from:http://miracle.blog.51cto.com/255044/69829
真的很长时间没有写些文字了,惭愧。不过真的很忙,管理工作不好做呀。不过“忙里偷闲”老本行不能丢。
前两天后台日志和线上程序都报了这样的ORACLE错误
ORA-00600: internal error code, arguments: [17059], [0xC0000000BD85D928], [], [], [], [], [], []
基本上ORACLE的600错误是非常让人讨厌的错误。查阅了一下:
ERROR:
ORA-600 [17059] [a]
VERSIONS:
versions 7.1 to 10.1
DESCRIPTION:
While building a table to hold the list of child cursor dependencies
relating to a given parent cursor, we exceed the maximum possible size
of the table.
ARGUMENTS:
Arg [a] Object containing the table
FUNCTIONALITY:
Kernel Generic Library cache manager
IMPACT:
PROCESS FAILURE
NON CORRUPTIVE - No underlying data corruption.
SUGGESTIONS:
One symptom of this error is that the session will appear to hang for a
period of time prior to this error being reported.
If the Known Issues section below does not help in terms of identifying
a solution, please submit the trace files and alert.log to Oracle
Support Services for further analysis.
Issuing this SQL as SYS (SYSDBA) may help show any problem
objects in the dictionary:
select do.obj#,
po.obj# ,
p_timestamp,
po.stime ,
decode(sign(po.stime-p_timestamp),0,'SAME','*DIFFER*') X
from sys.obj$ do, sys.dependency$ d, sys.obj$ po
where P_OBJ#=po.obj#(+)
and D_OBJ#=do.obj#
and do.status=1 /*dependent is valid*/
and po.status=1 /*parent is valid*/
and po.stime!=p_timestamp /*parent timestamp not match*/
order by 2,1
;
Normally the above select would return no rows. If any rows are
returned the listed dependent objects may need recompiling.
ORA-600 [17059] [a]
VERSIONS:
versions 7.1 to 10.1
DESCRIPTION:
While building a table to hold the list of child cursor dependencies
relating to a given parent cursor, we exceed the maximum possible size
of the table.
ARGUMENTS:
Arg [a] Object containing the table
FUNCTIONALITY:
Kernel Generic Library cache manager
IMPACT:
PROCESS FAILURE
NON CORRUPTIVE - No underlying data corruption.
SUGGESTIONS:
One symptom of this error is that the session will appear to hang for a
period of time prior to this error being reported.
If the Known Issues section below does not help in terms of identifying
a solution, please submit the trace files and alert.log to Oracle
Support Services for further analysis.
Issuing this SQL as SYS (SYSDBA) may help show any problem
objects in the dictionary:
select do.obj#,
po.obj# ,
p_timestamp,
po.stime ,
decode(sign(po.stime-p_timestamp),0,'SAME','*DIFFER*') X
from sys.obj$ do, sys.dependency$ d, sys.obj$ po
where P_OBJ#=po.obj#(+)
and D_OBJ#=do.obj#
and do.status=1 /*dependent is valid*/
and po.status=1 /*parent is valid*/
and po.stime!=p_timestamp /*parent timestamp not match*/
order by 2,1
;
Normally the above select would return no rows. If any rows are
returned the listed dependent objects may need recompiling.
非法的对象是没有了,应该是child cursor过多造成的。联想因为某种原因(不能说了),我修改了一个数据库参数:
cursor_sharing = TRUE
立刻捕捉这句SQL
SELECT sequence_id,sercode,content,area FROM XXX WHERE sendzt=:"SYS_B_0" AND Checkzt=:"SYS_B_1" AND starttime < :"SYS_B_2" AND starttime > :"SYS_B_3" AND rownum < :"SYS_B_4"
虽然看到的SQL貌似已经完成了绑定,但是一定还是有些问题的。当研发同事查询到这句SQL的时候告诉我:“没有问题是绑定的。”,但是这句SQL的rownum小于的是一个确切数字(10),没错就是它!让研发同事也按照其他变量的方式绑定这个确切的“变量”。将修改的程序重新上线,LOAD=1,VERSION_COUNT=1而执行次数超过12万(程序运行2个小时后),不再报600错!
我的这个解决办法只是根据数据库现有的模式因势利导给出的。这种cursor_sharing = TRUE方式其实一直存在很多争议。毕竟的“偷窥变量”是ORACLE一厢情愿的事情,时常不会太准确的,并且存在性能上的风险。-:)
本文出自 “Be the miracle!” 博客,请务必保留此出处