vPC Failure Scenario

[james21970]

Hello all,

So i've read this doc so many times trying to sort out a specific part of it, and it is about this statement (i've highligted in red the parts that aren't making any sense, and my comments are in paren in black between the highlighted red):

"

vPC Peer Keepalive Link Failure

The vPC keepalive link carries the heartbeat message between two vPC  peer switches. The failure of the vPC keepalive link alone does not  impact the vPC operation or data forwarding. Although it has no impact  on data forwarding, we recommend that you fix the keepalive as soon as  possible to avoid a double failure scenario that could impact the data  traffic.

When both switches come up together (such as after power gets restored  following a power outage) and only the mgmt/keepalive link fails, (ok, so the KEEPALIVE LINK has failed..got it) the  peers are unreachable. However, all other links, including vPC peer  links, are up (again, got it..keepalive down, peer-link up). In this scenario, reaching the vpc-peers through  keepalives are achieved through keepalive links (WHAT????..it just said the keepalive link was down) while the primary and  secondary role election is established through the vpc-peer link.(ok cool..the primary and secondary roles are established through the peer-link.got it) You  must establish the first keepalive (what do you mean 'the first keepalive'...i've only ever known of one, is there a second, or maybe a third?) for the role election to occur in the  case when a switch comes up and the vPC-peer link is up. (ok...your kidding me right???...it just said the role election is going to take place by using the peer link...Whiskey Tango)

When keepalives fail to reach the peer switches, role election does not  proceed and the primary or secondary role is not established on either  vPC peer switch and all vPC interfaces are kept down on both switches. "

Guys, I can't remember reading something that was more confusing. This is a horribly written paragraph, and if you're the guy at 3 AM tshooting a problem that is specific to this, a statement like the above is almost criminal in how unhelpful it is.

This can be read here:

http://www.cisco.com/en/US/docs/switches/datacenter/nexus5000/sw/operations/n5k_vpc_ops.html#wp425122

Do these docs get QA'd for accuracy?

[Ryan Newell]

Hello,

   I understand how this can be confusing. From what I understand the break down occurs here..

   In this scenario, reaching the vpc-peers through keepalives are achieved through keepalive links while the primary and secondary role election is established through the vpc-peer link. You must establish the first keepalive for the role election to occur in the case when a switch comes up and the vPC-peer link is up.

  In the sentences above they are stating what must occur for vPC peer election to happen.

   It reads like this to me: Keepalive messages are exchanged through keepalive links while the primary and secondary role election is established through the vpc-peer link. Role election can only occur across the vPC-peer link when keepalive messages are exchanged across the keepalive link. When keepalives fail to reach the peer switches because the keepalive link is down, role election does not proceed and the primary or secondary role is not established on either vPC peer switch and all vPC interfaces are kept down on both switches.

Regards,

Ryan

[LearnWithSalman]

A summary of possible VPC failure scenarios: VPC Failure Scenarios: What Can Happen in the VPC Domain?

[paul driver]

Hello
if the keepalive link is down but peer links are up post reload if BOTH switches even though keepalive messages do need to go through keepalive peerlink for vpc-peer connectivity  a role election can still take place via peer links