TMS zl Management and Configuration Guide ST.1.2.100916

8-6
High Availability
Overview
If there is only one link between the host switches of an inter-chassis cluster,
and that link fails, the cluster participant cannot receive the “heartbeat”
messages from the master. The participant therefore assumes that the master
has gone offline, so it assumes the role of master and begins to transmit
gratuitous ARP messages over the network to associate the cluster’s IP
addresses with the participant’s MAC addresses. In the meantime, the master
continues to respond to ARP requests by associating its MAC addresses with
the cluster’s IP addresses.
This situation is called a “split brain” situation because both modules are
attempting to perform the same function using the same IP addresses and yet
cannot detect each other. To prevent a split-brain situation, install redundant
links between the host switches.
If one or more failed links result in a split-brain situation and a link is later
restored, the cluster reforms. The master will be the module that was the
master before the link(s) failed.
You will also create a split-brain situation by deleting the HA VLAN on one
switch only. (When you delete an HA VLAN from the host switch, the HA traffic
moves onto the default VLAN without an interruption in traffic.)
Synchronization and Failover
Modules in an HA cluster synchronize:
■ Current startup-config
■ Static routes
■ IDS/IPS signatures
The participant will, therefore, have the current startup-config, the static
routes, and the IDS/IPS signatures if the master in a cluster fails. Many of the
current sessions will failover as well. Table 8-1 lists the connections that fail
over and those that do not, if one module in the cluster becomes unavailable.