We are having the same issue, and I have a case open with VMware and HP about it. (Our environments runs on HP's Simplivity platform)
Issue started occurring after upgrading our ESXi hosts and vcenter to 6.7, and Horizon to 7.6
When a non-persistant linked-clone attempts an automatic refresh after logout, it enters an error state. We then have to manually remove it so a new one is provisioned, and then it creates a new vm with the _# extension on the end of it.
