Hi all,

after some disastrous test with OBSD 4.5, i tried to upgrade to -current
 via cvs, then to download the latest snapshot of 4.6 release from the ftp.

My goal is to obtain a couple of firewall in active/active mode with
load balancing enabled, but at now I'm unable to make it works.

As I already verified in 4.5 (see my thread
http://marc.info/?l=openbsd-misc&m=124473589921871&w=2), my
configuration works only as failover.


Here is my new configuration:

---------------------------------------

### Host A: ###
   /etc/hostname.em0
   inet 192.168.10.2 255.255.255.0 NONE

   /etc/hostname.em1
   inet 10.20.30.2 255.255.255.0 NONE

   /etc/hostname.carp0
   inet 192.168.10.1 255.255.255.0 192.168.10.255 carpnodes 10:0,20:100
balancing ip

   /etc/hostname.carp1
   inet 10.20.30.1 255.255.255.0 10.20.30.255 carpnodes 30:0,40:100
balancing ip

   /etc/hostname.bge1
   inet 172.16.188.1 255.255.255.0 NONE

   /etc/hostname.pfsync0
   up defer syncdev bge1


### Host B: ###
   /etc/hostname.em0
   inet 192.168.10.3 255.255.255.0 NONE

   /etc/hostname.em1
   inet 10.20.30.3 255.255.255.0 NONE

   /etc/hostname.carp0
   inet 192.168.10.1 255.255.255.0 192.168.10.255 carpnodes 10:100,20:0
balancing ip

   /etc/hostname.carp1
   inet 10.20.30.1 255.255.255.0 10.20.30.255 carpnodes 30:100,40:0
balancing ip

   /etc/hostname.bge1
   inet 172.16.188.2 255.255.255.0 NONE

   /etc/hostname.pfsync0
   up defer syncdev bge1


On both machines:

OpenBSD hostX.domain.tld 4.6 GENERIC#50 i386
net.inet.ip.forwarding=1
net.inet.carp.allow=1
net.inet.carp.preempt=1
---------------------------------------

Carp works, pfsync works, but when I send packets to the carp address,
everytime the same machine reply (so there is not a load balancing). I
tried to overload the carp interface, generating a lot of traffic from
several machines, but nothing happens: replies awlays came from the same
machine.

I tried also the ip-stealth mode, but with same results.

I tried to put two web servers behind the firewall (ip 10.20.30.4 and
10.20.30.5), running pf with these rules:


    HOST A: rdr pass on em0 proto tcp from any to (carp0) port 80 ->
10.20.30.4

    HOST B: rdr pass on em0 proto tcp from any to (carp0) port 80 ->
10.20.30.5


but replies came always from the same webserver, that is loadbalancing
not working.

So, I'm asking again myself if I'm missing something or where is the
mistake with my configuration.

Every help will be appreciated.

Reply via email to