9600 switch failover

[carl.townshend]

Hi All

I would like to fail the active sup in chassis 1 to the standby sup in chassis 2.

The 9600s have quad sup and the secondary is in RPR mode.

If I type redudancy force-switchover, what happens to the active sup in chassis 1? does it reload and does the rpr sup in the chassis 1 take over as the standby sup in chassis 1?

Also, when we fail over, I assume the traffic in chassis 1 should not be affected ?

Many thanks

[Leo Laohoo]

@carl.townshend 

Is this the same VSS pair with a critical memory leak? 

In ideal situation, particularly when the memory utilization is low, the command "redundancy forced-switchover" will cause the Active chassis to handover all control- and data-plane over to the Secondary chassis.  Once this is completed, the entire chassis will reboot and the Secondary chassis will get promoted to "Active" chassis. The handover of the control- and data-plane, in ideal situation, can take a few seconds to a few minutes (depending on multiple factors such as memory- and CPU utilization, firmware bugs, etc).  

No one has ever witnessed performing a handover while the Active chassis has a critical memory leak.  No one has ever documented what the Active chassis is going to do.  Does it fail over "nicely" or "rough as guts".  Does it take minutes or hours to failover?  Does it even fail over cleanly?  

My recommendation is to engage TAC before doing the failover so as to ensure everything is going smoothly.  

[carl.townshend]

Hi Leo

Yes it is the same switch.

What would happen in an "ideal" scenario?

If I do a redundancy force-switchover, does the sup in chassis 1 reboot and the second up running as rpr in chassis 1 complete booting and take over?

Am I right in saying even in a good scenario, there will be loss of data plane forwarding in chassis 1 whilst the sup reboots and the rpr sup boots up?

cheers

[Leo Laohoo]

---

@carl.townshend wrote:

What would happen in an "ideal" scenario?

If I do a redundancy force-switchover, does the sup in chassis 1 reboot and the second up running as rpr in chassis 1 complete booting and take over?

---

Let us assume Chassis #1 is Active and Chassis #2 is Secondary/Stand-by. 

1. Chassis 1 will hand-over all control- and data-plane to Chassis 2.  
2. Chassis 1 will reboot.  
3. Chassis 2 gets promoted into "Active". 
4. Chassis 1 boots up and becomes "Secondary" or Standby. 

---

@carl.townshend wrote:

Am I right in saying even in a good scenario, there will be loss of data plane forwarding in chassis 1 whilst the sup reboots and the rpr sup boots up?

---

Always assume the worse.  In any given day, there will be a slight loss, like up to three pings.  

[carl.townshend]

Hi Leo

When you say chassis 1 reboots, you mean the current active supervisor, will the second sup in chassis 1 not take over, which is running RPR whilst the other sup reboots ?

[Leo Laohoo]

---

@carl.townshend wrote:

When you say chassis 1 reboots, you mean the current active supervisor, will the second sup in chassis 1 not take over, which is running RPR whilst the other sup reboots ?

---

Everything in that chassis reboots.  Every line card and every supervisor cards.  EVERYTHING.

[MHM Cisco World]

 

This from white paper of stackwise virtual' you need to read it

MHM

[Mancunian]

Hello,

Look for sections on redundancy mode.

https://www.cisco.com/c/en/us/td/docs/switches/lan/catalyst9600/software/release/17-11/configuration_guide/ha/b_1711_ha_9600_cg/quad_supervisor_with_route_processor_redundancy.html#id_127643

redudancy force-switchover command forces a failover from the current active supervisor (in Chassis 1) to the standby supervisor (in Chassis 2)
The active supervisor in Chassis 1 will reload, as RPR mode requires a reboot before it can become functional again.
The RPR standby supervisor in Chassis 1 will remain in its RPR standby state and will not take over.
And after the switchover
Chassis 2’s standby supervisor becomes the new active supervisor.
The reloading active supervisor in Chassis 1 will eventually come up as an RPR standby supervisor.
During a switchover, traffic handled by Chassis 1 is likely to experience disruption because the RPR standby supervisor in Chassis 1 cannot take over traffic immediately. This is a limitation of RPR mode—it does not support stateful switchover (SSO), so there’s no immediate failover for traffic within the same chassis.

[Leo Laohoo]

@carl.townshend

I would like to also highlight FN74222 - Full or Partial Cisco 9800 Series Wireless Controller Configuration Loss after High-Availability Stateful Switchover Failover (CSCwj73634) because the replication manager is present when VSS is enabled.  If this can happen to a 9800, it can also happen to 9400/9500 & 9600.