The topology can be referred to the attachment.
1. NE40-1 and NE40-2 enable VRRP and NE40-1 serves as master if it is normal.
2. MGW (Huawei) operation mechanism: MGW attaches two MPU boards (these two boards act as master/slave device mutually. Each MPU board has one upstream physical interface. Two upstream physical interfaces share one IP. Normally the interface of NE40-1 (master) is activated and it takes on forwarding service. Although physical interface of NE40-2 (backup) is up, it is not activated and there are no packets sent and received.
3. There are three conditions of master/slave switchover of MGW (Huawei): the master physical interface is down. MGW sends ARP detection packet every 10s. If it does not receive response 16 times straight, there is switchover again (the function is not started for some reasons). There is no traffic on master interface.
Shutdown physical interface of attached MGW on NE40-1. MGW can switch over backup link, but MGW service interrupts.
Change backup board of MGW and the problem is solved.
1. Shutdown physical interface of attached MGW on NE40-1 and VRRP on NE40-1 and NE40-2 does not switchover (there is transparent transmission of heartbeat packet).
2. MGW can normally switch over backup link. Check ARP table of NE40-1 and ARP of MGW can be correctly learnt on interconnected Eth-Trunk of NE40-1 and NE40-2. Check ARP table of NE40-2 and ARP of MGW is correctly learnt from the interface between NE40-2 and MGW.
3. On the interface between NE40-2 and MGW there is no packet inputs before switchover. After switchover the statistics can be referred: NE40-2 sends many unicast, multicast and broadcast packets. However, NE40-2 only receives broadcast packets of MGW and there is no unicast and multicast packet. It indicates that NE40-2 normally receives service packets, but MGW may only send ARP broadcast packet.
4. MGW may not learn ARP table of virtual MAC of gateway of NE40-1, so it does not correctly forward gateway traffic. The alarm log of MGW also proves the point. There are two possibilities: NE40-1 does not response ARP request of gateway of MGW. Or NE40-1 correctly responses, but MGW
5. Test it as follows: Replace PC with MGW and attaches it to the same interface of NE40-2. PC can correctly learn ARP of the gateway and can communicate, indicating NE40-1 can respond ARP request of gateway and the interface and related cables of NE40-2 can be ironed out.
6. Replace PC with the interface directly connecting with backup MPU board of MGW of NE40-2 and do switchover operation again. It is still problematic. PC and MGW cannot communicate.
7. MGW is problematic.
1. As maintenance engineers of IP bearer network, they should have a full knowledge of every kind of access service of IP bearer network and operation of related devices which is of great help for locating the problem of IP bearer network.
2. Simple, feasible and accurate test can rapidly locate the problem.