View PDF
Routing in a cluster
Routing in a cluster works in much the same way as routing in a standalone system. A few points to note:
-
All routing configurations must be performed from the cluster IP address and the configurations are propagated to the other cluster nodes.
-
Routing runs only on spotted SNIP addresses and NSIP addresses.
-
Routes are limited to the maximum number of ECMP routes supported by the upstream router.
-
Node-specific routing configurations must be performed by using the owner-node argument as follows:
! interface vlan97 ! router ospf owner-node 0 ospf router-id 97.131.0.1 exit-owner-node owner-node 1 ospf router-id 97.131.0.2 exit-owner-node owner-node 2 ospf router-id 97.131.0.3 exit-owner-node redistribute kernel network 97.0.0.0/8 area 0 ! -
Retrieve node-specific routing configurations by specifying the node(s) in the owner-node argument as follows:
> vtysh ns# owner-node 0 1 ns(node-0 1)# show cluster state ns(node-0 1)# exit-owner-node -
Clear node-specific routing configurations by specifying the node(s) in the owner-node argument as follows:
> vtysh ns# owner-node 0 1 ns(node-0 1)# clear config ns(node-0 1)# exit-owner-node -
Routing protocol daemons can run and adjacencies can be formed on active and inactive nodes of a cluster.
-
Only active nodes advertise host routes to striped VIP addresses. Spotted VIP addresses are advertised by active owner node.
-
Active and inactive nodes can learn dynamic routes and install them into the routing table.
-
Routes learnt on a node are propagated to other nodes in the cluster only if route propagation is configured. This is mostly needed in asymmetric topologies where the unconnected nodes may not be able to form adjacencies.
ns(config)# ns route-install propagateNote*
Make sure that route propagation is not configured in a symmetric cluster topology as it can result in making the node unavailable to the cluster.