A few days ago we had a quick request at a customer which had to replace a voting disk in an ASM disk group. The current ASM disk group (redundancy normal) had 3 disks: Two were located in the “VE” data center and one disk in the “SH” data center. The goal of this operation was to replace one voting disk from the “VE” data center with one disk coming from a third Data Center (“SH2”). Many thanks to Antoine Marty working at this customer for having performed these steps!

In order to achieve this “mission”, we first thought about the classical “crsctl replace votedisk” commands. Even the documentation evokes the “crsctl replace votedisk +asm_dg” synopsis, as you want to replace voting within the same ASM disk group, you do not have to use any “crsctl replace votedisk” anymore, Grid Infrastructure 11.2 & ASM collaborate together and simplify your life 🙂

Have a look below, at the begin, we had the following situation:

$ crsctl query css votedisk
##  STATE    File Universal Id                File Name Disk group
--  -----    -----------------                --------- ---------
 1. ONLINE   34a93f34e8094f4fbfe611fc3fdcd21a (/dev/oracle/vot_sh1_1122) [CRS11G]
 2. ONLINE   1758ce7cfd7a4f87bf4ec88716b7945e (/dev/oracle/vot_ve1_1123) [CRS11G]
 3. ONLINE   84b2c9d7352f4ffebf6d0ea0f5c3bdf6 (/dev/oracle/vot_ve1_1191) [CRS11G]
Located 3 voting disk(s).

Two voting disks in the “VE” data center and one in the “SH” data center. Of course, as we all know, if the VE data center is lost, the Grid Infrastructure 11.2 cluster goes down (majority of the voting disk is lost).

The strategy to replace one voting from the “VE” data center while using ASM is very easy, just add a new disk, remove the one you want to suppress and Grid Infrastructure 11.2 automatically uses this new disk to maintain the “normal” redundancy (that is 3 voting disks in the ASM disk group).

For this purpose we can use the Grid Control 11g interface. Below the original situation:

As we can see, we have 3 disks in the ASM disk group. We decided to add the new one (click the Add button at the right):

Add the name of the new disk (coming from the new data center “SH2”) and eventually its size:

We now have 4 disks in the disk group. However at “Grid Infrastructure” level, nothing happened yet, we still see the same voting disks:

$ crsctl query css votedisk
##  STATE    File Universal Id                File Name Disk group
--  -----    -----------------                --------- ---------
 1. ONLINE   34a93f34e8094f4fbfe611fc3fdcd21a (/dev/oracle/vot_sh1_1122) [CRS11G]
 2. ONLINE   1758ce7cfd7a4f87bf4ec88716b7945e (/dev/oracle/vot_ve1_1123) [CRS11G]
 3. ONLINE   84b2c9d7352f4ffebf6d0ea0f5c3bdf6 (/dev/oracle/vot_ve1_1191) [CRS11G]
Located 3 voting disk(s).

We removed one disk from the “VE” data center, still using Grid Control 11g (click on the disk and click remove):

As a result, we have only 3 disk in the disk group (one from the “VE” data center, one from “SH” and one from “SH2”) :

Here comes the real good news, at “Grid Infrastructure 11.2” level, we observe that Oracle automatically used the disk coming from the new Data Center “SH2” as the new voting disk (to maintain the normal voring disk redundancy : 3 voting disks) :

$ crsctl query css votedisk
##  STATE    File Universal Id                File Name Disk group
--  -----    -----------------                --------- ---------
 1. ONLINE   34a93f34e8094f4fbfe611fc3fdcd21a (/dev/oracle/vot_sh1_1122) [CRS11G]
 2. ONLINE   1758ce7cfd7a4f87bf4ec88716b7945e (/dev/oracle/vot_ve1_1123) [CRS11G]
 3. ONLINE   a558160426304fa0bfa18fdc5c56f3bb (/dev/oracle/vot_sh2_0287) [CRS11G]
Located 3 voting disk(s).

Grid Infrastructure 11.2, ASM and Grid Control 11g do really make the administrator’s life easier 🙂 !
Many thanks again to Antoine Marty for having performed the steps and having produced the nice screenshots.