I have finally discovered what is causing some issues on my network with random connectivity loss on some vlans and it is to do with policy routing and bandwidth control.

My setup:

Virgin Business UK, 5 static IPs, Hitrons Chita ISP modem which establishes a GRE tunnel to VMB datacentre, and the 5 statics are accessible over that GRE tunnel to the LAN ports on the modem without any extra NAT, which can then be used by third party routers for internal gateways, nat etc.  I would imagine your technical team are aware of how this works.

In my case, i have 3 links to my ER8411 WAN ports to facilitate my use of 3 of my public IPs to distribute over my VLans using policy routing to keep internal and public networks on seperate public IPs.  nothing unusual.  This works perfectly internally to assign different external IPs per network.  I have policy routing enabled to force different VLANs onto different wan ports on my ER8411 to, for example, ensure my public wifi network always goes out on a different IP to an internal network.

However, on both 1.2.1 and 1.2.2 beta firmware, as soon as the policy routing and bandwidth control "kicks in" after a router reboot, the upstream Hitrons modem ends up with at least one of the WAN ports on the ER8411 reporting 2 different IP addresses to the Hitrons, causing packet losses through NAT on traffic going back into my networks

Image below

As you can see, the WAN port with MAC ending 4E-14 is now reporting 2 IP addresses (both in my public IP range) one of which overlaps with WAN port MAC ending 4E-14

4E-13 is set on my ER8411 as static ending .90 and 4E-14 is set as static ending .92

Before policy routing is activated on the router when it boots, each WAN port has its own singular IP address with no overlap.