On Sat, Jan 13, 2007 at 12:54:24PM +0100, zutph3n@gmail.com wrote:
> Hi,
>
> I have a linux server running kernel 2.6.19 that is connected with 2
> seperate 100Mbit links to the same isp:
>
>
> +---+
> +---------------+ | I |
> +---------------+
> | | | S |
> | |
> | eth0 --+--------------+ P |
> | |
> | | | S |
> | |
> | linux 2.6.19 | | W |========================| ISP
> GATEWAY |
> | | | I |
> | |
> | eth1 --+--------------+ T |
> | |
> | | | C |
> | |
> +---------------+ | H |
> +---------------+
> +---+
>
> Both links have their own ip but have the same gateway. The problem is I
> can't seem to get egress traffic load balanced over the 2 nics.
>
> IP config after boot (dhcp from isp)
> ip a:
>
> 1: lo: <LOOPBACK,UP,10000> mtu 16436 qdisc noqueue
> link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> inet 127.0.0.1/8 brd 127.255.255.255 scope host lo
>
> 2: eth0: <BROADCAST,MULTICAST,NOTRAILERS,UP,10000> mtu 1500 qdisc
> pfifo_fast qlen 1000
> link/ether 00:00:00:00:00:0f brd ff:ff:ff:ff:ff:ff
> inet 10.0.0.110/24 brd 10.0.0.255 scope global eth0
>
> 3: eth1: <BROADCAST,MULTICAST,NOTRAILERS,UP,10000> mtu 1500 qdisc
> pfifo_fast qlen 1000
> link/ether 00:00:00:00:00:ed brd ff:ff:ff:ff:ff:ff
> inet 10.0.0.120/24 brd 10.0.0.255 scope global eth1
>
> Default routing table after boot
> ip r:
>
> 10.0.0.0/24 dev eth0 scope link
> 10.0.0.0/24 dev eth1 scope link metric 1
> 127.0.0.0/8 dev lo scope link
> default via 10.0.0.1 dev eth0
> default via 10.0.0.1 dev eth1 metric 1
>
> I enabled ip_forward and set arp_ignore to 1 for eth0 and eth1 to make
> sure the correct nic answers to arp requests.
>
> I tried to get the egress load balancing to work by replacing the above
> two default routes with:
>
> ip route add default mpath drr nexthop via 10.0.0.1 dev eth0 weight 1
> onlink nexthop via 10.0.0.1 dev eth1 weight 1 onlink
>
> I assumed that with mpath device round robin both nics would be used
> more or less equally, but the reality is only one of the nics actually
> works and the second nic even stops responding to arp requests.
>
> Am I doing something totally wrong or impossible here or is the device
> round robin code not working properly?
Curiosity but why use such a setup is your ISP link > 2Gbp/s ? Why not bond if
you want HA.
why its not round robining. I am going to guess but this line
default via 10.0.0.1 dev eth0
costs less to use than
default via 10.0.0.1 dev eth1 metric 1
so it should never use the second. I say guess cause I don't know what the
default metric is if you do add one.
What you want it to look something like is
default proto static metric 5
nexthop via 144.132.144.1 dev vlan2 weight 5
nexthop via 10.20.20.230 dev ppp0 weight 20
There is a link to a howto on the web site that steps out how to set this up
Alex
> _______________________________________________
> LARTC mailing list
> LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
>
signature.asc
Description: Digital signature
_______________________________________________
LARTC mailing list
LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
|