My Veeam server has multiple network interfaces: iSCSI, production server subnet, and an interface on the LAN subnet that the ESXi hosts and vCenter Server are on. The production server subnet is the only interface with a gateway.
When doing instant restores, the ESXi host automatically mounts the NFS datastore on via the production server LAN IP address on the Veeam server, even though it should be reachable via an IP address on the same subnet as the ESXi host's management vmk port. A vmkping confirms the ESXi host can reach the Veeam server's IP address on the same subnet.
How can I enforce that the ESXi host connects to the Veeam server's NFS service on the correct subnet, rather than routing to it via the gateway, which is an unnecessary additional hop that impacts performance?
I assume that I should just unmount the NFS datastore and manually add it, using the correct IP address? I also assume that the datastore must be named "VeeamBackup_SERVER_NAME"?
Interestingly, some of my hosts have the datastore mounted twice, once using each IP address, with "(1)" or "(2)" after the datastore name, but when mounting instant restore VMs, they are presented via the incorrect production server subnet address, rather than the dedicated ESXi subnet address.