John Reuning wrote:
>
> Actually, no, I was wrong. I think I misinterpreted the documentation
> when I reinspected after my original message. The DIP is the internal
> ip address that the real servers see, right?
yes
> So, the diagram should be:
>
> external --------------- internal
> | |
> 152.2.210.38 | | 192.168.210.38
> ---| |---
> | |
> 152.2.210.81 | | 192.168.210.81
> (VIP) | | (DIP)
> ---------------
yes
> The top two addresses are just the primary ip addresses of the
> director's internal and external interfaces. They shouldn't play a part
> in the lvs configuration. The real servers have the DIP as their
> default route, and the http requests all come to the VIP.
yes
> The problem occurs when http replies from the director are sent from the
> primary external ip address 152.2.210.38 instead of from the VIP.
this is not supposed to happen.
Julian,
2 people now have reported in LVS-NAT that the director replies
from the primary IP and not the VIP. Is this something you know about?
> In this case, should the lvs-start script remove the primary ip address
> from the external interface (152.2.210.38) after the secondary (VIP)
> addresses are added?
it shouldn't matter. All IPs should be independant
Joe
--
Joseph Mack PhD, Senior Systems Engineer, SAIC contractor
to the National Environmental Supercomputer Center,
ph# 919-541-0007, RTP, NC, USA. mailto:mack.joseph@xxxxxxx
|