Enabling cluster failover in a DAFS configuration

About the failover path

When a filer in a cluster fails, the DAFS Database Accelerator client must find the failover paths over the VI network to the takeover filer. Each VI NIC on the client must connect to a VI NIC on the takeover filer. The client uses the IP addresses of VI NICs on the takeover filer to identify the failover path. The IP addresses of the VI NICs are listed in the client's /etc/hosts file.

Configuration example

The following figure shows a sample configuration in which a client with two VI NICs connects to a clustered pair of filers through switches.

In this configuration, each VI NIC on a filer has a corresponding failover NIC on the takeover filer. When a filer fails, the following happens:

If Filer A fails, Filer B is the takeover filer. NIC4 on Filer B is the failover path for NIC2. NIC5 on Filer B is the failover path for NIC3.
If Filer B fails, Filer A is the takeover filer. NIC2 on Filer A is the failover path for NIC4. NIC3 is the failover path for NIC5.

Each VI NIC on the client uses the failover path to reach the takeover filer. For example, if Filer A fails, then the following happens:

NIC0 on the client connects to Filer B via NIC4.
NIC1 on the client connects to Filer B via NIC5.

Enabling failover paths

To enable failover paths between the DAFS Database Accelerator client and the takeover filer, you must configure the following:

On the client, define the IP addresses of failover VI NICs in the client's /etc/hosts file.
On the filer, use the dafs attach command to define a failover VI NIC as a partner interface.

Detailed information

For detailed information, see the following sections: