作者:刘安
爱可生测试团队成员,主要负责 DTLE 开源项目相关测试任务,擅长 Python 自动化测试开发。
本文来源:原创投稿
*爱可生开源社区出品,原创内容未经授权不得随意使用,转载请联系小编并注明来源。
前言:过年前 DTLE 发布了 4.22.01.0 版本,该版本最重要的特性是支持 Oracle-MySQL 增量数据同步。今天我就来给大家介绍一下这个功能。
一、现状
1. 目前只支持增量同步
a. 可以根据 SCN 节点开启增量复制
b. 从任务启动时间开启增量复制
2. 类型映射
a. 已支持类型
Oracle | MySQL | 限制 |
---|---|---|
BINARY_DOUBLE | float | mysql 不支持Inf/-Inf/Nan数据,用NULL来存储 |
CHAR(n), CHARACTER(n) | CHAR(n), CHARACTER(n) | |
DATE | datetime | |
DECIMAL(p,s), DEC(p,s) | DECIMAL(p,s), DEC(p,s) | |
DOUBLE PRECISION | DOUBLE PRECISION | |
FLOAT(p) | DOUBLE | |
INTEGER, INT | INT | 极值问题 (https://github.com/actiontech/dtle/issues/825) |
INTERVAL DAY(p) TO SECOND(s) | VARCHAR(30) | |
INTERVAL YEAR(p) TO MONTH | VARCHAR(30) | |
NCHAR VARYING(n) | NCHAR VARYING(n) | |
NCHAR(n) | NCHAR(n)/NVARCHAR(n) | |
NUMBER(p,0), NUMBER(p) | TINYINT/SMALLINT/INT/BIGINT/DECIMAL(p) | |
NUMBER(p,s) | DECIMAL(p,s) | |
NUMBER, NUMBER(*) | DOUBLE | |
NUMERIC(p,s) | NUMERIC(p,s) | |
NVARCHAR2(n) | NVARCHAR(n) | |
RAW(n) | VARBINARY(n) | |
REAL | DOUBLE | |
ROWID | CHAR(100) | |
SMALLINT | DECIMAL(38) | |
TIMESTAMP(p) | datetime | |
VARCHAR2(n) | VARCHAR(n) | |
VARCHAR(n) | VARCHAR(n) |
b. 待支持类型
Oracle | MySQL | 当前不支持原因 |
---|---|---|
BINARY_FLOAT | float | MySQL不支持Inf/-Inf/Nan数据, MySQL float类型无法精确匹配,导致更新失败 |
BLOB | BLOB | 当前实现逻辑,无法从redoSQL获取足够的值 |
CLOB | CLOB | 当前实现逻辑,无法从redoSQL获取足够的值 |
LONG | LONGTEXT | 只支持insert |
LONG RAW | LONGBLOB | 只支持insert |
NCLOB | TEXT | 无法从redoSQL获取足够的值 |
TIMESTAMP(p) WITH TIME ZONE | datetime | 时区问题未处理 |
c. 不支持类型
Oracle | MySQL | 不支持原因 |
---|---|---|
BFILE | VARCHAR(255) | logminer不支持 |
UROWID(n) | VARCHAR(n) | logminer读取的数据不足以构造新SQL |
XMLTYPE | VARCHAR(30) | logminer不支持 |
3. DML支持度
a. DML类型
DML类型 | Oracle SQL | MySQL SQL |
---|---|---|
INSERT | INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (0, NULL) | replace into ACTION_DB .CHAR_255_COLUMNS ( COL1, COL2`) values ('0', NULL) |
UPDATE | UPDATE ACTION_DB.CHAR_255_COLUMNS SET COL2='abcdefghijklmnopqrstuvwxyz' WHERE COL1=0 | update ACTION_DB .CHAR_255_COLUMNS set COL1 ='0', COL2 ='abcdefghijklmnopqrstuvwxyz' where ((COL1 = '0') and (COL2 is NULL)) limit 1 |
DELETE | DELETE FROM ACTION_DB.CHARACTER_255_COLUMNS WHERE COL1=0 | delete from ACTION_DB .CHAR_255_COLUMNS where ((COL1 = '0') and (COL2 = 'abcdefghijklmnopqrstuvwxyz')) limit 1 |
b. DML函数支持度
函数名 | 是否支持 | 其他 |
---|---|---|
CURRENT_TIMESTAMP | 是 | |
DATE | 是 | |
EMPTY_BLOB | 是 | 函数支持解析为NULL |
EMPTY_CLOB | 是 | 函数支持解析为NULL |
HEXTORAW | 是 | |
LOCALTIMESTAMP | 是 | |
RAWTOHEX | 是 | |
RAWTOHEX(CHR()) | 是 | |
SYSTIMESTAMP | 是 | |
TO_DATE | 是 | |
TO_DSINTERVAL | 是 | |
TO_TIMESTAMP | 是 | |
TO_YMINTERVAL | 是 | |
UNISTR | 是 |
4. DDL支持度
DDL | Target | Option |
---|---|---|
CREATE | 表 TABLE | DEFAULT CREATE |
ALTER | 表 TABLE | 增加字段 |
删除字段 | ||
重命名字段 (当前仅支持MySQL 8.0语法) | ||
变更字段类型 | ||
DROP | 表 TABLE | DEFAULT DROP |
二、环境准备
1. Oracle数据库开启归档日志
shell> su oracle
shell> mkdir /u01/app/oracle/oradata/archive_log
shell> sqlplus sys/oracle as sysdba
SQL> alter system set log_archive_dest_1='location=/u01/app/oracle/oradata/archive_log' scope=spfile;
SQL> alter system set db_recovery_file_dest_size = 10G;
SQL> shutdown immediate;
SQL> startup mount;
SQL> alter database add logfile group 3 '/u01/app/oracle/fast_recovery_area/XE/onlinelog/redo01.log' size 500m;
SQL> alter database add logfile group 4 '/u01/app/oracle/fast_recovery_area/XE/onlinelog/redo02.log' size 500m;
SQL> alter database add logfile group 5 '/u01/app/oracle/fast_recovery_area/XE/onlinelog/redo03.log' size 500m;
SQL> alter database archivelog;
SQL> alter database add supplemental log data (all) columns;
SQL> alter database open;
SQL> archive log list;
# Archive Mode表示已开启归档模式,Archive destination表示归档日志储存路径
2. 安装 LogMiner 工具(Oracle 安装时默认安装)
可查看系统中是否存在运行 LogMiner 所需要的 dbms_logmnr 、dbms_logmnr_d 包,如果没有安装 LogMiner 工具需要的包,需要运行下面两个命令:
shell> cat $ORACLE_HOME/rdbms/admin/dbmslm.sql | sqlplus sys/oracle as sysdba
shell> cat $ORACLE_HOME/rdbms/admin/dbmslmd.sql | sqlplus sys/oracle as sysdba
3. 创建 logminer 需要角色权限
shell> su oracle
shell> sqlplus sys/oracle as sysdba
SQL> create user roma_logminer identified by oracle default tablespace users;
SQL> GRANT CREATE SESSION TO roma_logminer;
SQL> GRANT SET CONTAINER TO roma_logminer;
SQL> GRANT SELECT ON V_$DATABASE TO roma_logminer;
SQL> GRANT FLASHBACK ANY TABLE TO roma_logminer;
SQL> GRANT SELECT ANY TABLE TO roma_logminer;
SQL> GRANT SELECT_CATALOG_ROLE TO roma_logminer;
SQL> GRANT EXECUTE_CATALOG_ROLE TO roma_logminer;
SQL> GRANT SELECT ANY TRANSACTION TO roma_logminer;
SQL> GRANT CREATE TABLE TO roma_logminer;
SQL> GRANT LOCK ANY TABLE TO roma_logminer;
SQL> GRANT CREATE SEQUENCE TO roma_logminer;
SQL> GRANT EXECUTE ON DBMS_LOGMNR TO roma_logminer;
SQL> GRANT EXECUTE ON DBMS_LOGMNR_D TO roma_logminer;
SQL> GRANT SELECT ON V_$LOG TO roma_logminer;
SQL> GRANT SELECT ON V_$LOG_HISTORY TO roma_logminer;
SQL> GRANT SELECT ON V_$LOGMNR_LOGS TO roma_logminer;
SQL> GRANT SELECT ON V_$LOGMNR_CONTENTS TO roma_logminer;
SQL> GRANT SELECT ON V_$LOGMNR_PARAMETERS TO roma_logminer;
SQL> GRANT SELECT ON V_$LOGFILE TO roma_logminer;
SQL> GRANT SELECT ON V_$ARCHIVED_LOG TO roma_logminer;
SQL> GRANT SELECT ON V_$ARCHIVE_DEST_STATUS TO roma_logminer;
SQL> alter user roma_logminer quota unlimited ON users;
4. 部署一个单节点的 4.22.01.0版本 DTLE
https://github.com/actiontech/dtle/releases/download/v4.22.01.0/dtle-ce-4.22.01.0.x86_64.rpm
三、创建 Oracle-MySQL 任务
1. 获取 Token
shell> curl -s -X POST "http://172.100.9.11:8190/v2/loginWithoutVerifyCode" -H "accept: application/json" -H "Content-Type: application/json" -d "{ \"password\": \"admin\", \"tenant\": \"platform\", \"username\": \"admin\"}" | jq
{
"message": "ok",
"data": {
"token": "eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJleHAiOjE2NDYxOTE2MzksImdyb3VwIjoicGxhdGZvcm0iLCJuYW1lIjoiYWRtaW4ifQ.-c_jVkxd_vP5Ka3gocdfGZIudWPujutdHpQYx8srX00"
}
}
2. 准备 job 文件
shell> cat job.json
{
"job_id": "test_oracle",
"src_task": {
"connection_config": {
"database_type": "Oracle",
"host": "172.100.9.31",
"port": 1521,
"user": "roma_logminer",
"password": "oracle",
"service_name": "XE"
},
"node_id": "96d28881-f91b-19f4-1614-4d8a0e718e2f",
"binlog_relay": false,
"repl_chan_buffer_size": 120,
"group_max_size": 1,
"group_timeout": 100,
"oracle_src_task_config": {
"scn": 0
},
"task_name": "src",
"replicate_do_db": [
{
"table_schema": "ACTION_DB"
}
]
},
"is_password_encrypted": false,
"dest_task": {
"connection_config": {
"database_type": "MySQL",
"host": "172.100.9.1",
"port": 3306,
"user": "test_dest",
"password": "test_dest"
},
"node_id": "96d28881-f91b-19f4-1614-4d8a0e718e2f",
"task_name": "dest",
"mysql_dest_task_config": {}
},
"task_step_name": "job_stage_full",
"failover": true,
"retry": 2
}
3. 创建 Oracle-MySQL job
shell> curl -s -X POST "http://172.100.9.11:8190/v2/job/migration/create" -H "accept: application/json" -H "Authorization: eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJleHAiOjE2NDYxOTE2MzksImdyb3VwIjoicGxhdGZvcm0iLCJuYW1lIjoiYWRtaW4ifQ.-c_jVkxd_vP5Ka3gocdfGZIudWPujutdHpQYx8srX00" -H "Content-Type: application/json" -d @job.json | jq
{
"job": {
"job_id": "test_oracle-migration",
"task_step_name": "job_stage_full",
"reverse": false,
"failover": true,
"is_password_encrypted": false,
"src_task": {
"task_name": "src",
"node_id": "96d28881-f91b-19f4-1614-4d8a0e718e2f",
"replicate_do_db": [
{
"table_schema": "ACTION_DB",
"table_schema_regex": "",
"table_schema_rename": "",
"tables": null
}
],
"replicate_ignore_db": null,
"skip_create_db_table": false,
"drop_table_if_exists": false,
"mysql_src_task_config": null,
"oracle_src_task_config": {
"scn": 0
},
"group_max_size": 1,
"group_timeout": 100,
"repl_chan_buffer_size": 120,
"chunk_size": 2000,
"connection_config": {
"host": "172.100.9.31",
"port": 1521,
"user": "roma_logminer",
"password": "*",
"service_name": "XE",
"database_type": "Oracle"
}
},
"dest_task": {
"task_name": "dest",
"node_id": "96d28881-f91b-19f4-1614-4d8a0e718e2f",
"database_type": "",
"mysql_dest_task_config": {
"parallel_workers": 1,
"use_my_sql_dependency": false,
"dependency_history_size": 2500
},
"connection_config": {
"host": "172.100.9.1",
"port": 3306,
"user": "test_dest",
"password": "*",
"service_name": "",
"database_type": "MySQL"
}
},
"retry": 2
},
"eval_create_index": 12,
"job_modify_index": 12,
"message": "ok"
}
4. 源端 Oracle 写入数据
SQL> create tablespace ACTION_DB datafile 'ACTION_DB.dbf' size 100M;
SQL> create user ACTION_DB identified by ACTION_DB default tablespace ACTION_DB;
SQL> grant unlimited tablespace to ACTION_DB;
SQL> CREATE TABLE ACTION_DB.CHAR_255_COLUMNS(col1 INT, col2 CHAR(255));
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (0, NULL);
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (1, 'abcdefghijklmnopqrstuvwxyz');
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (2, 'ABCDEFGHIJKLMNOPQRSTUVWXYZ');
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (3, '1234567890');
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (4, 1234567890);
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (5, '~`!@#$%^&*()-=_+{}[]|\:;<>,.?/');
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (6, '中文测试abc');
SQL> INSERT INTO ACTION_DB.CHAR_255_COLUMNS VALUES (7, '·~!@#¥%……&*()-=——+{}【】、|;‘:“,。/《》?');
5. 目标端 MySQL 检查同步情况
mysql> SHOW CREATE TABLE ACTION_DB.CHAR_255_COLUMNS\G
*************************** 1. row ***************************
Table: CHAR_255_COLUMNS
Create Table: CREATE TABLE `CHAR_255_COLUMNS` (
`COL1` int(11) DEFAULT NULL,
`COL2` char(255) DEFAULT NULL
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4
1 row in set (0.00 sec)
mysql> SELECT * FROM ACTION_DB.CHAR_255_COLUMNS;
+------+-----------------------------------------------------------------------------+
| COL1 | COL2 |
+------+-----------------------------------------------------------------------------+
| 0 | NULL |
| 1 | abcdefghijklmnopqrstuvwxyz |
| 2 | ABCDEFGHIJKLMNOPQRSTUVWXYZ |
| 3 | 1234567890 |
| 4 | 1234567890 |
| 5 | ~`!@#$%^&*()-=_+{}[]|\:;<>,.?/ |
| 6 | 中文测试abc |
| 7 | ·~!@#¥%……&*()-=——+{}【】、|;‘:“,。/《》? |
+------+-----------------------------------------------------------------------------+
8 rows in set (0.01 sec)
四、使用限制
因为 Oracle 和 MySQL 是异构数据库,所以在源端 Oracle 能执行的 Oracle SQL 语句通过 DTLE 转换到目标端的 MySQL SQL 语句后有可能无法正确执行。比如 Oracle 支持数值范围 MySQL 不支持,Oracle 的 DATE 类型支持公元前的年份而 MySQL 的 datetime 只能为公元后的年份等等。遇到这些情况,DTLE 的默认行为是报错并且停止同步。
如果需要跳过这种阻塞情况,可以修改 DTLE 的环境变量然后重新启动 DTLE 服务。
shell> vi /etc/systemd/system/multi-user.target.wants/dtle-nomad.service
# 添加
[Service]
Environment="SkipErr=true"
shell> systemctl daemon-reload
shell> systemctl restart dtle-nomad
如果在使用 DTLE 时发现了任何问题,请及时联系我们。
网友评论