OSPF on SVI's rerouting issue

[dvag-nsafe]

Hi!

We currently have a setup where OSPF is condfigured on the SVI's. My Question is simple: OSPF will not notice a failure of the physical links until the dead timer has run out, right? If you configure OSPF on the physical links, OSPF will reroute immidietely, correct?

Our problem is, that we had a link flapping three times for 5-10 seconds. The connection was not rerouted. My guess was that OSPF did not realize the link was down and therefor all traffic went in the black hole...

Let me know if im totally off here...

Christian

[Mark Malone · Accepted solution]

Ye nothing happens until the dead timer finishes , OSPF timers are per link so why not increase the convergence speed by changing the hello timers for those particular links only as a test and drop the svi see if it makes a difference for you or use bfd you could get it down to real quick convergence , just with OPSF though you need to be careful and make sure as its a LSDB setup that you don't cause issues with convergence in general in the topology as it all relies on each other but you can definitely tweak OSPF down from the defaults , the more stable your setup is the lower you could probably drop it but I would test it bit by bit

[dvag-nsafe]

Thanks Mark!

I was pretty sure but thanks for the confirmation. I already thought about BFD, since it works fine with OSPF and is really quick. I would not like the timers of OSPF tweaked, cause it will still be much slower than BFD.

What about Interface SLA tracking? Is there a way to track the physical port and have it take down multiple SVI's when a state change is recognized?

[Mark Malone]

What about Interface SLA tracking? Is there a way to track the physical port and have it take down multiple SVI's when a state change is recognized?

The way I would do that is through an EEM script tracking ip sla inside it , when the physical interface drops the script could trigger and then you could have it shut down all  your svis you need and bring them backup when the interface comes backup

it would look something like this in the post below but you could tweak it , as well if you don't want to track the physical interface you can track the syslog state change as you said instead in the EEM if that's a better option for you , you want the script though to automate the svis back up if the state change fixes itself , also would put a delay of a few seconds at least on it , you don't want to be triggering convergence changes instantly everytime or you could end up with interfaces going up and down constantly

https://supportforums.cisco.com/discussion/10794236/shut-interface-if-no-ping-response-using-ip-sla-eem

BFD would be my choice too is your kit supports it , better and ties in well with OSPF