[c-nsp] Dialer vs. dinamic routing

[Rodney Dunn]

When the OSPF learned route comes in over the primary the
static should come out and no redistribution should happen.

Even if you are running OSPF over the dial circuit once
it comes up if you learn the same prefix over the dialer
it should override the static also.

The floating is just to trigger the dial to start if you have
OSPF marked as noninteresting taffic.

You may need to look at other forms of dial such as dialer watch,
backup interface, IP SLA, etc...


Rodney


On Mon, Jan 23, 2006 at 04:47:14PM +0100, Piestaga wrote:
> Hi,
> 
> I would like to describe the problem I have and I would like to ask if
> that is the regular behaviour or is it a type of bug ?
> 
> There is a CE router that is connectied to PE using primary link with
> OSPF.
> There is also a floating static route that points the dialer to get
> the destination (eg.):  ip route 10.10.10.0 255.255.255.0 Dialer1 220
> when primary is down (220 is a distance metric higher that dafault
> ospf metric)
> Primary link is connected to PE-A whereas backup interface dials-in to
> PE-B
> 
> When primary link goes down, the backup session is established and
> there is no problem with that.
> Problem is when the primary links is up again.
> Because of the OSPF on CE is configured with "redistribute static",
> it also redistributes the floating static route build by dialer
> interface.
> 
> For me the dialer should terminate the connection when the 10.10.10.0
> is acccessible by primary link. But it does not happen. It looks like
> the CE sends via OSPF the update about 10.10.10.0 available through dialer
> interface before the PE-A sends to CE the update about the same
> network lerned from MBGP (MPLS in backbone).
> This causes that PE-A trying to send the whole traffic to 10.10.10.0 using
> dialer what causes the dialer up and running all the time (
> OSPF metric is more preferable than MBGP)
> 
> The temporary solution we use is to assigen a tag to the dialer
> floating static route and deny the redistribution for that tag on
> OSPF toward PE-A.
> 
> My question is whether this is a regular way it works, or I should
> watch for bug linked that case.
> 
> Thanks for any help or suggestions
> Sebastian
> 
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/