Simple 2 Node cluster config
Let me paint this simple scenario: I have two servers, each with local disks attached to a RAID controller. Based on those servers I want to create a highly available, resiliant file server.
The simple solution would be: Use either Storage Spaces Direct or Storage Replica. Unfortunately both options are not available right now.
S2D fails a) because of the use of a RAID controller and b) because of the use of only 2 servers.
Replica fails because I either have no automatic failover (server 2 server mode) or I don't have disks that support persistent reservation (which is required for cluster mode)
Now I would like to propose following solution:
Allow me to use disks that do not support SCSI persistent reservation for a clustered SR config. (Either through allowing them to come online in the cluster, or by not needing the disks to be part of the cluster in the first place.)
Nowadays there is no more need for PR in a cluster, except when using a Disk as a witness. Everything else in the cluster can these days rely on the quorum mechanism and just be good with it. So that change should be obvious anyway.
Also there is no need to have an SR disk in the cluster. We know that, after all SR works Server 2 Server as well... All I want is a Server 2 Server SR that has a floating IP on top. Why is this blocked in the first place?
I posted both scenarios (S2D and SR) on connect as well. Would appreciate your votes on it.
You can achieve an HA file server using SR? I have built this several times as a POC. On each physical server configure a VM with failover cluster, SR, & file server role. attach a couple of VHDX and you can use these as your CSV for data and logs, rinse and repeat on the other physical. works perfectly. cannot fault it.
Admittedly - this is not giving you SR storage for VMs on the physical themselves.
S2D is not an option with 2 nodes IMO. bit odd for MS to support S2D on 2 nodes.