Hello.
I have configured connection through two providers. It works in balancing mode, PCC, first mark connection, then mark routing, two tables, routes in IP-Routing.
Question. Is it correct to make one marking for VPN (no matter whose packet is IPS1 or IPS2), conventionally: first IPS1_route_mark and IPS2_route_mark, then VPN_route_mark. Or make two VPN labels (and two tables) - conventionally: first IPS1_route_mark and IPS2_route_mark, then IPS1-VPN_route_mark and IPS2-VPN_route_mark.
What confuses me is that essentially the VPN interface is running on top of the ISP (or rather two ISPs).
What is better is not to twist yourself into a pretzel about the config…
What you should do is communicate clearly your requirements
a. identify users
b. identify traffic they need ( for example LAN1 and LAN2 might need PCC, but LAn3 only WAN1, or a group of users or devices has specificity )
c. elaborate on WAN connections
d. decide how WANS will be used ( primary/backup, pcc load balance etc..)
-detail what happens when any of the WANs fails, what happens with its traffic.
Then you need to ensure you detail if there is any port forwarding going …which WAN
Then you need to ensure you detail if there is any VPNs going on…which WAN.
Then one can start to produce a coherent config.