We were facing this issue with disk errors and cluster node getting out for past one month since we have shifted our cluster disk from RDM to VMDK. Yesterday I got this info. that mcsc cluster across two ESX cluster cannot be configured on virtual disk and can only be on RDMs i.e.
And this thing does have a logic as we were not facing these issues when we hade them on RDMs. We need to perform a storage change activity where we cloned and decided to stay cluster shared disks as .vmdk.
My first question is, is this reason justified. Will this be resolved after moving them back to RDMs(possibly by backup restore)
Secondly, if this is the case, what is the significance of the Physical mode option available for shared scsi in case od virtual disks(vmdks). If they cant be used with shared across ESX servers why is the physcial mode option given there?
As you already saw in the documentation you mention, for the so called "Cluster across boxes" you definitely need Pass-through RDM's (physical compatibility mode). So the answer to your first question is "Yes"
Depending on what you want to achieve, follow the documentation (btw. http://kb.vmware.com/kb/1004617 contains links to the MSCS docs for each ESX version). Page 12-13 in the 4.0 doc shows all possible storage types for the different cluster types.
yes specifications are provided but the answer is not given.
With cluster on same base....no such requirement is there.
Wehereas on cluster across base, its mentioned only RDM will work.
And thats why I am asking, whats the use of giving physical option in case of virtual disks. When does it comes into use ?