Supervisor Redundancy

On modular switches with redundant supervisor modules, control of the switch can be transferred to the standby supervisor to minimize downtime and data loss in the case of a reset, reload, or failure of the active supervisor. How the switchover takes place is determined by the redundancy protocol configured on the active supervisor.

To display the state and the current redundancy protocol of both supervisors, use the show redundancy status command. To display the state of configuration file synchronization between the supervisors, use the show redundancy file-replication command.

Redundancy Supervisor Protocols

There are three available supervisor redundancy protocols.

Route Processor Redundancy (RPR)

The default redundancy protocol is Route Processor Redundancy (RPR), which synchronizes startup-config files between the supervisor modules and partially boots the standby supervisor to a standby warm state, but does not synchronize running-config. If the active supervisor fails, or a manual switchover is initiated with the redundancy manual switchover command, the standby supervisor will become active. Running state, including spanning tree, is lost, and all links are temporarily brought down.

Under RPR, the CLI of the standby supervisor can be accessed by SSH or through the console port, but the available command set is limited. Any configuration changes made to the standby supervisor will be lost when the supervisor reboots.

Stateful Switchover (SSO)

In Stateful Switchover (SSO) protocol, the switch synchronizes both startup-config and running-config files between the supervisor modules and fully boots the standby module to a standby hot state to speed the switchover process and minimize packet loss. If the active supervisor fails, or a manual switchover is initiated, the standby supervisor immediately becomes active, and L2 running state is maintained. An SSO switchover is largely transparent from the outside, but because L3 state is not synchronized the switchover can result in traffic loss for traffic forwarded on routes learned by a dynamic routing protocol. Enabling nonstop forwarding can eliminate most packet loss for BGP and OSPF.

Under SSO, the CLI of the standby supervisor can be accessed only through the console port, and the command set is limited. Any configuration changes made on the standby supervisor will be lost when the supervisor reboots.

Note: When upgrading the EOS on a dual-supervisor switch to an SSO-capable version (4.11.0 or higher) from a version that does not support SSO, both supervisors will reset simultaneously, causing several seconds of system downtime.

Simplex

When the switch is set to simplex protocol, the standby supervisor is disabled and switchover will not occur even if the active supervisor fails. Reloading the active supervisor results in system downtime while the supervisor reboots, and the standby supervisor remains disabled. To transfer control of the switch to the standby supervisor, the redundancy protocol must be changed to RPR or SSO.

Under simplex protocol, the CLI of the disabled supervisor can be accessed only through the console port, and the command set is limited. Any configuration changes made on the standby supervisor will be lost when the supervisor reboots.

Configuring Supervisor Redundancy

The supervisor redundancy protocol is configured using the protocol command in redundancy configuration mode (accessed with the redundancy command).

Changing the redundancy protocol on the active supervisor resets the standby supervisor regardless of redundancy protocol, and executing the write command on the active supervisor synchronizes the startup-config files between supervisors in RPR and SSO modes.

Examples

  • These commands display the current redundancy state of the switch and the most recent file synchronization information.
    switch#show redundancy state
      my state = ACTIVE
    peer state = STANDBY WARM
          Unit = Primary
       Unit ID = 1
    
    Redundancy Protocol (Operational) = Route Processor Redundancy
    Redundancy Protocol (Configured) = Route Processor Redundancy
    Communications = Up
    Ready for switchover
    
      Last switchover time = 7:23:56 ago
    Last switchover reason = Supervisor has control of the active supervisor lock
    switch#show redundancy file-replication
    0 files unsynchronized, 2 files synchronized, 0 files failed, 2 files total.
    
    File                   Status         Last Synchronized
    ---------------------- -------------- -------------------
    file:persist/sys       Synchronized   0:10:04 ago
    flash:startup-config   Synchronized   0:10:04 ago
    switch#
  • These commands set the redundancy protocol for the active supervisor to stateful switchover (SSO).
    switch#config
    switch(config)#redundancy
    switch(config-redundancy)#protocol sso
    Peer supervisor will be restarted.
    switch(config-redundancy)#