Select next-hop based on source address

Question

I'm trying to set up a Ubuntu-based gateway between clients and the core router, the ideal traffic flow should be:

ClientGroup 1 (10.0.0.0/24) -> Ubuntu gateway (10.0.0.2) -> Core router (10.0.0.1) -> Internet

ClientGroup 2 (10.0.1.0/24) -> Ubuntu gateway (10.0.1.2) -> Core router (10.0.1.1) -> Internet

ClientGroup 3 (10.0.2.0/24) -> Ubuntu gateway (10.0.2.2) -> Core router (10.0.2.1) -> Internet

Clients use Ubuntu instance as their gateway.

The following netplan configuration routes internal traffic properly, however for Internet traffic, it defaults to eth0.1 (10.0.1.1) because of gateway4.

network: version: 2 ethernets: eth0: addresses: [10.0.0.2/24] routes: - to: 10.0.0.0/24 via: 10.0.0.1 table: 101 routing-policy: - from: 10.0.0.0/24 table: 101 dhcp4: false vlans: eth0.1: id: 1 link: eth0 addresses: [10.0.1.2/24] gateway4: 10.0.1.1 routes: - to: 10.0.1.0/24 via: 10.0.1.1 table: 102 routing-policy: - from: 10.0.1.0/24 table: 102 dhcp4: false eth0.2: id: 2 link: eth0 addresses: [10.0.2.2/24] routes: - to: 10.0.2.0/24 via: 10.0.2.1 table: 103 routing-policy: - from: 10.0.2.0/24 table: 103 dhcp4: false 

Any ideas on how to route all traffic to its corresponding next-hop on the core router? Namely, all traffic (0.0.0.0/0) from 10.0.0.0/24 should be routed to 10.0.0.1 whereas 10.0.2.0/24 should be routed to 10.0.2.1.

I'm happy to use plain ip route rules as well.

Thanks in advance!

---

Edit 13/05/20:

I have added a default route to each routes, the routing is working as expected but traceroute output seems weird:

routes: - to: 0.0.0.0/0 via: 10.0.0.1 table: 101 

1st test:

traceroute to 1.1.1.1 (1.1.1.1), 64 hops max, 52 byte packets 1 10.0.0.2 (10.0.0.2) 0.896 ms * * 2 10.0.0.1 (10.0.0.1) 1.361 ms 1.126 ms 0.879 ms 

2nd test:

traceroute to 1.1.1.1 (1.1.1.1), 64 hops max, 52 byte packets 1 * * * 2 10.0.0.1 (10.0.0.1) 1.353 ms 1.062 ms 0.825 ms 

Answer 1

Easy enough, particularly since you're already using multiple routing tables:

(Sorry, didn't convert my example to netplan; but since you're game to use regular ip route rules, it should suffice - and this example assumes you're starting from scratch)

So, your netplan is already creating three routing tables, so we'll use the first routing table you have:

ip route flush 101 ip route add table 101 to 10.0.0.0/24 dev eth0 ip route add table 101 to default via 10.0.0.1 

And the third routing table you have:

ip route flush 103 ip route add table 103 to 10.0.2.0/24 dev eth0.2 ip route add table 103 to default via 10.0.2.1 

Then create two rules to send traffic to each table based on the traffic's source IP address:

ip rule add from 10.0.0.0/24 table 101 priority 101 ip rule add from 10.0.2.0/24 table 103 priority 103 

This should do it. Note that you can keep adding destination routes to the two tables and only traffic from the associated sources will hit those entries - enough of those and you may choose to remove the default routes on each table.