H3C S5120-EI Series Configuration Manual - page 15
8
IRF fabric topology maintenance
As soon as a member switch is down or an IRF link is down, its neighbor switches broadcast the leaving
of the switch to other members. When a member switch receives the leave message, it looks up its IRF
topology database to determine whether the leaving switch is the master. If yes, the member switch starts
a master election and updates its IRF topology database. If the leaving switch is not a master, the member
switch directly updates its IRF topology database.
NOTE:
An IRF port goes down only when all its physical IRF ports are down.
IRF multi-active detection
An IRF link failure causes an IRF fabric to split in two IRF fabrics operating with the same Layer 3
configurations, such as the same IP address. To avoid IP address collision and network problems, IRF
uses multi-active detection (MAD) mechanism to detect the presence of multiple identical IRF fabrics and
handle collisions. MAD provides the following functions:
Detection
MAD detects identical active IRF fabrics with the same global configuration by extending the Link
Aggregation Control Protocol (LACP) or the Gratuitous Address Resolution (ARP) protocol. For more
information, see ”
Configuring MAD detection
.”
NOTE:
For more information about LACP, see
Layer 2 — LAN Switching Configuration Guide
; for more
information about gratuitous ARP, see
Layer 3 — IP Services Configuration Guide
.
Collision handling
If multiple identical active IRF fabrics are detected, only the IRF fabric that has the lowest master ID can
operate in active state and forward traffic. MAD sets all other IRF fabrics in recovery state (disabled) and
shuts down all physical ports but the IRF ports and other ports you have specified with the mad exclude
interface command. See
Figure 6
.
Failure recovery
An IRF link failure causes an IRF fabric to divide into two fabrics and multi-active collision occurs. When
the system detects the collision, it holds a master election between the two collided IRF fabrics. The IRF
fabric whose master’s member ID is smaller prevails and operates normally. The state of the other IRF
fabric transits to the recovery state and temporarily cannot forward data packets. In this case, recover the
IRF fabric by repairing the IRF link first (The switch tries to automatically repair the failed IRF links. If the
reparation fails, manually repair the failed links.)
When the link is recovered, the IRF fabric in recovery state automatically reboots, and then the IRF fabrics
both in active state and in recovery state automatically merge into one. Service ports that were shut down
and belonged to the IRF fabric in recovery state automatically restore their original physical state, and the
whole IRF fabric recovers, as shown in
Figure 6
.