[Pacemaker] SMTP Sending via Pacemaker VIP

[Dejan Muhamedagic]

Hi,

On Mon, Mar 07, 2011 at 02:21:10PM +0100, DooMRunneR wrote:
> Hello,
>
> i test pacemaker and drbd for a HA Mail server solution with a primary  
> and a backup server.
>
> Now we have the following problem.
>
> Our primary server uses ip x.x.x.10, the backup x.x.x.20 and the virtual  
> IP which the services are available uses x.x.x.100
>
> At the moment the primary server is the active mailserver-node. When a  
> user sends an e-mail via the primary node the internal ip x.x.x.10 is  
> used for sending, not the x.x.x.100.

This sounds like an application issue (the mail server in this
case). Which interface it uses for tcp/ip is beyond cluster. I
can recall this issue coming up often on multi-homed hosts.

Thanks,

Dejan

> This collides with our Firewall, only the 100 IP is allowed to send  
> mails out of our mailserver network. Our Mail-Solution does not have a  
> logic implemented for detecting which interface to use for an outgoing  
> connection or any connection for that matter. The system's kernel  
> determines this by analyzing the IP headers and based on the destination  
> address of the packet it routes the traffic according to the routing 
> table.
>
> I can't find the Pacemaker VIP in any kernel related routing tables, is  
> it possible to get those entries via pacemaker in my redhat distribution  
> so that my mailserver knows the he also uses the ip 100?
>
> kind regards
>
> anton
>
>

> _______________________________________________
> Pacemaker mailing list: Pacemaker at oss.clusterlabs.org
> http://oss.clusterlabs.org/mailman/listinfo/pacemaker
> 
> Project Home: http://www.clusterlabs.org
> Getting started: http://www.clusterlabs.org/doc/Cluster_from_Scratch.pdf
> Bugs: http://developerbugs.linux-foundation.org/enter_bug.cgi?product=Pacemaker