今天发现alertsid.log文件里面每隔几分钟都会出现ORA-00600的错误:

file /oracle/app/admin/orcl/udump/orcl1_ora_745474.trc: ORA-00600: internal error code, arguments: [qcplgnt : no null terminator], [], [], [], [], [], [], []

trace文件里面看到这个相关的sql:

 select ta.id,tb.name||:"SYS_B_0"||nvl(tc.title,:"SYS_B_1")||:"SYS_B_2" as subject_name from subject_issue ta  inner join subject tb on ta.subject_id=tb.id left outer join (select ra.subject_issue_id,rd.title from subject_page ra,page rb, (select concat(st.name,pg.section_num) as title,pg.id as pid from page pg,paper pp,section st where pp.section_id=st.id and pg.paper_id=pp.id) rd, (select ta.subject_issue_id, min(tb.num) as minnum from subject_page ta,page tb where ta.page_id=tb.id group by ta.subject_issue_id) rc where ra.page_id=rb.id and ra.subject_issue_id=rc.subject_issue_id and ra.page_id=rd.pid and rb.num=rc.minnum) tc on ta.id=tc.subject_issue_id

怀疑和我使用了cursor_sharing有关，将这个参数改回到exact，然后reboot instance,flush shared_pool，没有再观察到ora-00600这个错误....

觉得很郁闷，因为数据库没有邦定变量，才用这个参数改善性能。

这个问题，需要我继续跟踪啊.