While we were doing POC, we were faced with RDM cluster and failover.
Let’s assume that there are two hosts clustered by RDM(using pacemaker for Linux) for DB server.
Plus, one more host for replicate and failover (not storage, just a normal server)
If so, target host will get VMDK file during replicating RDM clustered hosts.
And then, Here is one question.
Will target hosts’ VMs operate well if we failover source VM to target?
If this one works, There is the other question.
Will VMDK files turn into RDM if I failback those VMs?
If you hard to understand this, please let me know.
About whether a failed-over VM can work well, you can test it by initiating a Failover Test (FOT) or Offsite Clone, before starting the Failover Live (FOL).
After a FOL, the VPG can either be “reconfigured” or “removed & recreated”, as it will be in “Needs Configuration” status.
After reconfigured or recreated, the VPG might have the same name and some settings as prior to the FOL, but the VPG is practically a new one.
In the VPG reconfiguration or recreation process, you can select a datastore for the RDM to replicate a .vmdk file.
This means the RDM will be the source/protected disk.
There are some limitations of RDM being a protected disk. Please refer to Page 58 and other RDM related sections of the ZVM administration guide: