Re: [lvs-users] using fwmark with kernel 2.6/ IP ERR!

To: " users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: [lvs-users] using fwmark with kernel 2.6/ IP ERR!
From: Thomas Pedoussaut <thomas@xxxxxxxxxxxxxx>
Date: Wed, 23 Jul 2008 22:34:17 +0100
Perry Casson wrote:
> Hi,
> I've got two LVS-DR clusters using FWM running Centos 5.1
> (2.6.18-53.1.6.el5) for LB's and real servers.  Every few days I'm getting
> reports of http timeouts when going through the LVS but if the real servers
> are accessed directly the system works fine.
> Ipvsadm -Lnc reports several IP connections in ERR! state, I sniffed around
> the archives and found several threads on this the most recent leaves me
> hanging with the following question but no replies:
I'm using exactly the same configuration (LVS-DR, FWM, CentOS 5.1), an I
also have those ERR entries, only for the services using FWM, none for
services on traditional ip/port.

>From their characteristics, they are only 60 sec ttl, and seems to be
lacking the source port, which makes nearly impossible to track what
triggers it.

But at the same time I don't have any issues in connections, so I never
really bothered to dig further.
I was just assuming they were dummy entries for persistence stuff.

Sorry to no being able to help more.

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