After you configure a server host with a server name, you can use the
deadtime command to configure each server per server group. Configuring deadtime within a server group allows you to direct AAA traffic
to separate groups of servers that have different operational characteristics.
Configuring deadtime is not limited to a global configuration. A separate timer is attached to each server host in every
server group. Therefore, when a server is found to be unresponsive after numerous retransmissions and timeouts, the server
is assumed to be dead. The timers attached to each server host in all server groups are triggered. In essence, the timers
are checked and subsequent requests to a server (once it is assumed to be dead) are directed to alternate timers, if configured.
When the network access server receives a reply from the server, it checks and stops all configured timers (if running) for
that server in all server groups.
If the timer has expired, the server to which the timer is attached is assumed to be alive. This becomes the only server
that can be tried for later AAA requests using the server groups to which the timer belongs.
Note |
Because one server has different timers and might have different deadtime values configured in the server groups, the same
server might, in the future, have different states (dead and alive) at the same time.
|
Note |
To change the state of a server, you must start and stop all configured timers in all server groups.
|
The size of the server group will be slightly increased because of the addition of new timers and the deadtime attribute.
The overall impact of the structure depends on the number and size of the server groups and how the servers are shared among
server groups in a specific configuration.