Fault Monitoring and Failover

Failover means that when one FW in a hot standby network fails, the other FW takes over services from the faulty FW. The FW monitors different fault events based on the configured trigger conditions. When a fault occurs, a failover is triggered in a timely manner to ensure service continuity.

Failover Trigger Conditions

The conditions for triggering a failover include heartbeat loss and VGMP group status change.

Heartbeat loss

The FW monitors the heartbeat packets of the peer device to determine whether the peer device is alive and whether a failover is required. In hot standby networking, two FWs send heartbeat packets to each other through the heartbeat link. The default interval for sending heartbeat packets is 1000 ms. If a FW does not receive a heartbeat packet from the peer device for five consecutive heartbeat intervals, it determines that the peer device is faulty and triggers a failover.
VGMP Group Status Change

The FW receives VGMP packets from the peer device through the heartbeat link, learns the priority of the VGMP group on the peer device, and compares the priorities of the local and peer VGMP groups to determine whether to perform a failover. When an interface, or link of the FW is faulty, the priority of the VGMP group decreases. For details, see Table 1. If the priority of the local VGMP group is lower than that of the peer, the state of the local VGMP group changes to standby. Meanwhile, the FW sends a VGMP packet to the peer device to instruct the peer device to perform a failover.

**Table 1** Impact of different fault events on the VGMP group priority
Fault Event	Description
An interface monitored by a VGMP group is faulty.	The impact of the interface fault on the VGMP group priority depends on the configuration. The details are as follows: If a VRRP group is configured on an interface and the interface is faulty, the priority of the VGMP group decreases by 2 x number of VRRP groups on the interface. If you have run the hrp track interface command to configure a VGMP group to monitor the physical interface status, the priority of the VGMP group decreases by 2 for each faulty physical interface. If you have run the hrp track interface command to configure a VGMP group to monitor the status of an Eth-Trunk interface and some member interfaces of the trunk interface are faulty, the priority of the VGMP group decreases by 2 x faulty member interfaces by default. If all member interfaces of a trunk interface are faulty, the priority of the VGMP group decreases by 2 x (1 + number of member interfaces). If a VRRP group is configured on an interface and the hrp track interface command has been run to configure a VGMP group to monitor the interface status and the interface is faulty, the reduction of the priority of the VGMP group is accumulative. For example, two VRRP groups are configured on GigabitEthernet0/0/1, and the hrp track interface GigabitEthernet 0/0/1 command has been configured on the interface. When the interface is faulty, the priority of the VGMP group decreases by 6. If the hrp track vlan command has been used to monitor the VLAN status and an interface added to the VLAN is faulty, the priority of the VGMP group decreases by 2 for each faulty interface. NOTE: If a member interface of a trunk interface is faulty or a member interface is added to or deleted from the trunk interface, the VGMP group priority is affected. You can run the undo hrp track trunk-member enable command to cancel the monitoring of trunk member interface status by the VGMP group.
A link monitored by a VGMP group is faulty.	A VGMP group can monitor the link status in the following ways and adjust the priority of the VGMP group: Monitors the status of an IP-link to indirectly monitor link status. When an IP-link is down, the priority of the VGMP group decreases and failover is triggered. The priority of the VGMP group decreases by 2 every time when an IP-link monitored by the VGMP is down. Monitors the BFD status to indirectly monitor link status. When a BFD session is down, the priority of the VGMP group decreases and failover is triggered. The priority of the VGMP group decreases by 2 every time when a BFD monitored by the VGMP is down. Monitors the status of OSPF neighbors to indirectly monitor link status. When an OSPF neighbor changes from a full to a non-full state, the priority of the VGMP group decreases and a failover is triggered. The priority of the VGMP group decreases by 2 every time an OSPF neighbor monitored by the VGMP group changes to a non-full state. Monitors the status of BGP peers to indirectly monitor link status. When the state of a BGP peer changes from established to non-established, the priority of the VGMP group decreases and a failover is triggered. The priority of the VGMP group decreases by 2 every time when the state of a BGP peer monitored by the VGMP group changes to non-established.

Failover Operation

Table 2 shows the FW failover operations when different fault events occur.

**Table 2** Failover operations
Fault Event	Failover	Operation on the Faulty Device	Operation on the Normal Device	Description
The entire device is faulty.	Yes	N/A	After five heartbeat intervals, the state of the VGMP group changes to active.	When the state of the VGMP group changes to active, the FW performs the following operations to divert network traffic to itself: Changes the state of all VRRP groups that are in the Backup state to Master and starts sending VRRP advertisement packets and gratuitous ARP packets that announce the virtual IP address. Advertises dynamic routes, such as OSPF, OSPFv3, and BGP routes, according to the configurations. Enables the VLAN monitored by the VGMP group. If the device works in mirroring mode, the service interfaces change to the non-passive state and start to send and receive packets. When the state of a VGMP group changes to standby, the FW performs the following operations to divert network traffic to the peer device for forwarding: Changes the state of all VRRP groups in Master state to Backup and stops sending VRRP advertisement packets and gratuitous ARP packets that announce the virtual IP address. Increases the costs of the advertised dynamic routes (OSPF, OSPFv3, and BGP). Disables the VLAN monitored by the VGMP group. If the device works in mirroring mode, the service interfaces change to the passive state, and the service interfaces stop sending and receiving packets (except LACP and LLDP packets).
The heartbeat link is faulty.	Yes	After five heartbeat intervals, the state of the VGMP group changes to active.	After five heartbeat intervals, the state of the VGMP group changes to active.
An interface monitored by a VGMP group is faulty.	Yes	The priority of the VGMP group decreases. The VGMP group is switched to the standby state. A VGMP packet is sent to instruct the peer to perform failover.	After receiving the VGMP packet from the peer device, the device changes the VGMP group to the active state.
A link monitored by a VGMP group is faulty.	Yes	The priority of the VGMP group decreases. The VGMP group is switched to the standby state. A VGMP packet is sent to instruct the peer to perform failover.	After receiving the VGMP packet from the peer device, the device changes the VGMP group to the active state.