Would like your assistance please regarding NSF/NSR .. As I understand NSF/NSR is useful only during switchover from active RP to backup one
Now my question is
1- If router crashed and restarted or even I did normal reload, will NSF/NSR be useful or not as in this case both RPs will reload ?
2- If router really goes down or line is down due to telco problem how will neighbor router knows this is not RP switchover ; cause as I understand in this case neighbor will wait for a grace period then will declare neighbor is dead thus slow convergence
One of the basic assumptions of NSF is that the network is stable and that the router is still capable of forwarding IP traffic. One of the requirements to support this is that interfaces must remain up and neighbors must remain accessible. if you reload the router then these requirements are not satisfied and NSF is not operable.
Similarly if the line is down due to telco problem then NSF is not a factor since the topology of the network has changed and the existing forwarding table is no longer valid.
1) as already noted NSF applies to RP switchover, if both RPs are reloaded this is outside the domain of NSF there is no forwarding table to be kept, and the reload will make all linecards to reload and reboot and will make all links to go down,
In NSF RP switchover linecards are not restarted the new master RP takes control of the whole chassis on the fly.
2) when an RP switchover is started the restarting system advertises the neighbor with a special message (protocol dependent, but with a meaning of starting of RP switchover) indicating it is starting an RP switchover, If the line fails no message at all is heard by neighbor and standard IGP / BGP timer applies (dead interval for OSPF) and when hold timer expires the adjacency is turned down.
Hope to help
Thats make much sense now .. Actually I was confused cause of below document
OSPF Graceful Restart Functionality per RFC 3623
RFC 3623 graceful restart nonstop forwarding enhancement to the OSPF routing protocol. An OSPF NSF-capable router that is reloading and attempting a graceful restart originates grace-lsas to notify its neighbors that it will perform graceful restart within the specified amount of time or grace period. During this grace period, the neighboring OSPF routers, called helper routers, continue to announce the restarting router in their LSAs as if it were fully adjacent, as long as the network topology remains static
The original question seemed to be asking about generic behavior of NSF and that is the context in which Giuseppe and I answered it. Now you have clarified that you are really asking about an enhancement to NSF specific to OSPF. So now we have a different context for the question. While I am not particularly familiar with this enhancement it seems clear to me what is going on. In normal operation when a router reboots its OSPF neighbors will see the connecting link go down. They will withdraw all LSA learned from the neighbor and will run SPF to rebuild the routing table. Then when the reboot is completed they will re-learn the neighbor, receive the LSAs and run SPF again to rebuild the routing table (and will wind up with exactly the same content). So this enhancement is based on the assumption that after the reboot the topology and the LSA database, and the routing table will be just as they were. so this enhancement allows OSPF to tell its neighbors that it is going down for a brief time and that when it comes back the topology and the LSA data base should be the same. So the OSPF neighbors establish a grace period. During this grace period the neighbors can not communicate with the rebooting router but they continue to act as if the router were functional and they do not withdraw advertisements or run SPF. It makes the network more stable and reduces overhaed processing in OSPF.
Thanks for your reply
In this case .. I believe it would be useful only in case NSF helper router does not have a backup path as during grace period there will be no packet forwarding
Also there is a possiblity after NSF helper router waits for grace period it then recomputes its SPF again ... long time .. slow convergence .. Believe this enhancement is optimium when no backup path exists
What do you think ?