teql packets disappearing

I'm trying to set up a load balanced link per the instructions on the
advanced routing howto and similarly this document
http://www.sangoma.com/linux/linux-teql.htm, which i think describes it a
bit better but is taken directly from the section from the routing
howto, but not having a lot of luck. I'm using a RHEL 4.0 machine with 3
ethernet interfaces as a test before i do it on my machine with S518 DSL
card and a S514 T1 card. The setup looks like the following:

|------|192.168.3.79-------192.168.3.94|-----|
10.0.0.2---10.0.0.1| |teql 10.2.0.0/31 | |---
|______|192.168.3.101-----192.168.3.126|_____|


I follow the instructions verbatim but after i have the teql0 device set
up as the default route, it looks like nothing ever goes to it.

initial setup:
[root@localhost sysconfig]# ip addr show
1: lo: <LOOPBACK,UP> 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
inet6 ::1/128 scope host
valid_lft forever preferred_lft forever
2: eth0: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
link/ether 00:90:27:8f:4b:28 brd ff:ff:ff:ff:ff:ff
inet 192.168.3.101/27 brd 192.168.3.127 scope global eth0
inet6 fe80::290:27ff:fe8f:4b28/64 scope link
valid_lft forever preferred_lft forever
3: eth1: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
link/ether 00:0a:01:d4:96:6d brd ff:ff:ff:ff:ff:ff
inet 192.168.3.79/27 brd 192.168.3.95 scope global eth1
inet6 fe80::20a:1ff:fed4:966d/64 scope link
valid_lft forever preferred_lft forever
4: eth2: <BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast qlen 1000
link/ether 00:50:bf:72:51:c9 brd ff:ff:ff:ff:ff:ff
inet 10.0.0.1/24 brd 10.0.0.255 scope global eth2
inet6 fe80::250:bfff:fe72:51c9/64 scope link
valid_lft forever preferred_lft forever
5: sit0: <NOARP> mtu 1480 qdisc noop
link/sit 0.0.0.0 brd 0.0.0.0

Then i load the sch_teql module (it loads fine)

now i have the additional following int:

6: teql0: <NOARP> mtu 1500 qdisc noop qlen 100
link/void


I set up the tc qdiscs for both eth0 and eth1:
[root@localhost sysconfig]# tc qdisc add dev eth0 root teql0
[root@localhost sysconfig]# tc qdisc add dev eth1 root teql0

and then set an arbitrary IP for the teql0 device, up the device and
delete the route for it:
[root@localhost sysconfig]# ip addr add dev teql0 10.2.0.0/31
[root@localhost sysconfig]# ip link set teql0 up
[root@localhost sysconfig]# ip route del 10.2.0.0/31

The next part of the instructions for adding a default route through the
teql0 device fails because i already have a default route, so i delete
my current default route and add the teql0 device as the default:
root@localhost sysconfig]# ip route show 192.168.3.96/27 dev eth0 proto
kernel scope link src 192.168.3.101 192.168.3.64/27 dev eth1 proto
kernel scope link src 192.168.3.79 10.0.0.0/24 dev eth2 proto kernel
scope link src 10.0.0.1 169.254.0.0/16 dev eth0 scope link
default via 192.168.3.126 dev eth0

[root@localhost sysconfig]# ip route del default
[root@localhost sysconfig]# ip route add default dev teql0
[root@localhost sysconfig]# ip route show
192.168.3.96/27 dev eth0 proto kernel scope link src 192.168.3.101
192.168.3.64/27 dev eth1 proto kernel scope link src 192.168.3.79
10.0.0.0/24 dev eth2 proto kernel scope link src 10.0.0.1
169.254.0.0/16 dev eth0 scope link
default dev teql0 scope link

so far so good, no errors, i disable the revers path filtering on both
int. eth0 and eth1 and from what i can tell, at this point i should be
good to go.

The problem now is that from the 10.0.0.x network, which uses 10.0.0.1
as the default gw, i can't get anywhere. I can ping the 10.0.0.1
interface as well as either of the eth interfaces on my teql machine,
but if i try to go past that the packets just seem to disappear. I never
see anything at the 192.168.3.94 or 192.168.3.126 addresses.... (ip
forwarding IS enabled and iptables is accepting everything)

It's somewhat odd because if i try to ping an ip address on the outside
world, i get destination host unreachable, but if i try to telnet to
port 80 on some webserver i see dns requests coming from the 10.0.0.2
machine on the 192.168.3.94 machine which would indicate that the packets
are finding it's way through the teql device. however, it fails with a
"no route to host" error message.

Any ideas?

Thanks in advance,

Aaron Martinez