I’ve never researched anything about how route metric number work other than it’s probably “higher metric wins” (or the other way around?) Anyway, today I added new interface to one of my VMs and repatched it. My goal was to change the default route from eth0 to newly created eth1. NetworkManager has some rules to automatically detect these, however since I am unsure how all of this works I’ve decided to solve it this way.
Once the new interface named eth1 appeared, I configured it for DHCP:
nmcli con add type ethernet con-name eth1 ifname eth1
Of course, two default gateways appeared in my routing table, both eth0 and eth1 with metric of 100 and 101 respectively. Not what I wanted since I repatched the VM and I wanted only eth1 to be the default one. Here comes the trick, you can tell the NetworkManager to never create default route with this:
nmcli c mod eth0 ipv4.never-default true
nmcli c mod eth0 ipv6.never-default true
And that’s all! All you need to do is to restart existing connection to clear the routing table or manually remove the entry. Let’s do the safe method ensuring everything is set accordingly however watch out: the following can cut the branch if you are doing this over SSH connection (use screen or tmux):
nmcli c down eth0; nmcli c up eth0
Done.
# ip r
default via 192.168.122.1 dev eth1 proto dhcp metric 101
192.168.122.0/24 dev eth1 proto kernel scope link src 192.168.122.60 metric 101
192.168.199.0/24 dev eth0 proto kernel scope link src 192.168.199.11 metric 102
Easy peasy with NetworkManager.