Oracle与Postgresql在PLSQL内事务回滚的重大差异

mingjie

发布于 2022-09-19 15:39:12

66400

代码可运行

文章被收录于专栏：Postgresql源码分析Postgresql源码分析

运行总次数：0

代码可运行

相关《Oracle/Mysql迁移到Postgresql事务回滚行为差异及改造方法》《Oracle与Postgresql在PLSQL内事务回滚的重大差异》

这个差异点非常容易造成Oracle迁移到PG后业务逻辑出现重大差异。

1 总结

先放总结

Oracle：在PLSQL内如果语句执行失败，进入异常处理程序后，PL程序正常退出。那么在执行失败语句前面的SQL不会回滚，执行结果都正常提交了。
Postgresql：在PLPGSQL内如果语句执行失败，进入异常处理程序后，PL正常退出。那么整个PL内的所有SQL自动回滚，因为：
- PG不支持PL内写SAVEPOINT （Oracle在每个语句前有隐式的savepoint）
- PL整体包装在一个大事务内。

Oracle11g

2 测试用例Oracle11g

drop table employees;
create table employees(employee_id int, salary int, raise int, job_id varchar(16), last_name varchar(16), first_name varchar(16));
insert into employees values (1, 1000, 60, 'ST_CLERK', 'Geller', 'Rose');
insert into employees values (2, 2000, 100, 'ST_CLERK', 'Green', 'Rachel');
insert into employees values (3, 4000, 400, 'ST_AAAAA', 'Bing', 'Chanller');
insert into employees values (4, 6000, 630, 'AD_CLERK', 'Geller', 'Rose');
insert into employees values (5, 7000, 120, 'AD_DSFWR', 'Green', 'Rachel');
insert into employees values (6, 8000, 410, 'AD_AAAAA', 'Bing', 'Chanller');
DROP TABLE emp_name;
CREATE TABLE emp_name AS SELECT employee_id, last_name, salary FROM employees;
CREATE UNIQUE INDEX empname_ix ON emp_name (employee_id);
 
DECLARE
  emp_id        employees.employee_id%TYPE;
  emp_lastname  employees.last_name%TYPE;
  emp_salary    employees.salary%TYPE;
 
BEGIN
  SELECT employee_id, last_name, salary
  INTO emp_id, emp_lastname, emp_salary 
  FROM employees
  WHERE employee_id = 1;
  INSERT INTO emp_name (employee_id, last_name, salary) VALUES (100, emp_lastname, emp_salary);
  INSERT INTO emp_name (employee_id, last_name, salary) VALUES (emp_id, emp_lastname, emp_salary);
  
EXCEPTION
  WHEN DUP_VAL_ON_INDEX THEN
  DBMS_OUTPUT.PUT_LINE('Insert was rolled back');
END;
/


select * from emp_name;

结果

3 测试用例Postgresql14

drop table employees;
create table employees(employee_id int, salary int, raise int, job_id varchar(16), last_name varchar(16), first_name varchar(16));
insert into employees values (1, 1000, 60, 'ST_CLERK', 'Geller', 'Rose');
insert into employees values (2, 2000, 100, 'ST_CLERK', 'Green', 'Rachel');
insert into employees values (3, 4000, 400, 'ST_AAAAA', 'Bing', 'Chanller');
insert into employees values (4, 6000, 630, 'AD_CLERK', 'Geller', 'Rose');
insert into employees values (5, 7000, 120, 'AD_DSFWR', 'Green', 'Rachel');
insert into employees values (6, 8000, 410, 'AD_AAAAA', 'Bing', 'Chanller');
DROP TABLE emp_name;
CREATE TABLE emp_name AS SELECT employee_id, last_name, salary FROM employees;
CREATE UNIQUE INDEX empname_ix ON emp_name (employee_id);
select * from emp_name;

DO $$DECLARE
  emp_id        employees.employee_id%TYPE;
  emp_lastname  employees.last_name%TYPE;
  emp_salary    employees.salary%TYPE;
 
BEGIN
  SELECT employee_id, last_name, salary
  INTO emp_id, emp_lastname, emp_salary 
  FROM employees
  WHERE employee_id = 1;
  INSERT INTO emp_name (employee_id, last_name, salary) VALUES (100, emp_lastname, emp_salary);
  INSERT INTO emp_name (employee_id, last_name, salary) VALUES (emp_id, emp_lastname, emp_salary);
  
EXCEPTION
  WHEN others THEN
  raise notice 'Insert was rolled back';
END$$;
select * from emp_name;

执行结果