一个案例,查看跟踪文件发现如下错误信息
d:Oracleproduct10.2.0admindbserverudumporcl_ora_5888.trc
Bad header found during buffer read
Data in bad block:
type: 1 format: 5 rdba: 0x30322d2d
last change scn: 0xfffe.fffefffe seq: 0xfe flg: 0xff
spare1: 0x0 spare2: 0x30 spare3: 0x0
consistency value in tail: 0x2d2d3533
check value in block header: 0xfffe
computed block checksum: 0x91f9
Reread of rdba: 0x09848269 (file 38, block 295529) found same corrupted data
Fri Dec 27 08:28:08 2013
Corrupt Block Found
TSN = 36, TSNAME = HGHIS
RFN = 38, BLK = 295529, RDBA = 159679081
Fri Dec 27 08:54:18 2013
Hex dump of (file 38, block 295561) in trace file d:oracleproduct10.2.0admindbserverudumporcl_ora_6796.trc
Bad header found during buffer read
Data in bad block:
type: 48 format: 0 rdba: 0x020cc100
last change scn: 0xb500.080cc100 seq: 0xa5 flg: 0xba
spare1: 0x34 spare2: 0x2 spare3: 0xb0b8
consistency value in tail: 0x395e3031
check value in block header: 0xcfcb
block checksum disabled
Reread of rdba: 0x09848289 (file 38, block 295561) found same corrupted data
Fri Dec 27 08:54:19 2013
Corrupt Block Found
TSN = 36, TSNAME = HGHIS
RFN = 38, BLK = 295561, RDBA = 159679113
Fri Dec 27 09:11:20 2013
Hex dump of (file 38, block 295563) in trace file d:oracleproduct10.2.0admindbserverudumporcl_ora_5584.trc
Bad header found during buffer read
Data in bad block:
type: 71 format: 7 rdba: 0x064d0001
last change scn: 0x3038.43584400 seq: 0x31 flg: 0x07
spare1: 0x44 spare2: 0x4c spare3: 0x771
consistency value in tail: 0x000a0000
check value in block header: 0x7800
computed block checksum: 0xcbf8
Reread of rdba: 0x0984828b (file 38, block 295563) found same corrupted data
Fri Dec 27 09:11:21 2013
Corrupt Block Found
TSN = 36, TSNAME = HGHIS
RFN = 38, BLK = 295563, RDBA = 159679115
Fri Dec 27 09:27:59 2013
根据上述信息得知38号数据文件的295529、295561、295563为坏块，可以使用DBV工具或者RMAN来检查坏块信息
DBV FILE="d:oradataDATA.DBF" blocksize=8192
or
rman target /
backup validate check logical database;
select * from V$DATABASE_BLOCK_CORRUPTION ;
可以根据文件号和块号查出损坏的是对象,表还是LOB segment
select tablespace_name,segment_type,owner,segment_name from dba_extents where file_id=38 and 295529 between block_id AND block_id + blocks - 1;
38是文件号,295529是block号
如果是对象，可以重建
alter index indexname rebuild
如果是表，可以使用10231事件忽略坏块，然后使用CTAS方式重建表最后rename table，别忘记rebuild index
alter session SET EVENTS '10231 trace name context forever,level 10';
create table tab_new as select * from tab;
rename tab to tab_bak;
rename tab_new to new;
alter index indexname rebuild;
alter session SET EVENTS '10231 trace name context off';
如果损坏的是LOB segment,先找出segment信息
select owner, segment_name, segment_type from dba_extents where file_id = 38 and 295563 between block_id and block_id + blocks - 1;
输出如下
owner=HGHIS
segment_name=SYS_LOB0000119493C00006$$
segment_type=LOBSEGMENT
找到表明和LOB字段
select table_name, column_name from dba_lobs where segment_name = 'SYS_LOB0000119493C00006$$' and owner = 'HGHIS';
输出如下
table_name = EMR_CASE
column_name = WORD
找到坏块的bad rowid,使用以下plsql脚本
create table bad_rows (row_id ROWID,oracle_error_code number);
set concat off
set serveroutput on
declare
n number;
error_code number;
bad_rows number := 0;
ora1578 EXCEPTION;
PRAGMA EXCEPTION_INIT(ora1578, -1578);
begin
for cursor_lob in (select rowid rid, &&lob_column from &&table_owner.&table_with_lob) loop
begin
n:=dbms_lob.instr(cursor_lob.&&lob_column,hextoraw('889911')) ;
exception
when ora1578 then
bad_rows := bad_rows + 1;
insert into bad_rows values(cursor_lob.rid,1578);
commit;
when others then
error_code:=SQLCODE;
bad_rows := bad_rows + 1;
insert into bad_rows values(cursor_lob.rid,error_code);
commit;
end;
end loop;
dbms_output.put_line('Total Rows identified with errors in LOB column: '||bad_rows);
end;
/
Enter value for lob_column: WORD
Enter value for table_owner: HGHIS
Enter value for table_with_lob: EMR_CASE
可以查询bad rowid
select * from bad_rows;
更新空LOB字段来避免ORA-1578,ORA-26040,如果是CLOB类型，将empty_blob()改为empty_clob()
set concat off
update &table_owner.&table_with_lob set &lob_column = empty_blob() where rowid in (select row_id from bad_rows);
将bad rowid lob块移到其他表空间
alter table &table_owner.&table_with_lob move LOB (&&lob_column) store as (tablespace &tablespace_name);
最后别忘记rebuild index
相关阅读：
Oracle ORA-01555 快照过旧 说明
ORA-01078 和 LRM-00109 报错解决方法
ORA-01555超长的Query Duration时间
ORA-00471 处理方法笔记
ORA-00314,redolog 损坏，或丢失处理方法
ORA-00257 归档日志过大导致无法存储的解决办法