Job And Schedule (V8R6C3)

KingbaseES 数据库提供了 kdb_schedule 扩展，使得用户能通过类似oracle job 的方式进行job调用。kdb_schedule 提供了三个Schema ：dbms_job and dbms_scheduler 分别类似于 oracle 的 dbms_job 和 dbms_schedule 包，数据字典信息放于kdb_schedule 模式下。

一、通过dbms_job管理Job

1、创建Job

create table d_test(tid varchar2(64), insdate date);

create or replace procedure p_test() as
begin
  insert into d_test values(to_char(sysdate, 'yyyymmddhh24miss'), sysdate);
  commit;
end;
/

DECLARE
  v_jobid NUMBER;
BEGIN
  dbms_job.submit(v_jobid, 'call p_test()', now(), 'Freq=Minutely;Interval=1');
  COMMIT;
END;
/

2、查询Job

test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
 jobid |                   jobname                   | jobenabled | joblastrun |          jobnextrun           | jobrepeattimes
-------+---------------------------------------------+------------+------------+-------------------------------+----------------
     1 | internal_job1:2021-06-22 08:11:39.797681+08 | f          |            | 2021-06-22 08:11:39.797681+08 |              0

可以看到，Job 创建后，默认的状态是 jobenabled=false，也就是实际没有启用。

Note：早期版本 kdb_job 表是schema kdb_schedule下，后面移到pg_catalog schema下。

3、启用Job

job 创建后，默认是 broken(disabled) ，需要enable

test=# call dbms_job.broken(1,false);
CALL
test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
 jobid |                   jobname                   | jobenabled | joblastrun |          jobnextrun           | jobrepeattimes
-------+---------------------------------------------+------------+------------+-------------------------------+----------------
     1 | internal_job1:2021-06-22 08:11:39.797681+08 | t          |            | 2021-06-22 08:11:39.797681+08 |              0

enable job 后，等待一会再查，发现实际job 还是没有运行。这是为什么？这是因为kdb_schedule 后台进程没有启动。启动后台进程后，再查，可以看到job已经运行了。

test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
 jobid |                   jobname                   | jobenabled |          joblastrun           |       jobnextrun       | jobrepeattimes
-------+---------------------------------------------+------------+-------------------------------+------------------------+----------------
     1 | internal_job1:2021-06-22 08:11:39.797681+08 | t          | 2021-06-22 08:22:39.062009+08 | 2021-06-22 08:23:39+08 |              0
(1 row)

4、Job实际上也是Schedule

test=#  select * from kdb_schedule.kdb_schedule;
 scid | scname | scdesc | scenabled |          scstart           | scend |    screpeat_interval
------+--------+--------+-----------+----------------------------+-------+--------------------------
    2 |        |        | t         | 2021-06-22 10:11:25.586148 |       | Freq=Minutely;Interval=1
(1 row)

test=# select * from kdb_schedule.kdb_schedule_job;
 sjid | sjscid | sjjobid | sjstatus | sjlasttime |          sjnexttime
------+--------+---------+----------+------------+-------------------------------
    2 |      2 |       1 | s        |            | 2021-06-22 10:11:25.586148+08

可以看到，通过dbms_job创建的job信息，同样会在kdb_schedule 和kdb_schedule_job 显示。

5、其他Job维护操作

--Job删除
test=# call dbms_job.remove(1);
CALL

二、启动kdb_schedule 进程

使用过Oracle的人都知道，Oracle数据库启动时，会运行多个ora_j000_xxx 进程，这些进程实际上就是用于运行Job 的。同样KINGBASE 也需要后台进程来运行job

[kingbase@dbhost03 ~]$ kdb_schedule "dbname=template1 user=system"

Note：kdb_schedule只需要运行一个，后面跟的数据库连接信息可以是任何一个数据库，不影响其他库job的使用。

三、通过dbms_scheduler管理Job

1、创建program

begin
  dbms_scheduler.create_program(program_name        => 'prog_01',
                                program_type        => 'PLSQL_BLOCK',
                                program_action      => 'call public.p_test()',
                                acdbname            => 'test',
                                number_of_arguments => 0,
                                enabled             => true,
                                comments            => 'test program');
end;

这里有两个地方必须注意。'PLSQL_BLOCK' 必须大写，acdbname 必须指定。

acdbname 指明了action 所在的数据库，如果没有指定，默认指 kdb_schedule 运行时指定的数据库。成功创建program后，会有如下一行信息：

test=# select * from kdb_schedule.kdb_action;
 acid | acname  |    acdesc    | acenabled | ackind |        accode        | acconnstr | acdbname | acnextrun
------+---------+--------------+-----------+--------+----------------------+-----------+----------+-----------
    7 | prog_01 | test program | t         | s      | call public.p_test() |           | test     |
(1 row)

2、创建Schedule

begin
  dbms_scheduler.create_schedule(schedule_name   => 'schedule_01',
                                 start_date      => now(),
                                 repeat_interval => 'freq=minutely;interval=1',
                                 end_date        => null,
                                 comments        => 'test schedule');
end;

创建schedule 后，会有如下一行信息：

test=# select * from kdb_schedule.kdb_schedule;
 scid |   scname    |    scdesc     | scenabled |          scstart           | scend |    screpeat_interval
------+-------------+---------------+-----------+----------------------------+-------+--------------------------
    7 | schedule_01 | test schedule | t         | 2021-06-22 14:28:38.536823 |       | freq=minutely;interval=1
(1 row)

3、创建Job

begin
  dbms_scheduler.create_job(job_name         => 'job_01',
                            program_name     => 'prog_01',
                            schedule_name    => 'schedule_01',
                            job_class        => 'routine maintenance',
                            enabled          => true,
                            auto_drop        => true,
                            comments         => 'test job',
                            credentail_name  => null,
                            destination_name => null);
end;

创建后，下表会有相关记录信息，分别记录 job 与 schedule ， 以及 job 与 action 之间的关系。

test=# select * from kdb_schedule.kdb_schedule_job;
 sjid | sjscid | sjjobid | sjstatus | sjlasttime | sjnexttime
------+--------+---------+----------+------------+------------
    7 |      7 |       1 | s        |            | 

test=# select * from kdb_schedule.kdb_job_action;
 jaid | jajobid | jaacid | jastatus | jalasttime
------+---------+--------+----------+------------
    7 |       1 |      7 | s        | 

具体的job信息，可以看 kdb_job:

test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
jobid | jobname | jobenabled |          joblastrun           |       jobnextrun       | jobrepeattimes
-------+---------+------------+-------------------------------+------------------------+----------------
     1 | job_01  | t          | 2021-06-22 14:59:41.997986+08 | 2021-06-22 15:00:38+08 |              0
(1 row)

4、Enable and Disable Job

Job 创建时，可以指定enable or disable ，后续也可以手动修改。

begin
  dbms_scheduler.disable(name => 'job_01');
end;
/

test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
 jobid | jobname | jobenabled |          joblastrun           |       jobnextrun       | jobrepeattimes
-------+---------+------------+-------------------------------+------------------------+----------------
     1 | job_01  | f          | 2021-06-22 15:06:42.303189+08 | 2021-06-22 15:07:38+08 |              0
(1 row)

begin
  dbms_scheduler.enable(name => 'job_01');
end;
/

test=# select jobid,jobname,jobenabled,joblastrun,jobnextrun,jobrepeattimes from kdb_job;
 jobid | jobname | jobenabled |          joblastrun           |       jobnextrun       | jobrepeattimes
-------+---------+------------+-------------------------------+------------------------+----------------
     1 | job_01  | t          | 2021-06-22 15:08:07.365179+08 | 2021-06-22 15:08:38+08 |              0
(1 row)

5、其他操作

begin
  dbms_scheduler.drop_job(job_name         => 'job_01');
end;
/

begin
  dbms_scheduler.drop_schedule(schedule_name   => 'schedule_01');
end;
/

begin
  dbms_scheduler.drop_program(program_name        => 'prog_01');
end;
/

begin
  dbms_scheduler.run_job(job_name => 'job_01');
end;
/

四、故障排查

kdb_schedule.kdb_jobsteplog 记录了job 调用的具体信息，如果有错误的，jsloutput 会显示具体的错误信息。