LARTC - Jun 2002 - I: troubles with RIPd

> Hi,
> 
> I''m trying to test RIP but I have some problems:
> 
> 1. 	I connect two linux boxes running zebra and ripd, with a cross
> ethernet cable.
> 
>    	 looking at the sniffer (ethereal) all seems to go well, because I
> can see both boxes sending the Ripv2 response message in   	broadcast
> and receiving them. The fact is that on both the ripd.conf file I put the
> command "redistribute kernel" but only one 	box updates its
routing
> table. Why the other box doesn''t?
> 
> 2.	Substituting one of the linux boxes with a cisco2600 router, things
> get worst. The cisco router send a ripv2 request packet 	and then
> continues to send CDP packets without redistributing its static routes.
> Does anybady know a simple configuration  	for the cisco2600 rip? On
> cisco web siteI found all but this.
> 
> my zebra.conf:
> 
> hostname Router-A
> password zebra
> enable password zebra
> interface lo
> description Router-A loopback
> ip route 0.0.0.0/0 192.168.100.1
> log file zebra.log
> 
> my ripd.conf
> 
> hostname ripd-r1
> password zebra
> router rip
> network 192.168.100.0/24
> redistribute kernel
> log stdout
> 
> my cisco configuration
> 
> router rip
> version 2
> network 192.168.100.0
> redistribute static
> 
> Many thanks for help,
> 
> Luca Andreani

Hi,

I am testing IMQ from few days. I have quite complex setup managing two
external interfaces with many classes and over 350 users. IMQ works
reliable and stable :)

I tried to experiment and to direct flow to IMQ I used rule in mangle
table but in FORWARD chain like following:

$IPTABLES -t mangle -I FORWARD 1 -i $ext_if1 -o $internal_iface -j IMQ
--todev 0

It works fine.

I also tried to direct local destined traffic via following rule to have
opportunity to run some local services and not disturb shaping policy..

$IPTABLES -t mangle -A INPUT -i $external_iface -j IMQ --todev 0

I couldnt use PREROUTING to achieve this result due to my box is doing
NAT and I have some classes/filters related to local destination
IPs.


It looks that second [INPUT] chain have some impact on traffic shaping but
it was very strange. Then I put that rule my interactive traffic becomes
very poor [NAT box didnt generate any traffic itself]. Normal situation
RTT was about 60ms but when launched rule in "-t mangle INPUT" average
RTT
jumped to 500 or even 1000 !

OK, thats all my report today :)
Going to play more with QL IMQ driver.

Regards
tw
--