联系：手机/微信(+86 17813235971) QQ(107644445)
标题：ORACLE Instance XFF (pid = 18) – Error 600 encountered while recovering transaction
作者：惜分飞©版权所有[未经本人同意,不得以任何形式转载,否则有进一步追究法律责任的权利.]
分享一次由于一个表异常导致数据库报类似：ORACLE Instance XFF (pid = 18) – Error 600 encountered while recovering transaction故障的案例
一个10.2.0.4的数据库,正常运行的库突然出现如下错误
Sun Apr 07 11:07:12 2019 Thread 1 advanced to log sequence 602883 (LGWR switch) Current log# 3 seq# 602883 mem# 0: L:\ORADATA\XFF\REDO03.LOG Sun Apr 07 11:10:38 2019 Thread 1 advanced to log sequence 602884 (LGWR switch) Current log# 1 seq# 602884 mem# 0: L:\ORADATA\XFF\REDO01.LOG Sun Apr 07 11:11:56 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\udump\XFF_ora_22956.trc: ORA-00600: 内部错误代码, 参数: [ktspgfb-1], [], [], [], [], [], [], [] Sun Apr 07 11:12:46 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\udump\XFF_ora_27408.trc: ORA-00600: 内部错误代码, 参数: [kcbnew_3], [0], [1], [168354056], [], [], [], [] Sun Apr 07 11:13:57 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\udump\XFF_ora_6632.trc: ORA-00600: 内部错误代码, 参数: [ktspgfb-1], [], [], [], [], [], [], []
过一段时间报，然后实例直接crash
Tue Apr 09 07:47:35 2019 ORACLE Instance XFF (pid = 18) - Error 600 encountered while recovering transaction (1, 1) on object 113718002. Tue Apr 09 07:47:35 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_smon_12948.trc: ORA-00600: internal error code, arguments: [kcbgcur_3], [168454497], [8], [4], [0], [], [], [] Tue Apr 09 07:55:23 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_pmon_22652.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:24 2019 PMON: terminating instance due to error 474 Tue Apr 09 07:55:24 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_lgwr_28608.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:34 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_psp0_12544.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:34 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_j000_5216.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:35 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_ckpt_28204.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:36 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_mman_9320.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:44 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_q002_24384.trc: ORA-00474: SMON process terminated with error Tue Apr 09 07:55:53 2019 Errors in file c:\oracle\product\10.2.0\admin\XFF\bdump\XFF_reco_24124.trc: ORA-00474: SMON process terminated with error
根据以上报错，数据库crash的原因是由于undo异常导致，通过对undo进行重建，解决掉异常undo，但是业务运行之后，一样的问题又重现，最后通过分析确认是对象异常导致
SQL> create table XFF.T_XIFENFEI_xff as select * from XFF.T_XIFENFEI;
create table XFF.T_XIFENFEI_xff as select * from XFF.T_XIFENFEI
*
ERROR at line 1:
ORA-00600: internal error code, arguments: [kcbz_check_objd_typ], [0], [0], [1], [], [], [], []
屏蔽相关block obj的check之后
SQL> create table XFF.T_XIFENFEI_xff as select * from XFF.T_XIFENFEI;
create table XFF.T_XIFENFEI_xff as select * from XFF.T_XIFENFEI
*
ERROR at line 1:
ORA-00600: internal error code, arguments: [ktspScanInit-l1], [], [], [], [],[], [], []
比较明显该表对象出现逻辑异常，通过基于rowid的方式对该表数据进行抽取
SQL> create table XFF.T_XIFENFEI_new
2 as
3 select * from XFF.T_XIFENFEI where 1=0;
Table created.
SQL> set serveroutput on
SQL> set concat off
SQL> DECLARE
2 nrows number;
3 rid rowid;
4 dobj number;
5 ROWSPERBLOCK number;
6 BEGIN
7 ROWSPERBLOCK:=1000;
8 nrows:=0;
9 select data_object_id into dobj
10 from dba_objects
11 where owner = 'XFF'
12 and object_name = 'T_XIFENFEI'
13 ;
14 for i in (select relative_fno, block_id, block_id+blocks-1 totblocks
15 from dba_extents
16 where owner = 'XFF'
17 and segment_name = 'T_XIFENFEI'
18 order by extent_id)
19 loop
20 for br in i.block_id..i.totblocks loop
21 for j in 1..ROWSPERBLOCK loop
22 begin
23 rid := dbms_rowid.ROWID_CREATE(1,dobj,i.relative_fno, br , j-1);
24 insert into XFF.T_XIFENFEI_NEW
25 select /*+ ROWID(A) */ *
26 from XFF.T_XIFENFEI A
27 where rowid = rid;
28 if sql%rowcount = 1 then nrows:=nrows+1; end if;
29 if (mod(nrows,10000)=0) then commit; end if;
30 exception when others then null;
31 end;
32 end loop;
33 end loop;
34 end loop;
35 COMMIT;
36 dbms_output.put_line('Total rows: '||to_char(nrows));
37 END;
38 /
Total rows: 227000
PL/SQL procedure successfully completed.
再次观察数据库恢复正常,也不再crash和报错,恢复完成
