ASM磁盘头损坏如何修复

来源：这里教程网时间：2026-03-03 22:27:32 作者：

磁盘头信息备份恢复 / 无备份修复

一、 kfed 备份 / 恢复

1 、备份

kfed 
read 
/dev/data text=/tmp/data.head

2 、模拟破坏

[grid@rac1 ~]$ dd if=/dev/zero 
of
=/dev/data bs=4096 
count
=1

1+0 records 
in

1+0 records 
out

4096 bytes (4.1 kB) copied, 0.0013819 s, 3.0 MB/s

  

[grid@rac1 ~]$ kfed 
read 
/dev/data

kfbh.endian:                          0 ; 0x000: 0x00

kfbh.hard:                            0 ; 0x001: 0x00

kfbh.type:                            0 ; 0x002: KFBTYP_INVALID

kfbh.datfmt:                          0 ; 0x003: 0x00

kfbh.block.blk:                       0 ; 0x004: blk=0

kfbh.block.obj:                       0 ; 0x008: file=0

kfbh.
check
:                           0 ; 0x00c: 0x00000000

kfbh.fcn.base:                        0 ; 0x010: 0x00000000

kfbh.fcn.wrap:                        0 ; 0x014: 0x00000000

kfbh.spare1:                          0 ; 0x018: 0x00000000

kfbh.spare2:                          0 ; 0x01c: 0x00000000

7F03F92F5400 00000000 00000000 00000000 00000000  [................]

  
Repeat 255 times

KFED-00322: Invalid content encountered during block traversal: [kfbtTraverseBlock][Invalid OSM block type][][0]

  

SQL> 
alter 
diskgroup DATA mount;

alter 
diskgroup DATA mount

*

ERROR 
at 
line 1:

ORA-15032: 
not 
all 
alterations performed

ORA-15017: diskgroup 
"DATA" 
cannot be mounted

ORA-15040: diskgroup 
is 
incomplete

3 、修复

[grid@rac1 trace]$ kfed write /dev/data text=/tmp/data.head

  

[grid@rac1 trace]$ kfed 
read 
/dev/data

kfbh.endian:                          1 ; 0x000: 0x01

kfbh.hard:                          130 ; 0x001: 0x82

kfbh.type:                            1 ; 0x002: KFBTYP_DISKHEAD

......

kfdhdb.grptyp:                        1 ; 0x026: KFDGTP_EXTERNAL

kfdhdb.hdrsts:                        3 ; 0x027: KFDHDR_MEMBER

......

  

[grid@rac1 trace]$ sqlplus / 
as 
sysasm

SQL> 
alter 
diskgroup DATA mount;

Diskgroup altered.

二、使用 kfed 自动备份的恢复

从 10.2.0.5 、 11.1.0.7 、 11.2 之后 ASM 磁盘头会自动备份到 AU#1 的倒数第二个 block 。对于 AU size 是 1MB 的 DISKGROUP ，每个 AU 包括 block 数量 =1024KB/4KB=256 个，因此备份信息位于 AU#1 的第 254 号 block(block 从 0 号开始 ) 。

1 、模拟磁盘头损坏

dd if=/dev/zero 
of
=/dev/data bs=4096 
count
=1

  

SQL> 
select 
group_Number,disk_number,mount_status ,header_status 
from 
v$asm_disk;

  

GROUP_NUMBER DISK_NUMBER MOUNT_S HEADER_STATU

------------ ----------- ------- ------------

           
1           0 CACHED  CANDIDATE

           
2           0 CACHED  MEMBER

           
2           1 CACHED  MEMBER

           
2           2 CACHED  MEMBER

    

SQL> 
alter 
diskgroup DATA dismount 
force
;

  

Diskgroup altered.

  

SQL> 
alter 
diskgroup DATA mount;

alter 
diskgroup DATA mount

*

ERROR 
at 
line 1:

ORA-15032: 
not 
all 
alterations performed

ORA-15017: diskgroup 
"DATA" 
cannot be mounted

ORA-15040: diskgroup 
is 
incomplete

2 、使用自备份的修复磁盘头

[grid@rac1 trace]$ kfed repair /dev/data

  

[grid@rac1 trace]$ kfed 
read 
/dev/data

kfbh.endian:                          1 ; 0x000: 0x01

kfbh.hard:                          130 ; 0x001: 0x82

kfbh.type:                            1 ; 0x002: KFBTYP_DISKHEAD

......

kfdhdb.grptyp:                        1 ; 0x026: KFDGTP_EXTERNAL

kfdhdb.hdrsts:                        3 ; 0x027: KFDHDR_MEMBER

......

kfdhdb.grpstmp.hi:             33123181 ; 0x0e4: 
HOUR
=0xd DAYS=0x1b MNTH=0xa 
YEAR
=0x7e5

kfdhdb.grpstmp.lo:           1764439040 ; 0x0e8: USEC=0x0 MSEC=0x2cd SECS=0x12 MINS=0x1a

  

SQL> 
alter 
diskgroup DATA mount;

  

Diskgroup altered.

  

SQL> 
select 
group_Number,disk_number,mount_status ,header_status 
from 
v$asm_disk;

  

GROUP_NUMBER DISK_NUMBER MOUNT_S HEADER_STATU

------------ ----------- ------- ------------

           
1           0 CACHED  MEMBER

           
2           0 CACHED  MEMBER

           
2           1 CACHED  MEMBER

           
2           2 CACHED  MEMBER

  

SQL> 
select 
name
,path,STATE,MODE_STATUS 
from 
v$asm_disk;

  

NAME                           
PATH                                     STATE    MODE_ST

------------------------------ ---------------------------------------- -------- -------

DATA_0000                      /dev/data                                NORMAL   ONLINE

OCR_0000                       /dev/ocrvote1                            NORMAL   ONLINE

OCR_0001                       /dev/ocrvote2                            NORMAL   ONLINE

OCR_0002                       /dev/ocrvote3                            NORMAL   ONLINE

三、使用 dd 备份恢复

1 、 dd 备份磁盘头信息

[grid@rac1 trace]$ dd if=/dev/data 
of
=/tmp/data2.bak bs=4096 
count
=1

1+0 records 
in

1+0 records 
out

4096 bytes (4.1 kB) copied, 0.000103484 s, 39.6 MB/s

2 、模拟破坏

[grid@rac1 trace]$ dd if=/dev/zero 
of
=/dev/data bs=4096 
count
=1

1+0 records 
in

1+0 records 
out

4096 bytes (4.1 kB) copied, 0.00127311 s, 3.2 MB/s

3 、检查磁盘和磁盘组状态

SQL> 
select 
name
,path,STATE,MODE_STATUS 
from 
v$asm_disk;

  

NAME                           
PATH                                     STATE    MODE_ST

------------------------------ ---------------------------------------- -------- -------

DATA_0000                      /dev/data                                NORMAL   ONLINE

OCR_0000                       /dev/ocrvote1                            NORMAL   ONLINE

OCR_0001                       /dev/ocrvote2                            NORMAL   ONLINE

OCR_0002                       /dev/ocrvote3                            NORMAL   ONLINE

  

SQL> 
select 
group_Number,disk_number,mount_status ,header_status 
from 
v$asm_disk;

  

GROUP_NUMBER DISK_NUMBER MOUNT_S HEADER_STATU

------------ ----------- ------- ------------

           
1           0 CACHED  CANDIDATE

           
2           0 CACHED  MEMBER

           
2           1 CACHED  MEMBER

           
2           2 CACHED  MEMBER

  

SQL> 
alter 
diskgroup DATA dismount 
force
;

  

Diskgroup altered.

  

SQL> 
select 
group_Number,disk_number,mount_status ,header_status 
from 
v$asm_disk;

  

GROUP_NUMBER DISK_NUMBER MOUNT_S HEADER_STATU

------------ ----------- ------- ------------

           
0           0 CLOSED  CANDIDATE

           
2           0 CACHED  MEMBER

           
2           1 CACHED  MEMBER

           
2           2 CACHED  MEMBER

  

SQL> 
alter 
diskgroup DATA mount;

alter 
diskgroup DATA mount

*

ERROR 
at 
line 1:

ORA-15032: 
not 
all 
alterations performed

ORA-15017: diskgroup 
"DATA" 
cannot be mounted

ORA-15040: diskgroup 
is 
incomplete

4 、使用dd 备份恢复

[grid@rac1 trace]$ dd if=/tmp/data2.bak 
of
=/dev/data

8+0 records 
in

8+0 records 
out

4096 bytes (4.1 kB) copied, 0.00125491 s, 3.3 MB/s

  

SQL> 
select 
group_Number,disk_number,mount_status ,header_status 
from 
v$asm_disk;

  

GROUP_NUMBER DISK_NUMBER MOUNT_S HEADER_STATU

------------ ----------- ------- ------------

           
1           0 CACHED  MEMBER

           
2           0 CACHED  MEMBER

           
2           1 CACHED  MEMBER

           
2           2 CACHED  MEMBER