Upgrade Nexuses in VPC peering configuration and attached to FEX

[WGL_BK]

Hello all. 

I'm looking at upgrading 4 Nexus 9ks (93180YC-EX)  from 9.3.9 to 10.2.4 code, they're two separate sets of VPC peers at different sites. One peering is connected to a FEX. I've already given up on using ISSU for a smooth upgrade since I don't believe my 9ks have a second supervisor so I'll be looking for a maintenance window for that downtime but for that FEX connection, I wonder what gotchas I should be looking for, if any. Looking for guidance on the matter before I dive into reading NXOS guides again. Thanks for reading.

After further reading, it seems I may be able to proceed with ISSU - I checked the ISSU Support Matrix and found the 9.3.9 to 10.2.4 path. I believe I misunderstood the hardware requirements since I was unable to confirm the presence of two supervisor modules.

Edit: Added upgrade path and reconsideration re: ISSU.

[Georg Pauwen]

Hello,

from which version are you upgrading, and TO which version ?

[WGL_BK]

Oh - apologies. I'm going from 9.3.9 code to 10.2.4. 

[WGL_BK]

Hello all - I would like to revisit this item. I'm still looking out for any guidance that you might provide. I've since tested ISSU on a single unit, however I do still have to contend with vPC peers and FEX connections in the production environment. 

I have since added 92160YC-X units to be upgraded, also in vPC peering.

@Georg Pauwen would you be able to share any advice on this item? 

When I check the impact of the upgrade on vPC peers, it tells me that it'll be disruptive with below additional details.

Service "vpc" in vdc 1: Vpc is enabled, Please make sure both Vpc peer switches have same boot mode using 'show boot mode' and proceed

Service "stp" in vdc 1: Port: port-channel2 in Tree: VLAN0101: Bridge-assurance is enabled.

Please disable BA on the switch and retry.

 (modified the impact to <Hitful>  for module <1>)

Service "hsrp_engine" in vdc 1: Please configure 'hsrp timers extended-hold <> ' based on the control-plane downtime to avoid Hsrp state transitions.

 (modified the impact to <Hitful>  for module <1>)