[lvs-users] localnode question
dean.scothern at eduserv.org.uk
Wed Nov 30 09:41:39 GMT 2011
On Wrote, 30 Nov 2011, Joseph Mack wrote:
>On Tue, 29 Nov 2011, Dean Scothern wrote:
> > Hi,
> > Is there a way to either disable or bypass the localnode behaviour in ipvs?
> other than not using it?
What alternatives are there?
> > I'm trying to build a simple 2 node mail load balanced cluster.
> > I have a mail server on each node and load balance between the two via a
> > I'm not allowed to use the lvs-dr method as our network guys say that
> asynchronous routing is forbidden by our firewalls.
> > I can use lvs-nat, with a separate 'back-end' network between the two
> I don't understand your lvs. You have the director in localnoce listening on
> the VIP and you have a 2nd box which is listening on the RIP via LVS-NAT?
Apologies for not explaining the configuration clearly.
I have a two box solution in which both nodes are real servers.
The boxes are connected via their primary interfaces.
Exim(mail) is running on both. Pacemaker is used for HA and can migrate the two exims in the usual way.
I want to load balance between the two nodes and as I'm not allowed to use lvs-dr I've tried to use lvs-nat.
To this end I've added a secondary network and again used pacemaker to manage the secondary VIPS (and ldirectord).
= = = lvs nat =
= exim = = exim =
VIP1s,RIP VIP2s, RIP
In the normal situation, with exim running on both boxes load balancing works fine.
On the local node, exim can listen/reply on the primary vip (localnode behaviour) and on
the remote node exim can listen/reply on the secondary vip.
However in failover situations where both primary vips are on the same machine,
load balancing does not work, the original local exim can can listen/reply on the
primary vip (localnode behaviour) but the migrated remote node exim cannot receive, as traffic always goes
to the original local primary vip.
I can make this scenario work If I can disable or bypass the localnode behaviour.
Suggestions/work arounds would be welcome
> > However the localnode behaviour forces the mail server
> > (exim) to listen on the vip. When the second node's mail server is
> > migrated to the first node (pacemaker), it cannot be accessed as all
> > connections go to the vip and hence the first mail server.
> > It appears that there was some experimentation to address this some
> > years ago, but I'm not sure if there have been any recent
> > developments.
> I thought there might have been something like this by Ted Pavlic in the early
> days but I can't find it.
> I expect the simplest thing to do is to get a 2nd realserver
I really don't want to add a separate load balancer cluster for the sake of two machines.
> Joseph Mack NA3T EME(B,D), FM05lw North Carolina jmack (at) wm7d (dot)
> net - azimuthal equidistant map generator at
> http://www.wm7d.net/azproj.shtml Homepage http://www.austintek.com/
> It's GNU/Linux!
> Please read the documentation before posting - it's available at:
> LinuxVirtualServer.org mailing list - lvs-users at LinuxVirtualServer.org Send
> requests to lvs-users-request at LinuxVirtualServer.org
> or go to http://lists.graemef.net/mailman/listinfo/lvs-users
Dr Dean Scothern
E: dean.scothern at eduserv.org.uk T: +44 (0)1225 474379 F: +44 (0)1225 474301 www.eduserv.org.uk
Eduserv is a company limited by guarantee (registered in England & Wales, company number: 3763109) and a charity (charity number 1079456), whose registered office is at Royal Mead, Railway Place, Bath, BA1 1SR.
More information about the lvs-users