LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

RE: SYN_RECV LVS_NAT on 2.4.7 kernel

To: Jeremy Kusnetz <JKusnetz@xxxxxxxx>
Subject: RE: SYN_RECV LVS_NAT on 2.4.7 kernel
Cc: "'mack.joseph@xxxxxxx'" <mack.joseph@xxxxxxx>, "'lvs-users@xxxxxxxxxxxxxxxxxxxxxx'" <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
From: Julian Anastasov <ja@xxxxxx>
Date: Mon, 10 Sep 2001 19:48:25 +0300 (EEST)
        Hello,

On Mon, 10 Sep 2001, Jeremy Kusnetz wrote:

> This is my ipvsadm setup:
>
> ipvsadm -C
> ipvsadm -A -t 216.163.XXX.13:telnet -s rr
> ipvsadm -a -t 216.163.XXX.13:telnet -R 10.75.0.2:telnet -m

        Time to switch from -R to -r, it is changed in the next
ipvsadm versions. Before checking anything else, one stupid question
from me: is ipvsadm -Ln showing the right setup you entered? I.e.
are you using the right ipvsadm version, from the ipvs tarball and
not from any package that is not matching (probably) to the LVS kernel
part?

> Running tcpdump on the director on all the interfaces I do ***not see any
> traffic what so ever*** trying to go to the realserver.  It's like port
> forwarding isn't on.  But:
>
> # cat /proc/sys/net/ipv4/ip_forward
> 1

        If you really have now a test setup, can you rebuild the kernel
without nat and connection tracking support (netfilter is still required).
May be if you put everything on modules it will be more easy to play with
different setups. [I didn't saw such simple setup not to work, it seems
LVS is not working at all]

> So it appears to be ON to me.  Since packets don't appear to be making it to
> the realserver, I think there is a problem with the director, not the
> realserver.

        Hm, can you compile and run the LVS debugging and to show one
session from your kernel log.

Regards

--
Julian Anastasov <ja@xxxxxx>



<Prev in Thread] Current Thread [Next in Thread>