oracle row cache lock 之sequence

Posted 2021-03-02

今天遇到一个生产库产生大量row cache lock,以下是相应步骤：

1 查询当时P1的情况

select INSTANCE_NUMBER,p1,count(*) cnt from dba_hist_active_sess_history where event=‘row cache lock‘ and SAMPLE_TIME>=to_date(‘2018-08-31 10:00:00‘,‘yyyy-mm-dd hh24:mi:ss‘) and SAMPLE_TIME <=to_date(‘2018-08-31 10:40:00‘,‘yyyy-mm-dd hh24:mi:ss‘) group by INSTANCE_NUMBER,p1 order by cnt;

2 根据第一步查询的P1，代入到下面cache#,这里查询出来的结果是13，dc_sequences

select INST_ID, CACHE#,TYPE,GETS,PARAMETER from gv$rowcache where CACHE# in (?) order by gets;

3 查询当时用户会话量情况

select INSTANCE_NUMBER,USER_ID,count(*) cnt from dba_hist_active_sess_history 2 where SAMPLE_TIME>=to_date(‘2018-08-31 10:00:00‘,‘yyyy-mm-dd hh24:mi:ss‘) and SAMPLE_TIME <=to_date(‘2018-08-31 10:30:00‘,‘yyyy-mm-dd hh24:mi:ss‘) 3 group by INSTANCE_NUMBER,USER_ID order by 1,cnt;

4 查询对应SQL的情况

select INSTANCE_NUMBER,SAMPLE_TIME,event,sql_opname,sql_id,count(*) cnt from dba_hist_active_sess_history where SAMPLE_TIME>=to_date(‘2018-08-31 10:00:00‘,‘yyyy-mm-dd hh24:mi:ss‘) and SAMPLE_TIME <=to_date(‘2018-08-31 10:30:00‘,‘yyyy-mm-dd hh24:mi:ss‘) group by INSTANCE_NUMBER,SAMPLE_TIME,event,sql_opname,sql_id order by 2,1;

5 查询对应节点更详细的会话信息

select INSTANCE_NUMBER,SAMPLE_TIME,session_id,BLOCKING_SESSION,current_obj#,user_id,event,sql_id,P1 from dba_hist_active_sess_history where SAMPLE_TIME>=to_date(‘2018-08-31 10:17:00‘,‘yyyy-mm-dd hh24:mi:ss‘) and SAMPLE_TIME <=to_date(‘2018-08-31 10:20:00‘,‘yyyy-mm-dd hh24:mi:ss‘) and INSTANCE_NUMBER=2 and event=‘row cache lock‘;

最后定位是一个高频INSERT 语句引用SEQUENCE引起的，而这个序列是NOCAHE，改成cache恢复正常