asm disk header 的自动备份和恢复

本文参考：http://www.dbform.com/html/2012/1875.html
http://www.xifenfei.com/2012/06/%E4%BD%BF%E7%94%A8asm-disk-header-%E8%87%AA%E5%8A%A8%E5%A4%87%E4%BB%BD%E4%BF%A1%E6%81%AF%E6%81%A2%E5%A4%8D.html
   在Oracle 10.2.0.5之前，ASM磁盘的头块并没有自己的备份，因此一旦头块损坏，如果没有以前kfed read备份出来的信息，也就没有办法使用kfed merge来作头块恢复，特别是如果一个磁盘组中所有的磁盘头块都出现问题（比如被人为地创建了PV），恢复ASM磁盘头块的操作就会非常麻烦。

   但是从Oracle 10.2.0.5之后，ASM磁盘的头块会自动备份在另外一个块中，这实际上是Oracle 11g出现的功能，不过经过测试，在Oracle 10.2.0.5版本中，这个备份也是存在的。

   对于10.2.0.5.0以及以后版本，不管au size是多少,asm disk header自动备份存储的位置是第2个au的倒数第2个block.
计算方法:AU中包含的block
num[AU_SIZE/block_size]*2-2[因为从第一个块从0计数],通过该方法计算结论为:
         1M AU在510
         2M AU在1022
         4M AU在2046
         8M AU在4094
         16M AU在8190
         32M AU在16382
         64M AU在32766

   正是因为存在这个备份，所以Oracle 10.2.0.5之后的kfed程序才有了新的repair命令，该命令将备份块直接覆盖到磁盘头块，完成修复工作。

   在Oracle 10.2.0.4中，如果尝试执行kfed repair，则会报错说命令行参数不正确，此报错说明并不存在repair命令:

$ kfed repair
KFED-00101: LRM error [102] while parsing command line arguments

   但是在Oracle 10.2.0.5中，执行kfed repair，则会说无法打开文件空，而这正说明repair命令是存在的，报错是因为还需要明确指定要修复哪块磁盘：

$ kfed repair
KFED-00303: unable to open file ''

编译kfed程序

cd $ORACLE_HOME/rdbms/lib
cp ins_rdbms.mk ins_rdbms.mk.bak
make -f ins_rdbms.mk ikfed

数据库版本

SQL> select * from v$version;

BANNER
--------------------------------------------------------------------------------
Oracle Database 11g Enterprise Edition Release 11.2.0.4.0 - 64bit Production
PL/SQL Release 11.2.0.4.0 - Production
CORE    11.2.0.4.0      Production
TNS for Linux: Version 11.2.0.4.0 - Production
NLSRTL Version 11.2.0.4.0 - Production

ASM磁盘属性

SQL> set linesize 300
SQL> col PATH format a30
SQL> select group_number,DISK_NUMBER,PATH,HEADER_STATUS from v$asm_disk where group_number<>0;

GROUP_NUMBER DISK_NUMBER PATH                           HEADER_STATU
------------ ----------- ------------------------------ ------------
4           2 /dev/mapper/cdpfra1            MEMBER
4           3 /dev/mapper/cdpfra2            MEMBER
3           7 /dev/mapper/cdpdata4           MEMBER
3           6 /dev/mapper/cdpdata3           MEMBER
3           4 /dev/mapper/cdpdata1           MEMBER
3           5 /dev/mapper/cdpdata2           MEMBER

SQL> select group_number,name,BLOCK_SIZE,ALLOCATION_UNIT_SIZE from v$asm_diskgroup;

GROUP_NUMBER NAME                           BLOCK_SIZE ALLOCATION_UNIT_SIZE
------------ ------------------------------ ---------- --------------------
4 FRA                                  4096              1048576
3 DATA                                 4096              1048576

验证自动备份asm disk header功能

$ kfed read /dev/mapper/cdpfra1 blknum=510|>/tmp/diskhead.510
$ kfed read /dev/mapper/cdpfra1 blknum=0|>/tmp/diskhead.0

$ ll /tmp/diskhead.*
-rw-r--r-- 1 grid oinstall 0 Nov  5 09:47 /tmp/diskhead.0
-rw-r--r-- 1 grid oinstall 0 Nov  5 09:46 /tmp/diskhead.510

$ diff /tmp/diskhead.0 /tmp/diskhead.510
--通过对比发现两者无不同记录返回,证明他们记录内容完全相同

破坏asm disk header

[grid@hlwzf1 lib]$  kfed read /dev/mapper/cdpfra1
kfbh.endian:                          1 ; 0x000: 0x01
kfbh.hard:                          130 ; 0x001: 0x82
kfbh.type:                            1 ; 0x002: KFBTYP_DISKHEAD
kfbh.datfmt:                          1 ; 0x003: 0x01
kfbh.block.blk:                       0 ; 0x004: blk=0
kfbh.block.obj:              2147483650 ; 0x008: disk=2
kfbh.check:                    49
d376
794540 ; 0x00c: 0x02f7cdec
kfbh.fcn.base:                     2493 ; 0x010: 0x000009bd
kfbh.fcn.wrap:                        0 ; 0x014: 0x00000000
kfbh.spare1:                          0 ; 0x018: 0x00000000
kfbh.spare2:                          0 ; 0x01c: 0x00000000
kfdhdb.driver.provstr:         ORCLDISK ; 0x000: length=8

$ dd if=/dev/zero of=/dev/mapper/cdpfra1 bs=4096 count=1
1+0 records in
1+0 records out
4096 bytes (4.1 kB) copied, 0.00430849 s, 951 kB/s

$ kfed read /dev/mapper/cdpfra1 blknum=0
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

查看破坏后的磁盘属性和操作

SQL> set linesize 300
SQL> col PATH format a30
SQL> select group_number,DISK_NUMBER,PATH,HEADER_STATUS from v$asm_disk where group_number<>0;

GROUP_NUMBER DISK_NUMBER PATH                           HEADER_STATU
------------ ----------- ------------------------------ ------------
4           2 /dev/mapper/cdpfra1            CANDIDATE
4           3 /dev/mapper/cdpfra2            MEMBER
3           7 /dev/mapper/cdpdata4           MEMBER
3           6 /dev/mapper/cdpdata3           MEMBER
3           4 /dev/mapper/cdpdata1           MEMBER
3           5 /dev/mapper/cdpdata2           MEMBER

SQL>  alter diskgroup fra dismount;

Diskgroup altered.

SQL> alter diskgroup fra mount;
alter diskgroup fra mount
*
ERROR at line 1:
ORA-15032: not all alterations performed
ORA-15040: diskgroup is incomplete
ORA-15042: ASM disk "2" is missing from group number "4"

使用kfed repair修改损坏asm disk header

$ kfed repair '/dev/mapper/cdpfra1'

$ kfed read /dev/mapper/cdpfra1 blknum=0
kfbh.endian:                          1 ; 0x000: 0x01
kfbh.hard:                          130 ; 0x001: 0x82
kfbh.type:                            1 ; 0x002: KFBTYP_DISKHEAD
kfbh.datfmt:                          1 ; 0x003: 0x01
kfbh.block.blk:                       0 ; 0x004: blk=0
kfbh.block.obj:              2147483650 ; 0x008: disk=2
kfbh.check:                    49794540 ; 0x00c: 0x02f7cdec
kfbh.fcn.base:                     2493 ; 0x010: 0x000009bd
kfbh.fcn.wrap:                        0 ; 0x014: 0x00000000
kfbh.spare1:                          0 ; 0x018: 0x00000000
kfbh.spare2:                          0 ; 0x01c: 0x00000000
kfdhdb.driver.provstr:         ORCLDISK ; 0x000: length=8

SQL> alter diskgroup fra mount;

Diskgroup altered.

如果是手动备份asm disk header，比如前面的

kfed read /dev/mapper/cdpfra1 blknum=0|>/tmp/diskhead.0

还可以使用kfed merge恢复

kfed merge /dev/mapper/cdpfra1 /tmp/diskhead.0