Cross-connect protection upgrade is applicable to symmetrical networking. Upstream and downstream devices are symmetrically connected to master and backup chassis on a virtual cluster. The following figure shows the process.

Chassis A is the master, and chassis B is the backup. The upstream and downstream traffic is balanced through chassis A and B.

The upgrade process is as follows:

1. Load a target software package on both chassis and set the file to be loaded at a next startup to the target software package.
2. Manually close all external interfaces on chassis B. Traffic switches to chassis A. The following figure shows the process.
   
   
3. Reset chassis B and restore the configuration file.

   After chassis B is reset, it does not detect chassis A and becomes the master. Therefore, two master chassis exist. External connections to chassis B are disconnected, and the two-master-chassis situation poses no impact after chassis B restores the configuration file.

4. Prepare scripts used to start interfaces on chassis B.
5. Reset chassis A and run scripts to rapidly enable all external interfaces on chassis B. Traffic switches to chassis B.
   
   

   Services are interrupted in the duration between when the interfaces on chassis B go Up and when services are restored. This is the primary cause for service interruptions in cross-connect protection upgrade. The service interruption time is related to the service volume.

6. After chassis A is reset, it automatically joins the virtual cluster. Services are balanced between the two chassis again.