We are having a recurring issue (most recently Case # 00941337) where a replica VM becomes invalid on the destination vCenter. This has been happening since Veeam 7.X (we are currently on 22.214.171.1247) on a replication environment of about 40 VMs, some of which replicate as often as every 15 minutes. We are currently on vSphere 5.5, fully patched on both sides, Also, the VMs are relatively heavily provisioned with up to 24 vCPU and 32 GB of RAM. Out ISP is gigabit and we use VPN, and have very low packet loss.
Once or twice a month, one of the VMs will become invalid on the destination vCenter. We have determined that the *.vmx file is being truncated, with about half of it missing, but the rest of the lines intact. This has happened on many different VMs, with different configurations and replication frequencies. It has even happened on a different Veeam server, with different proxies.
To date, the only resolution is to delete everything but the virtual disks (*.vmdk's), create a new VM and attach the disks, then re-map the replication job. Veeam has to digest the disks, but thereafter replication will proceed until it happens again. This is obviously very inconvenient and places us well outside our target RPO when it occurs.