Table of ContentsView in Frames

Simple HA pairs with iSCSI

The following example describes how to implement the best practices for using iSCSI with HA pairs.

Consider the following simplified example. Storage System A has a two-port Ethernet card in slot 9. Interface e9a has the IP address 10.1.2.5, and interface e9b has the IP address 10.1.3.5. The two interfaces belong to a user-defined target portal group with tag value 2.

Simplified image of storage system connection to host through Ethernet switch using iSCSI.

Storage System B has the same Ethernet card in slot 9. Interface e9a is assigned 10.1.2.6, and e9b is assigned 10.1.3.6. The two interfaces are in a user-defined target portal group with tag value 2.

In the HA pair, interface e9a on Storage System A is the partner of e9a on Storage System B. Likewise, e9b on System A is the partner of e9b on system B. For more information on configuring interfaces for an HA pair, see the Data ONTAP High Availability and MetroCluster Configuration Guide for 7-Mode.

Now assume that Storage System B fails and its iSCSI sessions are dropped. Storage System A assumes the identity of Storage System B. Interface e9a now has two IP addresses: its original address of 10.1.2.5, and the 10.1.2.6 address from Storage System B. The iSCSI host that was using Storage System B reestablishes its iSCSI session with the target on Storage System A.

If the e9a interface on Storage System A was in a target portal group with a different tag value than the interface on Storage System B, the host might not be able to continue its iSCSI session from Storage System B. This behavior varies depending on the specific host and initiator.

To ensure correct CFO behavior, both the IP address and the tag value must be the same as on the failed system. And because the target portal group tag is a property of the interface and not the IP address, the surviving interface cannot change the tag value during a CFO.