08: Deploying OSPF

Configuring OSPF

Junos OSPF Support

Junos supports OSPF versions 2 and 3. OSPF 2 is for IPv4; OSPF 3 is for v4 and v6. In OSPF v2, authentication is available but disabled by default. Route summarisation is available.

External prefix limits can be used to stop too many external prefixes being acceped into OSPF. This is configured with the prefix-external-limit statement. Graceful restart is available, for a router to inform its neighbours before restarting. Its neighbours will continue to forward traffic to the router whilst it is restarting. A grace period specifies how long it should forward traffic for. This feature is not enabled by default.

Bidirectional forwarding detection (BFD) detects failures in the network. Hello packets are sent at a regular interval. Router failure is detected when a router does not receive a packet in a specified interval, which is shorter than the OSPF hello timer.

Configuring OSPF

OSPF v2 configuration is under the protocols ospf hierarchy. OSPF v3 configuration is under the protocols ospf3 hierarchy.

Every router has a single OSPF router id. It is configured under routing-options router-id, however, Junos will use an IP address which was configured at the time the OSPF process started. This is ideally a non-127/128 loopback address. It is strongly recommended that the router id is configured manually to avoid unpredictable behaviour.

The cost of a link in OSPF, called the ‘metric’, is equal to the reference bandwidth divided by the bandwidth of the link. The metric can be changed either by changing the reference bandwidth, which is defined under protocols ospf reference-bandwidth, or by specifying the metric directly on the interface.

By default, the reference bandwith is 100Mbit/s, specified as 100000000. This means that any interface with bandwidth of 100Mbit/s or higher will have an OSPF metric of 1.

No bandwith is associated with the loopback interface.

To add an interface to OSPF;

protocols {
    ospf {
        area 0.0.0.0 {
            interface ge-0/0/0;
        }
    }
}

Additional configuration can be specified on the interface under protocols ospf area;

interface ge-0/0/1 {
    metric 100;
    hello-interval 4;
    dead-interval 10;
}

To accept a specific, directly connected route into OSPF, a routing policy can be written;

policy-options {
    policy-statement to-ospf {
        term match-direct-route {
            from {
                protocol direct;
                route-filter 172.18.1.0/24 exact;
            }
            then accept;
        }
    }
}

The routing policy is applied to export the specified route into OSPF;

ospf {
    export to-ospf;
}

Monitoring OSPF

Check the states of OSPF adjacencies with;

show ospf neighbor

Output is terse by default. The extensive option will also display area, options, DR, BDR, uptime and adjacency uptime.

OSPF neighbours can be cleared with clear ospf neighbor.

This will cause a momentary disruption. Adjacencies should be reformed immediately.

Find information about individual routes with show route 172.18.1.0/24. The result will show that the route was learned from OSPF. It will also show the next hop IP address and interface.

Show the OSPF interface parameters with show ospf interface extensive. Interfaces with OSPF enabled are displayed.

Show routes learned by OSPF with show ospf route.

Show the OSPF database with show ospf database. For ABRs, the output is grouped by OSPF area.

Clear the OSPF database with clear ospf database. The purge option forces the local router to set all LSAs in its database to max-age, causing them to age out. The LSAs are then re-flooded, because the OSPF specification requires that any LSAs which have reached max-age must be re-flooded, regardless of whether the LSA was generated by the local router or not. The LSAs have max-age set, so the originating router which receives an LSA with max-age set will be compelled to refresh the LSA and itself re-flood it. Although disruptive, this technique is effective for removing stale routes.

Display OSPF logging information with show ospf log. SPF calculation instances are listed.

Show statistics with show ospf stastics. The number of packets of each type, and the number of LSAs, transmitted and received, are displayed.

Troubleshooting OSPF

Common Adjacency Problems

If no neighbour is detected, first check physical and data link (Layer 2) connectivity. Check basic Layer 3 attributes, such as mismatched IP address and subnet mask. Check that all OSPF variables match; in particular, area number, area type, hello & dead interval, and network type.

If the router is stuck in ExStart state, check the MTU settings.

If the router is stuck in 2-way state, check that the router is not a DR or BDR (it is DRother). It is normal for DRother routers to not progress beyond 2-way state.

Using Traceoptions

Enable traceoptions in the OSPF configuration;

protocols {
    ospf {
        traceoptions {
            file trace-ospf;
            flag error detail;
            flag event detail;
        }
    }
}

The traceoptions logging output will be stored in /var/log/trace-ospf. It can be viewed with show log trace-ospf or monitor start commands. The flag keyword determines what will be captured in the log.

Viewing OSPF Error Counters

Error counters, split by type of error, are included in the output of show ospf statistics. They can be reset with the clear ospf counters command.