Configuring interfaces for takeover

Interface takeover

When one filer takes over another, it also takes over the interfaces of the failed filer so that it can maintain the same network connections as before the takeover. You can configure interfaces to enable the takeover of interfaces to happen properly. When you configure a new filer, you use the setup command to configure interfaces, but if you are adding interfaces, you use the ifconfig command, as described in the next section.

Network connections

Each node must be able to assume the identity of its partner on the networks to which the filers are connected.

Each node in a cluster must have interface attachments to the same collection of subnetworks.
Each filer must have interfaces that can take over each IP address and MAC (Media Access Control) address on the other filer.

In a noncluster configuration, each interface acts in one role: supporting the identity of the local node. In a cluster configuration, each interface can act in one of two roles, as described in the following table.

Normal mode means operation when there is no takeover.
Takeover mode means that the node has taken over the failed filer.


Interface type	Normal mode	Takeover mode
Dedicated	Supports the identity of the local node	Idle.
Hot standby	Idle	Supports the identity of the virtual filer.
Shared	Supports the identity of the local node.	Supports the identity of both the local node and the failed filer.

Takeover configurations

You can have two kinds of interface configurations for takeover.

A configuration using separate mirrored NICs. You use this configuration with cards that support only single MAC addresses. For each interface, you configure a dedicated interface on one filer and a hot standby interface on the other.
A configuration using NICs with dual MAC capability. A pair of dual-mac interfaces, one interface to a filer, mirror each other.

FORE/IP and LANE interfaces

You can configure FORE/IP and LANE interfaces as follows:

You must configure FORE/IP interfaces as dedicated and hot standby interfaces.
For LANE, you can configure ATM adapter interfaces as mirrored interface pairs, provided that after takeover, a given ELAN is configured only once on an adapter.

You can specify FORE/IP and LANE interfaces using interface names.

Takeover configuration with dedicated and hot standby interfaces

In this configuration, you use a pair of NICs for each interface. Each NIC can use only one MAC address. One NIC acts as a dedicated interface and the other acts as a hot standby.

If Filer 1 fails, the following occurs on takeover:

e0 stops functioning, but e0 on Filer 2 substitutes for the failed interface and handles Filer 1's network connection with the 230 network.
e2 stops functioning, but e2 on Filer 2 substitutes for the failed interface and handles Filer 1's network connection with the 223 network.

If Filer 2 fails, e1 stops functioning, but e1 on Filer 2 substitutes for the failed interface and handles Filer 2's network connection with the 230 network.

The procedure for setting up a configuration with individual mirrored NICs is in "Configuring dedicated and hot standby interfaces."

Takeover configuration with shared interfaces

In this configuration, you use a pair of NICs that support two MAC addresses to mirror two interfaces, as shown below with the interfaces that connects to the 230 net.

With the configuration shown in the preceding figure, if Filer 1 fails, the following occurs on takeover:

e0 stops functioning, but the secondary address on e0 on Filer 2 handles Filer 1's network connection with the 230 network.
e1 stops functioning, but e1 on Filer 2 substitutes for the failed interface and handles Filer 1's network connection with the 223 network.

If Filer 2 fails, e1 stops functioning, but e1 on Filer 2 substitutes for the failed interface and handles Filer 2's network connection with the 223 network.

Part No. 210-02334
File: clust2.htm
Last updated: 04/26/99 08:11:41