[AusNOG] OSPF question

[Chris Kawchuk]

On 19 May 2017, at 1:48 pm, Alex Samad <alex at samad.com.au> wrote:

> Just realised I can also redistribute the prefixes as NSSA-ext1 or 2.

This results in a type-7 (inside NSSA) or a type-5 if it's propagated into 0.0.0.0; meaning it's a new table entry, new metrics (absolute vs calculated metric) and requires calculation in SFP tree. However, routing control planes are really beefy these days, so no issue there computatonal-wise. (Ive seen a million OSPF routes in inet.0 on a lowly M7i). it's useful if you want to force traffic to a certain router to access that subnet. (i.e. absolute metric); however If the network is single homed, you don't gain much.

It's easier to see in the OSPF database (i.e. "show ospf database"); as it'll show the originating router who's announcing that type7/ (type5 after propagation by the ABR) into the network.

It will also show up as an OSPF-external (i.e. protocol preference = 150) route in your main routing table; so not as preferred.

> Whats the difference , which is considered to be best practices, I have also setup interfaces with passive OSPF.

If you just add it into the area as a "passive OSPF" then it gets attached to the type-1 router advertisement. It results in no entries in the OSPF external DB. 
It'll will still show up in "show ospf routes" and naturally as a ospf-internal (protocol preference = 10).

I've done both; I like the first method only because it's easier for me to see which router advertised that subnet; ("show ospf database | match 10.10" ) although you can still look at the type-1 for a match too; but a bit of digging.

- Ck.





-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20170519/55815762/attachment.html>