LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

NFCT configuration/usage

To: Julian Anastasov <ja@xxxxxx>
Subject: NFCT configuration/usage
Cc: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
From: ken <lists@xxxxxxxxxxxx>
Date: Mon, 15 Dec 2003 12:53:58 -0500
Quoting Julian Anastasov <ja@xxxxxx>:
> 
> > (1)  Would the Antefacto patch allow a director to do destination NAT
> (DNAT)
> > onto the same network for a VIP?
> 
>       Yes but it is not related to this feature

I've successfully patched and rebuilt Horms' RedHat 9 Ultramonkey kernel 
(kernel-2.4.20-24.9.um.1.src.rpm) with the 2.4.23 NFCT patch.  I've also 
enabled connection tracking and SNAT reroute runtime support.  I'm trying to 
figure out how to get DNAT to work.  In a simple NAT LVS setup with 2 
realservers, the realservers (running both Apache and a JAVA app server) must 
occationally make a HTTP request to a VIP living on the director.

I currently have an IPTABLES rule that makes each server view the VIP as local, 
but if the Apache server goes down, I have no failover.  How can I have 
requests sent to the director's VIP redirected back into the cluster and the 
same network?

Assuming this would require a DNAT-type setup, I've added the following 
IPTABLES rule on the the director, with no success.  192.168.0.40 is the VIP.  
The realservers live on 10.0.5.0/24.

-A POSTROUTING -s 10.0.5.0/255.255.255.0 -d 192.168.0.40 -p tcp -m tcp --dport 
80 -j SNAT --to-source 10.0.5.1

Regards,
Ken




<Prev in Thread] Current Thread [Next in Thread>
  • NFCT configuration/usage, ken <=