16.2.3 Network configuration
The multi-node function uses the three networks listed below. Keep these networks physically separated.
-
Client-server network
-
Inter-node network
-
Reset path
- Organization of this subsection
(1) Client-server network
This network is used for communication between HADB clients and HADB servers.
An HADB client uses an alias IP address to connect to the HADB server. Therefore, set the alias IP addresses based on the explanation in Inheriting a LAN in the manual HA Monitor for Linux(R) (x86).
The following table shows an example of IP address and port number settings for the system configuration shown in Figure 16‒1: Example of a system configuration using the multi-node function:
|
No. |
Setting location |
IP address |
Port number |
|---|---|---|---|
|
1 |
Client machine |
10.196.108.111 |
Setting not required |
|
2 |
Server machine hadb01 (master node) |
10.196.108.11 |
23650 |
|
3 |
Server machine hadb02 (slave node) |
10.196.108.12 |
23650 |
|
4 |
Server machine hadb03 (slave node) |
10.196.108.13 |
23650 |
|
5 |
Alias IP address |
10.196.108.143 |
23650 |
(2) Inter-node network
This network is used for communication between HADB servers, and as HA Monitor's monitoring path.
When you build a network, we recommend that you achieve NIC redundancy using Linux's bonding function. Furthermore, when you use the bonding function, have multiple LAN cards available.
For details about the bonding function, see the documentation for the operating system you are using. For details about how to set HA Monitor's monitoring path, see Configuring the monitoring paths in the manual HA Monitor for Linux(R) (x86).
The following table shows an example of IP address and port number settings for the system configuration shown in Figure 16‒1: Example of a system configuration using the multi-node function:
|
No. |
Setting location |
IP address |
Port number |
|---|---|---|---|
|
1 |
HADB inter-server communication for server machine hadb01 (master node) |
172.16.0.11 |
23651 |
|
2 |
HA Monitor's monitoring path for server machine hadb01 (master node) |
172.16.0.11 |
7777 |
|
3 |
HADB inter-server communication for server machine hadb02 (slave node) |
172.16.0.12 |
23651 |
|
4 |
HA Monitor's monitoring path for server machine hadb02 (slave node) |
172.16.0.12 |
7777 |
|
5 |
HADB inter-server communication for server machine hadb03 (slave node) |
172.16.0.13 |
23651 |
|
6 |
HA Monitor's monitoring path for server machine hadb03 (slave node) |
172.16.0.13 |
7777 |
Normally, the network used for communication between HADB servers is separated from the network used as HA Monitor's monitoring path. However, in such a case, HA Monitor cannot detect failures in the network used for communication between HADB servers. When the same network is used for communication between HADB servers and for HA Monitor's monitoring path as indicated in the table above, HA Monitor can detect network failures. Therefore, we recommend that you configure the inter-node network as shown above.
However, because the target network becomes a single point of failure, you need to enhance the network's resiliency by providing a redundant LAN adapter. You can do this by using a program such as the bonding package in Linux, which is a kernel extension package.
(3) Reset path
When using host reset, a reset path is required. SCSI reservation for shared disk does not require a reset path.
A reset path is a network used, when HA Monitor detects a failure, to reset the input/output of the node on which the failure occurred. This network needs to be dedicated to the reset path. For details, see Reset path configuration in the manual HA Monitor for Linux(R) (x86).
The following table shows an example of IP address and port number settings for the system configuration shown in Figure 16‒1: Example of a system configuration using the multi-node function:
|
No. |
Setting location |
IP address |
Port number |
|---|---|---|---|
|
1 |
Reset path of server machine hadb01 (master node) |
192.168.0.11 |
11111 |
|
2 |
SVP of server machine hadb01 (master node) |
192.168.0.21 |
22222 |
|
3 |
Reset path of server machine hadb02 (slave node) |
192.168.0.12 |
11111 |
|
4 |
SVP of server machine hadb02 (slave node) |
192.168.0.22 |
22222 |
|
5 |
Reset path of server machine hadb03 (slave node) |
192.168.0.13 |
11111 |
|
6 |
SVP of server machine hadb03 (slave node) |
192.168.0.23 |
22222 |
As the shared disk data protection method used by HA Monitor, we recommend that you use host reset. In this case, a reset path is required. The server machine on which the HADB server operates must also be equipped with a failure management processor such as an SVP. For details, see Required hardware in the manual HA Monitor for Linux(R) (x86).