1 row updated.
Elapsed: 00:00:00.01
I1S1> update t_deadlock set name = 'b1' where id = 2;
下面在实例2上的会话2，引入死锁：
I2S2> update t_deadlock set name = 'a2' where id = 1;
update t_deadlock set name = 'a2' where id = 1
*
ERROR at line 1:
ORA-00060: deadlock detected while waiting for resource
Elapsed: 00:00:06.07
I2S2> update t_deadlock set name = 'a2' where id = 1;
update t_deadlock set name = 'a2' where id = 1
*
ERROR at line 1:
ORA-00060: deadlock detected while waiting for resource
Elapsed: 00:00:05.95
I2S2> update t_deadlock set name = 'a2' where id = 1;
update t_deadlock set name = 'a2' where id = 1
*
ERROR at line 1:
ORA-00060: deadlock detected while waiting for resource
Elapsed: 00:00:06.63
I2S2> update t_deadlock set name = 'a2' where id = 1;
update t_deadlock set name = 'a2' where id = 1
*
ERROR at line 1:
ORA-00060: deadlock detected while waiting for resource
Elapsed: 00:00:05.89
显然由于不同实例的_lm_dd_interval参数的值设置不同，现在每次死锁都会在设置值更小的实例2上被检测，实例2上的会话每次都会被死锁牺牲掉。尝试设置不同的参数值在不同实例上设置死锁检测优先级获得成功。