nick at inex
Apr 15, 2012, 8:31 AM
Post #2 of 2
On 15/04/2012 06:07, Naser Al Hattab wrote:
> When I set this first, all is OK, and both servers announce the external
> routes (VIPs) out with a "Forwarding address" of *Router ID* However, while
> by testing the failover of the interfaces per server by shutting down one
> interface and keeping the other interface one is UP,the external OSPF route
> is advertised by the dns servers with forwarding address of the physical
> interface ,since the interface is down the next hop IP is not valid and
> hence all the traffic destined to this VIP will be blackhole.By restarting
> NIC interfaces“service network restart” the scenario becomes normal and the
> forward address getting the router ID.
> Can you assist us why this scenario happed on quagga and not in zebra.
Naser, I'm guessing that you have two DNS servers, each with two uplink
interfaces. Can you post your configuration (topology and quagga config)
because it's not really clear what's happening here?
This sort of config is pretty standard for quagga, and it works well. All
you need is two (or more) DNS servers, with the same virtual IP attached to
the loopback interface. There's no real requirement for multiple uplinks
from the DNS servers. Also, you would want to keep the ospf timers very
low (e.g. hello/dead == 1:4 or something) so that failure is detected quickly.
Quagga-users mailing list
Quagga-users [at] lists