Brocade Access Gateway Administrator's Guide Supporting Fabric OS v7.0.0 (53-1002156-01, April 2011)

Access Gateway Administrator’s Guide 45
53-1002156-01
Failover
3
NOTE
All the device data must be persistent in case of a reboot. During a reboot, the tables will be dumped
to the persistent_NPIV_config file.
Persistent ALPA policy considerations
The Persistent ALPA policy is not supported in the following situations:
When AG N_Ports are connected to the shared ports of 48-port Director blades.
Cisco fabrics. Enable Persistent FCID mode on the connecting Cisco switch to achieve the
same functionality.
Persistent ALPA configuration will not change to the default when the configDefault command
is used, but will retain the previous configuration.
Failover
The Access Gateway Failover policy ensures maximum uptime for the servers. When a port is
configured as an N_Port, the Failover policy is enabled by default and is enforced during power-up.
The Failover policy allows hosts and targets to automatically remap to another online N_Port if the
primary N-Port goes offline.
NOTE
For port mapping, the Failover policy must be enabled on an N_Port for failover to occur. For device
mapping, if a device is mapped to an N_Port in a port group, the device will always reconnect to the
least-loaded online N_Port in the group (or secondary N_Port in the group if configured) if the
primary N_Port goes offline. This occurs regardless of whether the Failover policy is enabled or
disabled for the primary N_Port.
Failover with port mapping
The Failover policy allows F_Ports to automatically remap to an online N_Port if the primary N_Port
goes offline. If multiple N_Ports are available for failover, the Failover policy evenly distributes the
F_Ports to available N_Ports belonging to the same N_Port group. If no other N_Port is available,
failover does not occur and the F_Ports mapped to the primary N_Port go offline as well.
AG provides an option to specify a secondary failover N_Port for an F_Port.
Failover configurations in Access Gateway
The following sequence describes how a failover event occurs:
An N_Port goes offline.
All F_Ports mapped to that N_Port are temporarily disabled.
If the Failover policy is enabled on an offline N_Port, the F_Ports mapped to it will be
distributed among available online N_Ports. If a secondary N_Port is defined for any of these
F_Ports, these F_Ports will be mapped to those N_Ports. If the Port Grouping policy is enabled,
then the F_Ports only fail over to N_Ports that belong to the same port group as the originally
offline N_Port.