It is likely that the VMDK files are presented in a different order on each system and that the names given by Volume Manager may vary. The recommended best practice for a consistent deployment is to rename the disk so the configuration is clear.

As an example of the initial discrepancies between cfs01 and cfs03, cfs01 the disk name associated to device ending on serial number 226 is vmdk0_5:

[root@cfs01 ~]# /etc/vx/bin/vxgetdmpnames
enclosure vendor=VMware product=disk serial=vmdk name=vmdk0
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_1
								dmpnode serial=6000c29ac083abd0a86fa46b509d69f5 name=vmdk0_2
								dmpnode serial=6000c29e13f6aff58ac3d543b022dfe2 name=vmdk0_3
								dmpnode serial=6000c29f2a8d6030ddf71042d4620cec name=vmdk0_4
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_5

Observe how cfs03 named the same device vmdk_0_0:

[root@cfs01 ~]# /etc/vx/bin/vxgetdmpnames
enclosure vendor=VMware product=disk serial=vmdk name=vmdk0
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_1
								dmpnode serial=6000c29ac083abd0a86fa46b509d69f5 name=vmdk0_2
								dmpnode serial=6000c29e13f6aff58ac3d543b022dfe2 name=vmdk0_3
								dmpnode serial=6000c29f2a8d6030ddf71042d4620cec name=vmdk0_4
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_5

In order to get the same names across all the cluster nodes the command vxddladm is used. For each node of the cluster, run the command:

# vxddladm assign names

Observe now how cfs03 got the right name for device ending at 226 serial number:

[root@cfs01 ~]# /etc/vx/bin/vxgetdmpnames
enclosure vendor=VMware product=disk serial=vmdk name=vmdk0
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_1
								dmpnode serial=6000c29ac083abd0a86fa46b509d69f5 name=vmdk0_2
								dmpnode serial=6000c29e13f6aff58ac3d543b022dfe2 name=vmdk0_3
								dmpnode serial=6000c29f2a8d6030ddf71042d4620cec name=vmdk0_4
								dmpnode serial=6000c2993a8d6030ddf71042d4620cec name=vmdk0_5