In a scenario where the customer only has a single ESX host at each site in a VMware cluster with NO HA or DRS enabled and it is utilizing only localized storage (no shared datastores), the DM is not discovering the locally attached datastore as a viable option to place the repository volume in the final steps.
Understanding there will be warnings about the vRPA's not being isolated on their own ESX hosts (2 vRPAs in this case), is this a supported and viable configuration? I suspect it can be made to work and the DM is just looking for shared datastores since the 2 vRPA's will technically be able to see the same local datastore to create the repository volume it needs. I don't see anything explicit saying it can't work, but I am also not seeing the datastores visible to complete the cluster install
This is the only blurb indicating a shared datastore where multiple ESX hosts are involved, in our case there is only a single ESX host for a POC test:
Thanks for any help.
I had similar configs working before in a virtual lab env, not sure if we’ve blocked that in DM now, but will check & someone will get back to you.
Thanks Zahid! That is greatly appreciated. I expected it to work and will validate the vRPA placement on the same ESX with the customer tomorrow so it's good to hear that it was working in your lab from a technical perspective.
Thanks again for the help!
I was able to get around the issue. Once I moved the host to its own cluster (isolated from the rest) it was able to discover the datastores that were locally attached. It seems that if the single host is part of an ESX cluster, even with HA/DRS off, that the DM won't show the local datastores - I'm guessing due to how it presents the "splitter" at the cluster level. So once I moved the host to it's own isolated new cluster, we were able to complete the install. Not sure if this is expected from the DM side, but the workaround seemed to work for this test.
What you're seeing looks like the expected behavior. DM will always show only the shared datastore if there's more than 1 ESX in a cluster. If there's only 1 ESX on a given HA cluster then all of the datastores will be displayed.
This is to protect against configuring the repository on dedicated datastore which in vMotion/HA restart will cause a definite access loss to the repository volume.
Hope that helps,
RecoverPoint Corporate Systems Engineering
Cool...thanks Idan...that is good to note. That was exactly my thinking trying it out today so I appreciate the follow-up!
Hope you guys are doing well!