Knowledge Search


[EX] ERPS and VC Mastership Failover

  [KB31481] Show Article Properties

When a virtual chassis (VC) is connected to Ethernet Ring Protection Switching (ERPS) ring topology via physical interfaces, if the master member to which one of the interfaces physically belongs is switched off, the ERPS may not work properly. 

When the VC master member is switched off, and the interface that belongs to that member is the part of ERPS ring, the ERPS should be re-initialized on other VC members. However, ERPS cannot complete initialization properly and it will stay in INIT state. The other interface may not converge to a proper state and may stay in discarding state, not allowing traffic to be forwarded. This is corrected once the VC master member is switched on and operational.


This behavior is per design of EX switches. 


The basic requirements to have a complete ERPS support in a ring topology and to perform a VC mastership failover are the following:

1. Enable the following on the VC:

set chassis redundancy graceful-switchover
set ethernet-switching-options nonstop-bridging
set system commit synchronize

2. Interfaces of the VC which are part of ERPS link should be configured as aggregated interfaces. Ideally physical interfaces that are part of this aggregated interface should be spread across all members of the VC, including the line cards.

However, this is not a mandatory requirement. If an aggregated interface is configured only on one physical interface the problem will not occur during the failover of the VC. While physical interface is created only if the corresponding line card is present and up, the aggregated interface is not dependent on the physical availability of the member interfaces. During the initialization of ERPS, aggregated interfaces will be present as down, even though the child interface is not present. In this case ERPS will initialize correctly and mark the failed interface as down while the other interface will converge to proper states.

Related Links: