Technical Considerations for a Serviceguard Cluster that Spans Multiple IP Subnets, July 2009

13
Figure 7 – Recommended configuration on a multi-site cross-subnet cluster with dual dedicated heartbeat subnets and a
separate package subnet
Quorum Service
Datacenter 1 Datacenter 2
3rd Location
R1
5.1.1.1
5.1.2.1
5.1.2.2 5.1.1.2 5.2.1.1 5.2.2.1
5.2.2.2
5.2.1.2
R3
R2
R5
R4
5.1.1.0 5.2.1.0
5.2.2.0
5.1.2.0
Primary network router
Fully redundant inter-site link
R#
Backup network router
R#
1
st
dedicated hearbeat
2
nd
dedicated hearbeat
Primary package network
Network switch
Highly available network
for user, client and quorum server access
Standby package network
5.1.3.1
5.1.3.2
5.1.3.0
5.2.3.1
5.2.3.2
5.2.3.0
R6
Node A Node B
Node C Node D
Gateway IPs
5.1.1.254
5.1.2.254
Gateway IPs
5.2.1.254
5.2.2.254
Example 4: Package subnet IP Monitor recommendations
You can add another layer of protection for your package subnets 5.1.2.0 and 5.2.2.0 in Example 3
by configuring the IP Monitor on these subnets.
If subnet 5.1.2.0 has the gateway IP address 5.1.2.254 and subnet 5.2.2.0 has the gateway IP
address 5.2.2.254, Serviceguard will configure the IP Monitor on these subnets by default, using
each subnet’s gateway IP address as the polling target for that subnet.
Serviceguard will send polling messages at the IP level from each of the IP addresses on the subnet to
the configured polling targets and make sure each polling message receives a polling reply from the
polling targets. In this example, polling messages will be sent from 5.1.2.1 and 5.1.2.2 to
5.1.2.254. Similarly, polling messages will be sent from 5.2.2.1 and 5.2.2.2 to 5.2.2.254.
You can use any IP address for the subnet’s polling target as long as the IP address is on an IP route
that you want to monitor. You should not use an IP address on a cluster node on the same subnet that
you want to monitor, however. This is because polling messages will be sent from an IP address to
itself when that IP address is configured as the polling target of the subnet. When the IP address fails,
it might still be able to ping itself and the failure will not be detected. For example, 5.1.2.1 and
5.1.2.2 should not be configured as the polling targets for subnet 5.1.2.0.