JING JIE UniPBX-2000, Административное руководство

Страница: 59 / 219

Using the System
59
Active/Standby Redundant for 2 Ethernet Legs
Above case but IPV4 only
Some system architecture examples will be showing in the following topic. For high
available, if the fail-over is happened, the call will be continue for 2-5 second silence
and voice recording will became 2 separate recorded file. However, the HA will try to
keep the server running on the same server as long as possible, it will only failed to
the standby node when the following is happened:
1. System is crashed several times (3 times in default) without continues working for
30 seconds.
2. The machine is crashed or had hardware failure.
3. If the virtual IP cannot be added into the node.
Each cluster need had an identifier which is Cluster ID. It is recommended to use
different UDP broadcasting port to reduce the CPU usage for verifying each
heartbeat message. You can use the default port which is port number 654 or higher
than port number 60000 to avoid some port conflicts.
The following is the system requirements before you can install or start the
HA service:
1. Both servers need to have 2 Ethernet interfaces which need connect to a
separate Ethernet switch for redundant purpose.
2. If you are using CISCO switch, please make sure the Ethernet port's is set
to become a fast port (turn off spanning tree).
3. Both servers need use NTP to synchronize the time. Otherwise the HA
service will not working correctly.
Please check the above requirements before you can move on.
2.2.15.1 Active/Standby (1 Ethernet)
In this mode, the system is serving for WAN interface and another node is a standby
node. In the normal case, the SIP CPE is register to their server through VIP. If one
of the machine is down, the another peering node will taking over the VIP and
continue the service.