Some OSPF clarifications under BMA environment.

[eldeezy29]

Hi.

First let me say i tried reading Cisco & non-Cisco info, Googled and read tutorials about OSPF before I sent my question here. My native language isn't English so sometimes its hard for me to understand English reading metirials.

The scenario is like this:

5 Routers Connected to a LAN BackBone, Running OSPF under a single AS. A DR/BDR are already elected and the 6'th Router that will join - won't be a DR/BDR (no new election).

Router A -> DR.

Router B -> BDR.

Router C,D,E -> DROTHER.

I'm about to connect the 6'th OSPF router (Router F) to the segment for the first time. Now to my questions:

1. All SPFRouters sends Hello packets periodically, to who? To all SPFRouters (224.0.0.5) or just to the DR/BDR (224.0.0.6)?

2. Router F just joined the segment. it starting to send an Hello packets to all SPF Routers(224.0.0.5). Who replies to him? The DR or Any OSFP router in the segment (NOT a DR necessarily)?

3. "Routers become neighbors as soon as they see themselfs listed in the neighbor's Hello packet". Who adds Router F to the neighburs list, The DR or any SPF Router on the segment?

4. Does Router F need to see himself listed only under one Hello packet or is it waiting to be known to all neighbors Hello Packets?

5. Who tells to who when Router F will go down? and then what will happen?

I know i need some order and simple/clear info about the steps and procedure OSPF is working.

Thanks a lot,

Eldad.

[Richard Burts]

Eldad

We will try to answer your questions and hope that English will not present too much of a problem.

1) all OSPF routers send hello packets to the 224.0.0.5 address. Hello packets perform several functions including that they allow routers to recognize who is on the segment, they are used to form neighbor relationships, and they act as keepalives so that routers can recognize when a router is no longer active on that segment.

2) router F joins the segment and sends hello. It will receive hello from each of the other routers on the segment (both DR, BDR, and DR/other).

3) Each router sends hello which reflects its own view of the network. Within the hello packets are several fields. For this discussion it is important to recognize that in the hello is a field in which a router lists every neighbor that it knows about, a field in which the router lists the DR (if one has been elected) and a field in which the router lists the BDR (if one has been elected). When router F joins the segment and sends a hello each of these fields will be blank because router F does not yet know about any neighbors or the DR or the BDR. Other routers will send hello messages in which they list the existing neighbors (but not yet router F) and list the DR and BDR. When router F receives them it knows that there is an existing DR and existing BDR. As other routers have received hello from router F they will list router F in their next hello.

Each router is responsible for maintaining its own neighbor list.

4) router F will need to see itself listed in the hello from all of the neighbors.

5) each neighbor keeps track of its neighbors. An OSPF router has a dead timer which defines the amount of time a router will wait without hearing hello from one of its neighbors before it decides that the neighbor has failed. So each router is independently evaluating the state of each of its neighbors. (note that there is a field for timers including dead timer in the hello packet and all routers on the segment must have the same value of dead timer).

If hello messages from a neighbor stop the router will use the dead timer to discover that this neighbor is no longer valid. When the dead timer expries the router marks the neighbor as down, and it purges from its topology table any information learned from that neighbor. It will then re-run the SPF algorithm to determine if any of its routes have become invalid and should be removed from the routing table.

HTH

Rick

[aravindhs]

Hi Rick

Thanks for the nice explanation.

I have an OSPF Question for you please.

I understand that an ABR starts flooding Type3 LSAs only after it realises that the area is synchronised, by means of 'sequence number'.

My question is, how does the ABR or for that matter, any router in an area come to a conclusion that the area is synchronised ? What specific field in what TYPE of OSPF packet is this inferrence made from ?

Thanks in advance !

Arav

[Richard Burts]

Arav

I am not familiar with what you are talking about where the ABR checks for synchronized by means of sequence number. If you can send some reference link or quote perhaps we can explain it.

In my experience an OSPF router realizes that it is an ABR when it has interface(s) in area 0 and in some other area. (I have taught people who thought that it was enough to be an ABR for the router to have interfaces in two areas (not necessarily in area 0) but in Cisco's implementation it is necessary that one of the areas be area 0). The ABR maintains a separate Link State Data Base (LSDB) for each area. The ABR will have (and will learn more) type 1 and type 2 LSAs in the LSDB for each area. The ABR will generate type 3 LSAs in each LSDB for the type 1 and type 2 LSAs in the other area LSDB. When the ABR is forming an adjacency with any of its neighbors it will forward the LSAs from the appropriate LSDB including the type 3 LSAs.

When you talk about synchronization, I wonder if you are talking about the process of forming an adjacency. In the process of forming an adjacency OSPF routers send each other Data Base Descriptor (DBD) packets. Each DBD has fields that identify each LSA in the LSDB. As each router receives DBD records from its neighbor it checks to see if it already has that LSA in its LSDB. When it finds a DBD for an LSA that it does not already have, it sends a request for the LSA and the other router sends the LSA. After all the DBDs have been exchanged and all requests for LSAs have been completed the two LSDBs are considered synchronized.

HTH

Rick