最近一周之内,两个客户由于错误修改asm_diskstring=’/dev/dm*’,’/dev/mapper/*’,导致后续集群无法启动成功,查看asm的alert日志发现类似错误(为了说明类似问题，实际中情况比现在复杂)

WARNING: detected duplicate paths to the same disk:
  '/dev/mapper/mpathi' and
  '/dev/dm-3'
ERROR: diskgroup CRS was not mounted
ORA-15032: not all alterations performed
ORA-15017: diskgroup "CRS" cannot be mounted
ORA-15040: diskgroup is incomplete

由于asm的vote磁盘组无法mount,从而使得crs无法正常启动,进而导致集群无法正常启动和工作.而在系统中mpathi和dm-3是同一块磁盘

[root@rac1 mapper]# ls -ltr mpathi
lrwxrwxrwx 1 root root       7 May 24 13:09 mpathi -> ../dm-3

GPnP profile中关于asm_diskstring配置

[grid@rac1 trace]$  gpnptool get -o-
……
<orcl:ASM-Profile id="asm" DiscoveryString="/dev/dm*,/dev/mapper/mpath*" 
SPFile="+DATA/cimp-cluster/asmparameterfile/registry.253.889969697"/>
……

对于这样的情况,需要把profile中值修改了才行,通过重建asm spfile来实现

---pfile
large_pool_size= 12M
instance_type= 'asm'
remote_login_passwordfile= 'EXCLUSIVE'
asm_diskstring= '/dev/mapper/*'
asm_power_limit= 1
diagnostic_dest= '/opt/app/grid'

创建spfile(create spfile=’+CRS’ from pfile=’/tmp/pfile’)

NOTE: Attempting voting file refresh on diskgroup CRS
NOTE: Refresh completed on diskgroup CRS. No voting file found.
NOTE: Voting file relocation is required in diskgroup CRS
NOTE: Attempting voting file relocation on diskgroup CRS
NOTE: voting file deletion on grp 1 disk CRS_0000
NOTE: voting file deletion on grp 1 disk CRS_0001
NOTE: voting file deletion on grp 1 disk CRS_0002
NOTE: No voting file found on diskgroup CRS
Sun May 24 13:23:25 2026
NOTE: updated gpnp profile ASM SPFILE to 
NOTE: updated gpnp profile ASM diskstring: /dev/mapper/*
NOTE: updated gpnp profile ASM diskstring: /dev/mapper/*
NOTE: updated gpnp profile ASM SPFILE to +CRS/cluster/asmparameterfile/registry.253.1234099407

主要注意在create spfile的过程中voting file的信息被删除,使用kfed进行验证

[root@rac1 dbs]# kfed read /dev/asm_mpathi|grep vf
kfdhdb.vfstart:                       0 ; 0x0ec: 0x00000000
kfdhdb.vfend:                         0 ; 0x0f0: 0x00000000
[root@rac1 dbs]# kfed read /dev/asm_mpatha|grep vf
kfdhdb.vfstart:                       0 ; 0x0ec: 0x00000000
kfdhdb.vfend:                         0 ; 0x0f0: 0x00000000
[root@rac1 dbs]# kfed read /dev/asm_mpathd|grep vf
kfdhdb.vfstart:                       0 ; 0x0ec: 0x00000000
kfdhdb.vfend:                         0 ; 0x0f0: 0x00000000

需要执行replace votedisk

[root@rac1 dbs]# crsctl replace votedisk +CRS
Successful addition of voting disk 99df870869c64f78bf944b40bfff8644.
Successful addition of voting disk e6db086a74d64f49bf0aff02944bcac5.
Successful addition of voting disk e88a412fe04b4fe6bf1ca067bca779b8.
Successful deletion of voting disk 615579e778684f54bf31c0ce83709f37.
Successful deletion of voting disk 9e48b89f95084fefbfb7648897508684.
Successful deletion of voting disk 274b913b3b874f7cbfb6b025028b4eaa.
Successfully replaced voting disk group with +CRS.
CRS-4266: Voting file(s) successfully replaced

这样操作完成之后,再次查看GPnP profile中关于asm_diskstring配置

[grid@rac1 trace]$  gpnptool get -o-
……
<orcl:ASM-Profile id="asm" DiscoveryString="/dev/asm_*" 
SPFile="+DATA/cimp-cluster/asmparameterfile/registry.253.1234099407"/>
……

然后再次重启集群,一切恢复正常,解决了asm_diskstring=’/dev/dm*’,’/dev/mapper/*’误操作引起的WARNING: detected duplicate paths to the same disk:故障