump: FATAL: error in WAL record at 0/22B8B28: invalid record length at 0/22B8BC8: wanted 32, got 0
//或者不指定事物id,只指定表对应的filenode
mog_xlogdump -o 16384 -R int,varchar,timestamp 000000010000000000000002 'insert','tuple':{'int':'1','varchar':'ysl','timestamp':'2023-12-15 14:49:35.975264'}mog_xlogdump: FATAL: error in WAL record at 0/22B8C48: invalid record length at 0/22B8CE8: wanted 32, got 0
四、关于REPLICA IDENTITY和解码数据对比
REPLICA IDENTITY {DEFAULT | USING INDEX index_name | FULL | NOTHING} 调整逻辑复制时写入WAL日志中的信息量,该选项仅在wal_level配置为logical时才有效。
当原数据表发生更新时,默认的逻辑复制流只包含主键的历史记录,如果需要输出所需字段更新或删除的历史记录,可修改本参数。
- DEFAULT(非系统表的默认值): 会记录主键字段的旧值。
- USING INDEX : 会记录名为index_name索引包含的字段的旧值,索引的所有列必须NOT NULL。
- FULL : 记录了所有列的旧值。
- NOTHING(系统表默认值): 不记录旧值的信息。
在REPLICA IDENTITY默认未修改的情况下,update的解码数据对比如下
数据更新:未设置wal_level=logical,未修改replica identity
‘update’,‘new_tuple’:{‘(null)’:‘2’,‘(null)’:‘tt’,‘(null)’:‘2022-04-25 20:05:28.159008’}
数据更新:设置wal_level=logical,未修改replica identity
‘update’,‘new_tuple’:{‘(null)’:‘2’,‘(null)’:‘ttt’,‘(null)’:‘2022-04-25 20:05:28.159008’}
更新非主键列
‘update’,‘new_tuple’:{‘(null)’:‘2’,‘(null)’:‘tttt’,‘(null)’:‘2022-04-25 20:05:28.159008’}
更新主键列
‘update’,‘old_tuple’:{‘(null)’:‘2’,},‘new_tuple’:{‘(null)’:‘3’,‘(null)’:‘ttttt’,‘(null)’:‘2022-04-25 20:05:28.159008’}