29.5.3 Operation in the event of a member switch failure and recovery

(a) Operation in case of master switch failure

The following figure shows the operation when a failure occurs in the master switch.

Figure 29-25: Operation when a master switch failure occurs on the master node

When the master switch fails and stops, the health check frame (HC(M)) sent by the master switch is stopped. The backup switch switches to the new master switch and operates in the following order:

1. Changing VLAN status of data-transferring rings

2. Transmitting Flush Control Frames

3. Clearing the MAC address table

The master switch then starts monitoring the ring status and resumes sending the health check frame (HC(M)). However, the master switch does not receive the health check frame (HC(M)) it sends, so it detects a ring failure. The master switch that detects the failure changes the monitoring status.

When the master switch is switched, the new master switch does not inherit the ring state of the original master switch.

(b) Operation in case of backup switch failure

The following figure shows what happens if the backup switch fails.

Figure 29-26: Operation when a backup switch failure occurs on the master node

If the backup switch fails and stops, the master switch cannot receive a health check frame (HC(M)) in both directions and detects a ring fault. The master switch that detects a failure operates in the following order:

1. Changing VLAN status of data-transferring rings

2. Transmitting Flush Control Frames

3. Clearing the MAC address table

4. Changing the Monitoring State

(c) Operation in Recovery from a Member Switch Failure

The following figure shows the operation when a member switch recovers from a fault.

Figure 29-27: Operation during recovery from a member switch failure on the master node

When a member switch recovers from a failure, it becomes a backup switch and configures a stack of two member switches.

Once the backup switch is restored, the master switch can receive the health check frame (HC(M)) it sends. If a health check frame (HC(M)) sent by the master switch is received while a ring fault is detected, the master switch determines that the ring fault has been restored and performs recovery operations in the following order:

1. Changing VLAN status of data-transferring rings

2. Transmitting Flush Control Frames

3. Clearing the MAC address table

4. Changing the Monitoring State

(a) Operation in case of a member switch failure

The following figure shows what happens when a member switch fails.

Figure 29-28: Operation when a member switch failure occurs on a transit node

When a member switch fails, it becomes a single member switch stack.

Because the ring port of the failed member switch goes down, the master node cannot receive health check frames (HC(M)) in both directions and detects a ring failure. The master node that detects the failure operates in the same order as in a ring configuration that does not include the stacked nodes.

When the master switch of the failed transit node receives a flush control frame sent from the master node, it clears MAC address table entry for the ring port. Clearing the MAC address table entries allows paths to be switched quickly.

(b) Operation in Recovery from a Member Switch Failure

The following figure shows the operation when a member switch recovers from a fault.

Figure 29-29: Operation during recovery from a member switch failure at a transit node

When a member switch recovers from a failure, it becomes a backup switch and configures a stack of two member switches.

Restoring the ring port on the backup switch allows the master node to receive health check frames (HC(M)) in both directions. If a health check frame (HC(M)) sent by the master node is received while a ring fault is detected, the master node determines that the ring fault has been restored and performs recovery operations in the same order as in a ring configuration that does not include nodes in a stacked configuration.

When the master switch of a transit node that has recovered from a fault receives a flush control frame sent from the master node, it clears MAC address table entry for the ring port. Clearing the MAC address table entries allows paths to be switched quickly.