LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Balancing outgoing traffic

To: Alexandre CASSEN <alexandre.cassen@xxxxxxxxxxxxxx>
Subject: Re: Balancing outgoing traffic
Cc: <mack.joseph@xxxxxxx>, <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
From: Julian Anastasov <ja@xxxxxx>
Date: Wed, 26 Sep 2001 16:54:11 +0300 (EEST)
        Hello,

On Wed, 26 Sep 2001, Alexandre CASSEN wrote:

> >> As we said in different anteriors post, the only way to use VRRP on a
> >> director
> >> (LVS or other) owning multiple NICs is to run a VRRP instance per NICs
> >> since
> >> VRRP can be view as a routing protocol & synchronization between VRRP
> >> instances
> >> must be handled using part of VRRP protocol (synchronization using high
> >> level
> >> priority advert).
> >
> >    Yes, may be it is difficult to me to understand all details
> >of your work, may be because I'm busy with other patches :)
>
> Or probably my poor english :)

        No, the problem is that I have to look more deeply into VRRP
one day, I know your goals from previous postings :)

> Well in fact it can be quickly summarize as :
>
> I am concidering LVS-NAT director box with 2 NICs (=> LVS1, eth0 for WAN &
> eth1 for LAN). I want to setup a second LVS-NAT director (=> LVS2) to
> handle director failover. So I setup exactly the same hardware and software
> configuration on both LVS1 & LVS2. Then to handle director failover I want
> to use VRRP. So I create VRRP VIP on LVS1 where VIP is master on LVS1 eth0
> interface (symetricly VRRP VIP backup on LVS2 eth0 interface).
>
> The VIP is created (a VRRP VIP), then I use this VRRP VIP to configure my
> LVS ipvsadm rules on the both LVS1 & LVS2. So if LVS1 eth0 fails, LVS2 eth0
> takeover.
>
> The question is here : when takeover appear (LVS1 eth0 fails) we need to
> synchronize eth1 too. I was thinking on this synchronization : to handle a
> good synchronization, IMHO, we need to run another VRRP instance onto the
> eth1 interface. that way, when eth0 fails our software implementation tells
> to set VRRP instance running on eth1 into backup state. And so VRRP VIP
> will be master on the same director at a time (eth0 & eth1). So we will
> never have the case where VRRP VIP is master on LVS1 eth0 and backup on
> LVS1 eth1.

        Yes, may be there are many possible scenarios. May be the user
should specify different reactions to a specific failure - may be
something like a group of conditions to be primary or backup or to
change the state. IMO, the routing topology can play big role in
such decisions. I still can't summarize these things.

> => So postulat : We need to run a VRRP instance per physical interface to
> handle director failover using VRRP.

        I think, you will change this axiom very soon :) Of course, you
have to monitor all possible links between two hosts but more things will
play in the decision. For example, the state of some third link. May be
some higher layer checks could help to maintain the right state. This
is something like a big expression from conditions or at least I see it
in this way :) If we want to make it so complex :)

> This is the part of soft I am currently working on.
>
> Regards,
> Alexandre


Regards

--
Julian Anastasov <ja@xxxxxx>



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