I just tried that, but it broke routing a bit.

I have two WAN connections. My main one is FTTC, giving me about 12 Mb/s, on eth2. The other is a backup over a complex wireless network on eth1 at 192.168.50.1. I also use this connection to administer the wireless network, and to do this I need to reach the various machines on the network, whose addresses are in the ranges 10.251.0.0/16 and 172.16.0.0/16.

To that end, I have create /etc/sysconfig/network-scripts/route-eth1, containing:

172.16.0.0/16 via 192.168.50.1

10.251.0.0/16 via 192.168.50.1

If I leave the system load balancing, with mumtiwan.conf containing:

MULTIPATH="on"

MULTIPATH_WEIGHTS="ppp0|200 eth1|1"

that works fine. But if I add the line:

EXTIF_BACKUP="eth1"

I can no longer connect to the 172.16 and 10.251 networks.

Is there anything I can do to keep the routing to those networks, but also force eth1 to be backup only?

i've made a similar change now and it seems to be working -
Still see a few warnings come through - but at least it's not timing out and restarting everything now

Maybe - but changing the last line works...

I edited the last line and used one of my other Internat IPs and www.apple.com (118.215.117.15) and this is now in my log

...

Fri Oct 22 15:34:16 2010  info:  config - offline interval - 10 seconds

Fri Oct 22 15:34:16 2010  info:  config - always use remote server for IP detection - disabled

Fri Oct 22 15:34:16 2010  info:  config - ping server auto-detect - disabled

Fri Oct 22 15:34:16 2010  info:  config - try pinging gateway - yes

Fri Oct 22 15:34:16 2010  info:  config - ping server - 118.215.117.15

Fri Oct 22 15:34:16 2010  info:  config - ping server - 144.136.112.190

...

...

Fri Oct 22 15:35:13 2010  info:  system - current WANs in use - eth0 eth3

Fri Oct 22 15:35:13 2010  info:  system - restarting firewall

Fri Oct 22 15:35:13 2010  info:  system - updating intrusion prevention whitelist

Fri Oct 22 15:35:13 2010  info:  system - adding ping server 144.136.112.190

Fri Oct 22 15:35:13 2010  info:  system - adding ping server 118.215.117.15

Fri Oct 22 15:35:13 2010  info:  system - adding DNS server 192.168.1.99

...

Edit - managed to get a failure to www.apple.com

...

Fri Oct 22 15:40:28 2010  info:  system - heartbeat...

Fri Oct 22 15:45:29 2010  info:  system - heartbeat...

Fri Oct 22 15:47:31 2010  info:    eth0 - ping check on server #1 failed - 118.215.117.15

Fri Oct 22 15:47:34 2010  info:    eth0 - ping check on server #2 passed - 144.136.112.190

Fri Oct 22 15:47:55 2010  info:    eth0 - ping check on server #1 passed - 118.215.117.15

Fri Oct 22 15:50:35 2010  info:  system - heartbeat...

...

syswatch log is where i found the evidence of the problems that were being experienced.
i've looked at /etc/syswatch, but found it to be totally standard - the IP addresses that syswatch is using as ping target is obviously from elsewhere in the system config - but where??

I figure i could add manual entry into /etc/syswatch and see what happens - hopefully that will override any other config.

I feel that this should be a user-configurable item... (perhaps in an advanced menu)
But i guess this would detract from the 'simple beauty' of ClearOS...

I am in Sydney and getting the odd ping failure - but not enough to cause a problem though...

Look at "/etc/syswatch" - note the warning at the top...