We need more examples. If you don't see what you need here, and you create a working config file that you're willing to share, please send it to us. (Also, if you find mistakes in the examples here, well, ahem, just tell us quietly and we'll fix them.)
Gateway configurations are more complicated than a simple host configuration as gateways have multiple interfaces and may run multiple protocols. This sample configuration file connects two nets within an autonomous system. This configuration can be used for any gateway that uses only RIP and that connects a single subnet to a larger network.
# # enable rip # rip yes ; # # using rip, announce subnet 3 via interface 136.66.1.2 # export proto rip interface 136.66.1.2 { proto direct { announce 136.66.3.0 metric 0 ; } ; } ; # # using rip, announce all routes learned from interface 136.66.1.2 # export proto rip interface 136.66.3.1 { proto rip interface 136.66.1.2 { announce all ; } ; } ;
The configuration contains two export statements. The first one tells
GateD to announce, via the RIP protocol and interface 136.66.1.2, a direct
route to subnet 136.66.3.0. The second route announces to subnet 3 all
routes learned from the interface 136.66.12. This includes subnet routes
and any routes to the rest of the world, including default routes. The
first export statement was not required since, by default, GateD announces
every network that is directly connected to the gateway. The first export
statement was explicitly entered for two reasons: to document the intended
structure and to avoid relying on defaults which may be changed in future
releases. In any export statement, the main proto clause and the main interface
clause may define the protocol and interface through which the routes are
advertised. Subsidiary proto and interface clauses define the protocols
and the interfaces from which the routes must be learned.
If the gateway only runs RIP and connects a local backbone to a subnet ,which in turn provides a gateway to the outside world, the file is configured differently. In this example, the gateway announces a default route to the backbone and announces all of the individual subnet routes to the outside world.
# # enable rip # rip yes ; # # using rip, announce all local subnets via 136.66.12.3 # export proto rip interface 136.66.12.3 metric 3 { proto rip interface 136.66.1.5 { announce all ; } ; }; # # using rip, announce default to the local backbone via 136.66.1.5 # export proto rip interface 136.66.3.1 { proto rip interface 136.66.12.3 { announce 0.0.0.0 ; } ; } ;
The first export statement explicitly directs GateD to announce all
of the routes it learns from interface 136.66.1.5 out through interface
136.66.12.3. The explicit announce all is not required. The metric value
is specified in the main statement and applies to every route announced
via this protocol and interface. When the metric is specified in the export
clause, it only applies to the routes specified in that clause. In this
example, metric 3 applies to every route announced over interface 136.66.12.3
via RIP.
This configuration for AS 283 enables both RIP and OSPF protocols, and can be used for testing RIP and OSPF.
#options noinstall ; interfaces { interface le0 passive ; } ; autonomoussystem 283 ; snmp yes ; rip yes { broadcast ; defaultmetric 5 ; interface le version 2 multicast ; } ; ospf yes { traceoptions lsabuild protocol ; monauthkey "ZZZZZZZZ" ; backbone { interface all { priority 2 ; } ; interface le { auth simple "YYYYYYYY" ; } ; | ; } ; static { default gateway 132.236.200.200 preference 140 retain ; }
In the RIP version 2 case, multicast packets are generated on all ethernet
interfaces (le0, le1...). This is a safe way to test RIP since the production
gateways on this network do not support RIP version 2. In the OSPF case,
all interfaces are configured to be Priority 2, so the Proteon routers
(Priority 0) will become designated routers. A simple password is specified
"YYYYYYYY" for authentication of all Ethernet interfaces (which
are the only kind of interfaces in this configuration.)
In this sample configuration, an HP server is configured to be both the area border router and the backbone router.
traceoptions parse ; routerid 195.1.1.2 ; # RIP no ; OSPF yes { defaults { preference 10 ; cost 5 ; type 1 ; } ; area 0.0.0.1 { networks { 195.1.1.0 ; } ; interface 195.1.1.2 nonbroadcast { routers { 195.1.1.1 eligible ; 195.1.1.2 eligible ; } ; priority 15 ; enable ; hellointerval 30 ; routerdeadinterval 30 ; pollinterval 30 ; } ; } ; area 0.0.0.2 { networks { 193.2.1.0 ; 194.1.1.0 ; } ; interface 194.1.1.3 nonbroadcast { routers { 194.1.1.2 eligible ; 194.1.1.3 eligible ; 194.1.1.1 ; } ; priority 15 ; enable ; hellointerval 30 ; routerdeadinterval 30 ; retransmitinterval 30 ; pollinterval 30 ; } ; } ; backbone { interface 15.13.115.156 nonbroadcast { enable ; transitdelay 20 ; priority 10 ; hellointerval 30 ; routerdeadinterval 30 ; retransmitinterval 30 ; pollinterval 30 ; } ; } ; } ;
This example has two areas (0.0.0.1 and 0.0.0.2) besides the OSPF backbone area. In area 0.0.0.1, one other routers besides this router (interface at 196.1.1.2) is eligible across the NMBA network to be polled every 30 seconds. In area 0.0.02, two other routers are listed (194.1.1.2 and 194.1.1.1), only one other router can engage in OSPF packets with this router (194.1.1.2).
In the backbone area, all routers are eligble by default to be polled at interval 30.
Last updated November 30, 1997.
gated@gated.merit.edu