vPC auto-recovery

We have vPC auto-recovery enabled on a pair of Nexus 7Ks. The two switches are in two different computer rooms.

The N7K is in Computer Room 1 (N7K1) was vPC primary and the N7K in Computer Room was vPC secondary (N7K2).

Over the weekend the power to Computer Room 1 was taken down for maintenance and then restored this morning.

As exepected the N7K2 has transitioned to the role of secondary, operational primary.

When N7K1 rebooted it has transitioned to vPC primary, operational secondary.

I have two questions:

1. In order to re-establish N7K1 as vPC primary does it need to be rebooted?

2. Is there are any real reasons to do this if the network is functioning as expected, for example any impact on spanning-tree or HSRP? There will be another power shutdown in a couple of weeks.

 

Thanks

Sean

 

Comments

  • Change the role priority of the secondary (opt primary) to 32768 and then bounce the peer-link. The primary should resume opt primary role.

     

    Best practice says to keep your STP root and primary lined up. Im thinking there are some failure scenarios that will cause issues if they dont I just cant think of them off the top of my head.

    The above steps were from page 34

    http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps9670/design_guide_c07-625857.pdf

     

    You might also check out the 7000 best practices guide for more information.

  • You need to shut/no shut the vPC Peer-Link. Be careful because of the implications of this: when the vPC Peer-Link goes down, the vPC secondary will bring down all the SVIs. This will have impact on HSRP. You may end up with the vPC primary where you want but the HSRP master on the vPC secondary unless you have the preempt option. The best is to have the STP root, HSRP primary and vPC primary in the same box. It's much easier to manage and troubleshoot.

    Antonio Soares, CCIE #18473 (RS/SP/DC)
    [email protected]
    http://www.ccie18473.net

  • Flapping the peer-link will have some pretty serious implications to your production traffic. You'll probably end up with much shorter break if you simply reboot the switch in data room 2, that is currently operational primary. If you have utilized peer-switch feature, your convergence time for the reboot should be pretty close to zero, at least better than flapping the peer-link.

Sign In or Register to comment.