Re: [lvs-users] Problem: Occasionally all traffic gest sent to a single

To: " users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: [lvs-users] Problem: Occasionally all traffic gest sent to a single backend node
From: "kotobuki intl" <kbintl2@xxxxxxxxx>
Date: Fri, 18 Jan 2008 20:15:05 +0900
Yeah your right, in fact I thought we had already addressed the ARP
issue in our init.d scripts, however upon looking at them I see they
were echoing the parameters to the location of the interfaces under a
2.4 kernel, as opposed to 2.6.  I can't believe I missed that before

On 1/18/08, Graeme Fowler <graeme@xxxxxxxxxxx> wrote:
> On Fri, 2008-01-18 at 18:35 +0900, kotobuki intl wrote:
> > Anyhow does anyone else have this problem and ideally a fix?
> Assuming that your monitoring graphs are being taken from the
> realservers themselves and not the director(s), I'd say that you have
> the ARP problem.
> This means that one of your realservers has managed to tell the upstream
> router that it's got the VIP. The router then bypasses the director(s)
> and sends all traffic for a given VIP to the realserver directly. Your
> clients see no difference - connections still work - but all traffic
> being handled directly rather than via LVS.
> Have you ensured that all of your loopback aliases on the realservers
> are hidden?
> You'll need to amend the instructions there accordingly.
> Graeme
> _______________________________________________
> mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
> Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
> or go to

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