I am trying to configure double sided vPC from a pair of N5K (N5020 with NX-OS version 4.1(3)N2(1)) to a pair of Nexus 7K in a double sided or back-to-back vPC fashion.
For some reasons, whenever I complete the config and apply, everything seem fine and after 10mins or so one of the N5K falls apart, reboots and Bridge Assurance kicks in and starts blocking ports.
I have this same setup and configuration on some other locations which works fine.
The only difference is that the NX-OS version on the other sites is different from this particular site running 4.1(3)N2(1).
Can someone tell me if its not possible to run double-sided (back-to-back) vPC from the N5Ks to the N7Ks when running on this version 4.1(3)N2(1) and the only solution is to upgrade to the latest code?
Thanks for any feedback.
AFAIK double-sided vPC has always been supported. I would look into upgrading to a more recent version of NX-OS, as that version is really old. I have a few customers right now running this same scenerio and I have moved them to 5.1 and 5.2 trains. If you want to send me your configurations I can take a look and see if I notice anything off, but I would imagine this will be resolved with an upgrade if you have the exact same setup working on a later version of NX-OS.
Thanks for your feedback. I actually double checked and can see that the other site where the double-sided vPC work is running the same version of code of 4.1(3)N2(1).
I will look to upgrade this site though to the latest version. Is there anything I need to be aware of from moving from version 4.1(3)N2 to 5.2.1.N1.4? Can I upgrade direct or do I have to upgrade to a previous code before upgrading to the latest one?
Unfortunately its not possible to directly upgrade. From looking into a bit here is an upgrade path option (there are others if you want to investigate):
4.1(3)N2(1) to 4.2(1)N1(1) - Disruptive Upgrade
4.2(1)N1(1) to 5.0(3)N1(1c) - ISSU
5.0(3)N1(1c) to 5.2(1)N1(4) - ISSU
Here is some documentation on upgrading:
Some Caveats on ISSU:
- Cannot be STP root
- No L3 card, which the N5020 doesnt support anyway.
- No fast timers for LACP
May I add this:
If you want to keep the downtime as little as possible, verify the master/slave state of your node regarding your FEX.
If you have a host dual homed on two FEXs and FEXs have the same master, you will experience a reboot of your FEXs at the same time.