OSPF LSA Expiry..!

[vramanaiah]

While i understand that OSPF LSA is flushed by the originator always, i don't understand how all the OSPF routers in a given area come to know about failure of a router (going down abruptly).. Do all the routers in the domain have to wait under MaxAge to flush out all of its LSAs..?

Can anyone advise?

[paul driver]

Hello

"Do all the routers in the domain have to wait under MaxAge to flush out all of its LSAs..?"

A change in the topology equals a change to the link state database and as you know this database need to be identical with all ospf domain.

The change initates the flooding of new LSA's ( which are then set  by ospf to the max age) to neighboring routers which in turn does the same to its ospf  neighbors.

Single or multiple LSA's can be sent via LSU's ( link state updates) wihich include just the headers of the LSA not the whole LSA packet and each LSA  needs to be acknowledge.

For the routers sending the LSA 's this validates thay their LSA's  have been received and for the routers receiving these LSA's this validates that they have the most recent uptodate copy.- hense why the flooding througout the ospf domain.

Hope this helps-?

res

Paul

Please don't forget to rate any posts that have been helpful.

Thanks.

[vramanaiah]

Hi Paul,

Thanks for your reply, however my question is not about expiring the links..

Let us say the topology is like this

R1 ------- R2 ------- R3

Let us say all three routers are in Sync.. Now if R1 goes down, how does R3 know that R1 has gone down and hence flush all the LSAs associated with it..   Should R3 wait until MaxAge to expire them all or is there any other mechanism through which it would come to know about R1's failure so it can flush its LSAs.

Regards

[Grant Curell]

Hey vramanaiah

So OSPF detects a change in one of two different ways:

1) A failure or change of status on a physical interface. In your example R2 would note the failure on its left interface and would realize that a topology change had occurred. Router 2 will send an LSA to R3 containing the topology change in the network and both R2 and R3 will run the SPF accordingly to update their routes.

2) The router fails to receive hellos for the dead timer. By default on a broadcast network the hello timer is 10 seconds and 30 seconds for a nonbroadcast network. After 4x the hello time the dead timer will pop and the router will realize there has been a topology change and generate LSAs accordingly.

The routers will not run the SPF algorithm until after the SPF timer expires (default time is 5 seconds). This is a hidden timer not often mentioned. Therefore the minimum time for convergence in OSPF is 5 seconds. Additionally, hello packets contain a list of all routers that the advertising router has heard from in less than the dead time or one that it knows to have failed. Therefore, R3 will know immediately upon receipt of the LSA from R2 that it can flush R1's LSAs

Hope that helped!

Grant

[paul driver]

Hello

"Now if R1 goes down, how does R3 know that R1 has gone down and hence flush all the LSAs associated with it.. "

As previosly stated - The change in the toplogy initates the flooding of new LSA's ( which are then set  by ospf to the max age) to neighboring routers which in turn does the same to its ospf  neighbors.

res

Paul

Please don't forget to rate any posts that have been helpful.

Thanks.

[vramanaiah]

I am not sure if i have conveyed my question right..!

My question is how will R3 come to know of R1's failure.. As per OSPF rules (as per my understanding), only the originator of the LSA can flush the LSAs. In this case, if R1 goes down abruptly how does R3 come to know about it..?

R2 would only intimate the change in topology with regards to the links connected to it. Correct me if i am wrong, It is not going to announce about R1's failure..

[Grant Curell · Accepted solution]

It sounds like you're asking two separate questions:

1) How will R3 identify R1's failure?

View my previous answer. Yes, R2 will announce R1's failure. Recall that OSPF views everything as a link or a node. R2's type 1 (router) LSA will not include R1 as an attached node. When R3 receives R2's new LSA it will overwrite the old one and rerun the SPF algorithm based on the new information. Since this new information does not include R1 as an attached node R3 will not select any paths utilizing R2->R1 when it runs the SPF algorithm (because R2 told R3 that the connection between itself and R1 no longer exists)

2) When will R3 flush R1's LSA?

I believe it will just have to wait until the max age timer hits. As far as I know you are correct in that OSPF doesn't allow for flushing unless it is self originated. However, if R1 is completely down then no one else will list a connection to it in their LSAs so the SPF algorithm will never select a route attempting to use R1. Point being, other then R1's LSA's negligable memory consumption it doesn't really matter that it's there.

[vramanaiah]

Hello Grant,

Thanks for sharing your views.. I tested your hypothesis in the lab and found your explanation is valid.

As you rightly pointed out the Routes of R1 vanish instantly (after R1 failure is detected by R2), however the LSAs of R1 remain in the database of R2 and R3. They will probably be flushed after 60 min (i.,e when MAXAGE is reached).

Regards