Some simple network issues - how to properly route two ISP with bridges?

1 post / 0 new
#1 Fri, 10/25/2013 - 06:02
fakemoth
fakemoth's picture

Some simple network issues - how to properly route two ISP with bridges?

Hello, as the subject suggests it's about something like that: I have 2 ISP, for now both getting into the same CentOS 6.4 server in eth0 (ISP1) ip 1.1.1.143 gateway 1.1.1.129 netmask 255.255.255.192 and in eth2 (ISP2) ip 2.2.2.223 gateway 2.2.2.193 netmask 255.255.255.192. By default only one default route works so had to do something about it.

Now, before going Cloudmin PRO, I added in network-scripts 2 files for each eth: route-eth0, rule-eth0, route-eth2, rule-eth2, and created in rt_tables two tables named ISP1 and ISP2. Worked fine.

The files were containing something like:

route-eth0:
1.1.1.0/24 dev eth0 src 1.1.1.143 table ISP1
default via 1.1.1.129 dev eth0 table ISP1

rule-eth0:
from 1.1.1.143/32 table ISP1
to 1.1.1.143/32 table ISP1

Everything was allright until I installed Cloudmin and created a br0 interface, bridged to eth2, and running a KVM machine with IP 2.2.2.221. So how can I achieve the same setup with bridges involved? I mean each ISP using it's own gateway, both active so I can login and use the other if the main goes down (ISP1 it's the default for Virtualmin domains), and still using the bridge for VMs with ISP2 (default for Cloudmin's VMs)? Getting to ping 1.1.1.143, 2.2.2.221 and 2.2.2.223 would be a nice start :)

Adding a BR0 table and file with the same recipe did not work even if I deleted the ISP2 table and files... Never encountered a situation like this and the Google searches throw garbage at me :P

Thanks!