Map each of the created VMDK files to each VM. The example procedure illustrates mapping the VMDKs to the cfs01 node: all steps should be followed for each of the other nodes.
To map the VMDKs to each VM
and click Next.
Modify the type to Paravirtual, if that is not the default, and check that SCSI Bus Sharing is set to None, as this is key to allow vMotion for the VMs.
For the example configuration, the parameters for steps 5-7 are given in the table below:
Data Store |
VMDK Name |
Virtual Device |
---|---|---|
DS1 |
cfs0/shared1.vmdk |
SCSI 1:0 |
DS2 |
cfs0/shared2.vmdk |
SCSI 1:1 |
DS3 |
cfs0/shared3.vmdk |
SCSI 1:2 |
DS4 |
cfs0/shared4.vmdk |
SCSI 1:3 |
DS5 |
cfs0/shared5.vmdk |
SCSI 1:4 |
The final configuration for the first node of the example cluster (cfs01):
Now follow the same steps for each node of the cluster and map each VMDK file to the VM following the instructions above. Once all the steps are completed, all the VMs should have access to the same VMDK files. Note that at this point, all the VMs are still powered off and that multi-writer flag has not been enabled yet (it will be done in the next step). Any attempt to power on the VMs in this state will prevent a second VM start because it will violate the restrictions to access a VMDK by only a host at a time.