Optimizing OSPF: Troubleshooting Paths and LSAs

1. The Logic of OSPF Path Selection

Unlike adjacency failures, which are often caused by simple configuration mismatches, suboptimal routing issues require a deeper understanding of the OSPF decision-making process. OSPF is deterministic; it follows a strict set of rules to select the best path. If a path seems "wrong," it's because it is optimal according to these rules, even if it's not what you intended.

The path selection process is a hierarchy. OSPF routers prefer routes in the following order, from most to least preferred:

1. Intra-Area (O): Routes to destinations within the same area.
2. Inter-Area (O IA): Routes to destinations in other OSPF areas.
3. External Type 1 (E1 / N1): Routes redistributed from another routing protocol, where the cost accumulates as it travels through the OSPF domain.
4. External Type 2 (E2 / N2): The default type for redistributed routes, where the cost is fixed and does not change within the OSPF domain.

Only if routes are of the same type does OSPF compare the cost to break the tie.

To see this hierarchy explained clearly, let's watch a short segment from PM Networking.

This strict preference (Route Type > Cost) is a common source of "suboptimal" routing. A low-cost Inter-Area route will always be preferred over any External route, regardless of the external route's cost.

2. Symptom: A Route is Missing from the Routing Table

One of the most common trouble tickets is "I can't reach network X, even though OSPF is running." Your first step is to determine if the route is simply missing or if it's present but pointing the wrong way.

Let's assume you run show ip route <network> and it's not there. The troubleshooting process now has two distinct phases:

1. Is the corresponding Link-State Advertisement (LSA) present in the OSPF database? (show ip ospf database)
2. If the LSA is in the database, why didn't the SPF algorithm install it into the routing table (RIB)?

Case 1: LSA is Missing from the Database

If an LSA never makes it to a router's database, the router can't possibly calculate a path. LSAs are the "puzzle pieces" of the OSPF topology map. If a piece is missing, that part of the map is blank.

Based on the video, common reasons for a missing LSA include:

• Area Type Restrictions: A router in a stub area will never see Type 5 or Type 4 LSAs. It receives a default route from the ABR instead. A router in a totally stubby area won't even see Type 3 LSAs from other areas. Always check the area type if external or inter-area routes are missing.
• No Path to the Originating Router: If an ABR or ASBR is down or unreachable, its LSAs cannot be propagated.
• Filtering: While less common for intra-domain routing, it is possible to filter LSAs at an ABR using area <id> filter-list.

Case 2: LSA is in the Database, but Not in the RIB

This is a classic CCNP-level troubleshooting scenario. You run show ip ospf database external <network> and see the LSA, but show ip route <network> yields nothing. This means the SPF algorithm ran, saw the LSA, but decided not to install the route.

The official Cisco guide, "Troubleshoot Open Shortest Path First Route Database Issues," is the definitive resource for these problems.

Let's summarize the key reasons and add a few more:

• Filtering with distribute-list in: As the document and the video below show, this command doesn't stop the LSA from entering the database, but it does prevent the route from being installed in the RIB. It's a "policy filter" applied after the SPF calculation.
• Unreachable Forwarding Address (for Type 5 LSAs): This is a critical and frequently tested concept. For a router to install a Type 5 LSA, it must have an intra-area or inter-area route to the LSA's forwarding address. If the only route to the forwarding address is external (or a default route), OSPF considers it a potential loop and will not install the route.
• Unreachable Advertising Router: For any LSA, if the router that originated it (the ADV Router) is considered unreachable by the local SPF algorithm, any routes within that LSA will not be installed. This is often indicated by an "Adv Router is not-reachable" message in the show ip ospf database output.
• Better Administrative Distance (AD): If the router is learning the exact same prefix from another source with a lower (better) AD, that route will be installed instead. For example, EIGRP (AD 90) or a static route (AD 1) will be preferred over OSPF (AD 110). Check with show ip route <prefix>.

The following video shows a practical example of filtering routes using a route-map with a distribute-list, a powerful technique for controlling which routes get installed.

3. Symptom: Traffic is Taking a Suboptimal Path

Now, let's address the second major issue: the route is in the routing table, but it's not the path you want. This means you need to influence the OSPF path selection process.

To fix a suboptimal path, you must make your desired path "better" in the eyes of OSPF. You can do this by:

1. Manipulating Cost: The most common method. You can lower the cost of the desired path or increase the cost of the undesired path. This is the primary tool for influencing intra-area and inter-area routes.
2. Manipulating Route Type: Influencing whether a route is seen as E1 or E2 can change the path selection when multiple ASBRs exist.
3. Using Summarization: A more advanced technique where you can make one path less specific, causing routers to prefer another, more specific path.

This video demonstrates all three techniques brilliantly.

Key Path Manipulation Techniques:

• Changing Cost: Use the ip ospf cost <value> command on an interface. Remember to also consider the auto-cost reference-bandwidth command on all routers to ensure costs are calculated appropriately for modern high-speed links.
• Changing Metric Type: During redistribution, you can specify metric-type 1 to make the cost cumulative (E1), or leave the default metric-type 2 for a fixed cost (E2). E1 is often preferred in networks with multiple ASBRs to make a cost-aware decision.
• Longest Match Routing: As shown in the summarization example, routers always prefer a more specific route. Advertising a /24 from one path and a summary /21 from another will cause traffic for that /24 to always choose the first path.

Conclusion

Troubleshooting OSPF routing involves moving beyond simple connectivity and applying your knowledge of the protocol's internal logic. When a route is missing or suboptimal, you must think like the SPF algorithm.

Key Takeaways:

• Path Selection is Hierarchical: OSPF's decision is always Route Type > Cost. Understanding this hierarchy is essential.
• Missing Routes? Check LSDB First: If an LSA isn't in the database, it's a propagation problem (often due to area type restrictions). If it's in the database but not the RIB, it's an installation problem (often due to filtering, AD, or an unreachable forwarding address).
• Suboptimal Paths? Manipulate the Metrics: To change a path, you must make the desired path more attractive to the SPF algorithm by lowering its cost, changing its route type, or making it more specific.
• External Routes are Special: Pay close attention to the forwarding address and the metric-type (E1 vs. E2) when troubleshooting routes that have been redistributed into OSPF.

Preview of the Next Lesson:

We have now completed our deep dive into OSPF, a foundational Interior Gateway Protocol (IGP). In the next module, we will shift our focus to the protocol that runs the global internet: Border Gateway Protocol (BGP). We'll start by exploring the fundamental differences between IGPs and EGPs, and you will learn to differentiate between eBGP and iBGP relationships and their use cases.