How to reload primary nexus device in vPC scene ?

Rps-Cheers · ‎09-30-2019

Hi everyone, I need your help. we want to reload the primary and secondary nexus device in vPc scene,and it is best to avoid business interruptions. There are my plans: 1\ I plan to shutdown all vPc member ports on primary vPC nexus deivce. 2\ reload the primary device. 3\ after primary device operation normal,i will open the vpc member ports on primary one,and shut all vpc member ports on secondary. 4\ finally, i will reload the secondary device. How about this plan, or do you have some suggestions? thanks

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Rps-Cheers | If it solves your problem, please mark as answer. Thanks !

Alex.M · ‎10-01-2019

Hi SYGY,

Thanks for reaching out on Cisco's Community forums.

It's to my understanding you are wanting to avoid business disruption altogether when reloading a pair of N3k's that belong to the same vPC domain; however, downtime is to be expected without the help of a secondary supervisor module installed on both switches (as possible in the 5k's, 7k's, and 9k's).

Business disruption can be minimized if conducted in a maintenance window with the plan you mentioned in your original unedited post. If in your network the operational primary switch could be hosted by either switch without a negative impact to the environment, it would be best to start the upgrade process with the secondary switch. This will remove a vPC election from the reload operation bringing the total number of elections from two to one. To add on, using these settings will help smooth out the moments during the promotion of the secondary switch to operational primary:

Delay restore (default)
Graceful consistency check (default)
ARP synchronization
Auto-recovery, and auto-recovery reload delay
Peer gateway

Regards,

Alex

Rps-Cheers · ‎10-01-2019

HI Alex,

Thanks for your response.
As your mention，business disruption is expected。So can we think if we shutdown all downstream ports first on primary n3k, how to be transit the downstream traffric ? Whether the downstram traffic will be interrupted ?

Because when we operate, we encounter a situation. When all the downstream ports of the primary device are shut down (the uplink port, peer link, and peer keepalive are not shutdown), the downstream server traffic is unreachable.Is this operation normal?
If we only need to restart this primary device, is there a way to avoid traffic interruption ?

Regards,

SYGY

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Rps-Cheers | If it solves your problem, please mark as answer. Thanks !

Alex.M · ‎10-01-2019

Hi SYGY,

I'm having difficulty understanding your second paragraph. Do you mind rephrasing your thought? A diagram of your use-case would be helpful as well in understanding the premise of your situation.

As for your question "If we only need to restart this primary device, is there a way to avoid traffic interruption?", only mitigation to traffic interruption is possible on the n3ks. A complete avoidance of traffic interruption is not possible on this series of nexus devices. Reload of software in a stateful manner can however be accomplished on certain builds of the n5k, n7k, and n9k series. I would talk to your Cisco accounts manager if your network would benefit from improved high availability features as well as others like VDC, modularity, ACI compatibility, etc.

Regards,

Alex

Rps-Cheers · ‎10-01-2019

So, if i shutdown all downstream ports on primary, there will be a traffic interrupt,Is it right? Thanks for you answer.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Rps-Cheers | If it solves your problem, please mark as answer. Thanks !