Advertising a default route into OSPF

Why can R1 reach the exit when the other routers cannot?

Let's say PC-1 and PC-2 can reach each other, but neither can reach a network beyond R1. R1 already has a default route. In IPv4, that is prefix 0.0.0.0/0: it matches every destination, but any matching longer prefix takes priority. It is the fallback when the router has no more specific route. The missing step is telling R2 and R3 to use that exit.

We'll follow one network through four situations. OSPF (Open Shortest Path First) is the routing protocol the internal routers use to share network information. This IPv4 lesson uses OSPF version 2 (OSPFv2).

Keep two questions separate: does R1 know a way out, and have the other routers learned to use it? A working answer to the first question does not automatically answer the second.

What does the network look like before the change?

PC-1 uses R2 at 192.168.2.1 as its gateway. PC-2 uses R3 at 192.168.3.1. The router links already run OSPF in normal area 0, a part of the OSPF network that can carry external route records. R1 connects to the upstream router labelled ISP, or Internet Service Provider.

In this isolated lab, the ISP router has a software interface, Loopback0, at 198.51.100.10/32. We use this documentation address as an external test target, not a real Internet website. R1's static default points to the ISP at 203.0.113.1.

Gi means GigabitEthernet, a physical network interface. A router ID (RID) identifies a router in OSPF; it does not have to be the IP used to reach it. R1 uses RID 1.1.1.1, R2 uses 2.2.2.2 and R3 uses 3.3.3.3. The diagram labels the interface IPs separately.

A default in R1's routing table is not automatically advertised through OSPF. Without another matching route, R2 and R3 cannot send traffic to our external target.

Figure 1 separates the three decisions. Read the numbered callouts from R1 on the right towards R3 on the left. Solid grey lines are physical connections. The small table shows only R3's selected default, not its complete routing table.

At number 1, R1 already knows an exit. Its next hop, 203.0.113.1, is the ISP's interface on the directly connected /30 link. R1 can use that route for a destination outside its known networks. R1's own ISP-facing address is 203.0.113.2; the two ends of this cable have different addresses.

At numbers 2 and 3, a connected cable is not enough. R2 and R3 know their internal OSPF routes, but neither has a route that matches 198.51.100.10. R3's empty default row therefore means “no fallback exit”, not “no routes at all”. PC-2 can still reach PC-1 through the internal network.

Check R1 first. This shortened output confirms its local default:

R1#show ip route 0.0.0.0
S*  0.0.0.0/0 [1/0] via 203.0.113.1

Follow a packet from PC-2 before we change anything. Its gateway R3 checks destination 198.51.100.10. None of its internal routes matches, and no default is installed. It drops the packet there. The packet does not reach R1 just because R1 has an ISP cable.

How do we tell the other routers about the exit?

On R1, enter default-information originate under the OSPF process. Here, originate means create and announce the route information.

R1 creates a Type 5 link-state advertisement (LSA), an OSPF record describing an external route. It describes 0.0.0.0/0. R2 receives the information and passes it towards R3.

R1 now has an additional OSPF role: autonomous system boundary router (ASBR). In this example, it introduces a route from outside OSPF into the OSPF network. R2 does not become an ASBR merely because it passes R1's record on.

Each router chooses its own next hop towards R1. R2 uses R1 at 10.0.12.1; R3 uses its immediate neighbour R2 at 10.0.23.1.

Figure 2 uses exactly the same cables, interfaces and addresses. Only the default-origination policy has changed. The dashed arrows point left because they show OSPF information moving away from R1; they are not the path of a browser request.

Number 1: R1 creates the external route record. The Type 5 LSA identifies prefix 0.0.0.0/0, external metric type E2 and metric 1. Its advertising-router field identifies R1 as 1.1.1.1. That router ID identifies the originator; it is not an instruction to use 1.1.1.1 as the next-hop IP.

Number 2: R2 calculates its own forwarding choice. R2 receives the LSA, calculates its path towards the originator and installs the default through R1's adjacent interface, 10.0.12.1.

R2 also floods the same external route information towards R3. It is passing on R1's record, not originating a different external default under R2's router ID.

Number 3: R3 selects its immediate neighbour, R2. The table in Figure 2 shows 10.0.23.1 through R3's Gi0/0. R3 cannot use the ISP address as a directly connected next hop: the ISP is several links away. The E2 metric stays 1 in this baseline; it does not mean the ISP is one hop away.

Now repeat PC-2's request to 198.51.100.10. The forwarding direction is PC-2 → R3 → R2 → R1 → ISP.

R3 resolves R2's local MAC with ARP if it is not cached, then sends a frame to R2. The IP packet still names the external target as its destination. Each router checks that destination and builds the next Ethernet frame; it does not put the whole route into the packet.

The reply also needs the return routes prepared in the lab below.

R3#show ip route 0.0.0.0
O*E2  0.0.0.0/0 [110/1] via 10.0.23.1, GigabitEthernet0/0

Read the advertisement as well as the route

The routing table shows R3's selected forwarding route. Its OSPF database shows the record R1 announced. Look for these fields:

R3#show ip ospf database external 0.0.0.0
Link State ID: 0.0.0.0
Advertising Router: 1.1.1.1
Network Mask: /0
Metric Type: 2
Metric: 1
Forward Address: 0.0.0.0

What happens if R1's link to the ISP fails?

Now disconnect R1's Gi0/2 link. We assume this removes R1's only installed default, while its internal OSPF link stays working.

Removing a broken exit does not create another exit. Without an alternative route, the external destination remains unreachable.

Figure 3 marks the failed ISP cable with a red cross. The dashed arrows now represent the withdrawal information travelling through the still-working internal OSPF network.

Number 1: R1 loses the condition for plain origination. The interface failure removes the usable path to the static next hop in this lab, so R1's only default leaves its routing table. With no always or route-map, R1 stops originating that default and flushes its external LSA.

Numbers 2 and 3: the other routers process the change. R2 receives and floods the withdrawal; R3 receives it through R2. Each router recalculates the affected route. After convergence, the default row is gone on R3.

A later packet to the external target has no matching route and is dropped, rather than being sent towards a known failed exit. The directly connected links and internal OSPF routes remain available.

Withdrawal means removing previously announced information. The change is carried in a Link State Update (LSU), an OSPF packet that can carry LSA records. An LSA is a record; an LSU is the packet carrying it.

What changes if we add always?

The alternative default-information originate always keeps announcing the default even when R1 has no default in its own routing table. We are not applying a route-map, a policy that can change the conditions for advertisement.

Always keeps the advertisement; it does not repair the connection. Keeping a default stable can reduce changes in its advertisement, but a failed exit can keep attracting traffic. It needs a deliberate routing design.

Figure 4 keeps the failed cable from Figure 3 and changes only the origination policy. Compare the retained row in R3's table with the empty row in Figure 3.

Number 1: R1 keeps originating despite its missing exit. The always keyword removes the normal requirement for an installed default in this unqualified example. It does not add a replacement static route, bring the interface up or repair the ISP.

Numbers 2 and 3: R2 and R3 still have a valid-looking advertisement. They retain their next hops towards R1 and keep forwarding otherwise unmatched traffic along that path.

R1 receives the packet but has no usable onward or more-specific route, so it drops it. This is a black hole: an advertised path attracts traffic that cannot be delivered. An installed default on R3 alone therefore does not prove that the external test will succeed.

What if the link stays up but service beyond it fails?

R1's untracked static default may stay installed if the ISP next hop is still reachable. Even plain origination can then continue. An interface showing up/up confirms local link state; an installed route confirms a routing entry. Neither proves that an external service is available.

Try the change, then verify it

Use an isolated Cisco IOS or IOS XE practice lab. Assign the diagram's addresses, enable the interfaces and establish the internal OSPF neighbours first. The router links use OSPF point-to-point; the LAN interfaces are enabled in area 0 but passive, meaning they advertise their networks without forming neighbours with PCs. The ISP link is outside OSPF.

Start at R1's privileged EXEC # prompt. The commands below are documentation-based examples; they have not been run on a physical router.

configure terminal enters configuration mode. The ip route line creates R1's local exit. router ospf 1 selects the local process; its number does not have to match R2's process number. The next line shares the default. end returns to the # prompt.

Check the result in this order

1. On R1, use show ip route 0.0.0.0 to confirm the installed static default.
2. On R3, use show ip ospf database external 0.0.0.0 to find R1's Type 5 record.
3. On R3, use show ip route 0.0.0.0 to confirm the selected next hop is R2 at 10.0.23.1.

ARP (Address Resolution Protocol) finds the next device's MAC address on a local IPv4 Ethernet link. A route can be present while that delivery fails. The ARP lesson explains the exchange.

Test the failure, then restore the link

After the change is processed, R1 loses its static default and R3 loses this learned default. PC-1 and PC-2 should still reach each other. Restore Gi0/2 with no shutdown and confirm the routes return.

For further investigation, follow OSPF troubleshooting. For a complete internal-router build, use the single-area configuration lesson.

References: Cisco default-route origination, IOS command reference, and RFC 2328, sections 14.1 and 16.4.