Re: [lvs-users] Source Hashing (Again)

To: <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: [lvs-users] Source Hashing (Again)
From: <Darren.Mansell@xxxxxxxxxxxx>
Date: Wed, 29 Sep 2010 14:31:27 +0100
-----Original Message-----
From: lvs-users-bounces@xxxxxxxxxxxxxxxxxxxxxx
[mailto:lvs-users-bounces@xxxxxxxxxxxxxxxxxxxxxx] On Behalf Of Malcolm
Sent: 29 September 2010 13:32
To: users mailing list.
Subject: Re: [lvs-users] Source Hashing (Again)

On 29 September 2010 11:36,  <Darren.Mansell@xxxxxxxxxxxx> wrote:
> Hello everyone.
> We have a 2 node load-balanced MySQL cluster running in active-active 
> mode. Linux-HA is running on each with LVS balancing connections on 
> the nodes themselves.
> So for this we use source hashing investigating after looking through 
> the archives of this list to see that as long as you give the 
> weighting a fairly high number it should be able to balance 
> connections when one node is unavailable and the weighting is set to 
> 0. Unfortunately this doesn't seem to happen. These are my current


Why can't you just use Source IP persistence with WLC?


Malcolm Turnbull.

Hi Malcom.

Thanks for the reply (and sorry for the Outlook-quoting).

I'd discounted persistence as I understood it to be only for making each
connection persistent, but thinking about it again that doesn't make a
lot of sense!

Am I right in thinking that if a single client makes its first
connection to node1 then as long as there are connections made before
the persistence time-out they will be made to the same real server
again? But then if the weight changes to 0 on node1 new connections will
be made to node2 regardless of the persistence time-out? (as should
happen on SH but doesn't).


Please read the documentation before posting - it's available at: mailing list - lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Send requests to lvs-users-request@xxxxxxxxxxxxxxxxxxxxxx
or go to

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