Nexus 5596 direct upgrade

[Ramirov]

Hi All, 

I have a pair of Nexus 5596 running in vpc with old firmware( 5.1) and there are 6 fex connected to the N5K.  Basically the N5K is only used for basic layer 2 function only (without FC). 

We are planning to upgrade the N5K to the version 7.3 and i have gone through the release note. If follow the release note, we will need to follow below upgrade path. 

5.1(3)N1(1) -> 5.1(3)N2(1) -> 5.2(1)N1(5) -> 7.0(6)N1(1) -> 7.1(4)N1(1) -> 7.3(3)N1(1)

It will take quite some times if i follow upgrade path. I tried googled around and saw someone mention about direct upgrade from version 5.1 to 7.3 to reduce the maintenance windows. 

Please advise if this is possible to do a direct upgrade and is there anything need to take note for this direct upgrade ? 

Somebody has experience with this direct upgrade? 

I am also check that ISSU fails in the STP check because the NX is not a leaf (Its means I have trunks ports) so the path recommended by Cisco will be disruptive anyway.

Thanks

[balaji.bandi]

You need to follow the upgrade path. (it upgrade other upgrade part of step approach), if you do not follow the steps and you do shortcuts, you may see a negative outcome.

for 7.3 release notes you need to have 7.X version :

If you want to upgrade from a release on Cisco NX-OS release 7.1 train or earlier, you must first upgrade to Cisco NX-OS release 7.1(4)N1(1) and then to 7.3(5)N1(1). If you want to upgrade from a release on Cisco NX-OS release 7.2 train or earlier, you must first upgrade to Cisco NX-OS release 7.3(2)N1(1) and then to 7.3(5)N1(1).

https://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus5500/sw/upgrade/735_N1_1/n5500_upgrade_downgrade_735_n1_1.html

[Ramirov]

Thanks Balaji , can you clarify what is a negative outcome?

In documentation you have a section where say 

Upgrading from Cisco NX-OS Release 5.1(3)N2(1c) and Earlier Releases (Disruptive Upgrade)

Why we cannot follow that path?

If I follow the original path I have to do 6 reload to the NX and FEX , I think that is not good.

Thanks.

[balaji.bandi]

You miss the bus when the new release is available if you have upgraded that time, you would not in this situation.

yes you need downtime, but if you are in vPC you can non-disruptive upgrade each side and reload only specific FEX to reload, onl;y if the end device is dual-homed, if they are orphan ports, you need downtime.

That is best approach if we do not follow as Cisco suggested, (risk is down to you, it may break the upgrade and you will have more downtime than planned)

Now its your decision what is the safe approach you like, one we can see upgrade path clear and required more reboots or controlled reboots, one may have a different outcome ( never tried nor i took that risk).

[Ramirov]

Thanks Balaji ,

one question more . For an upgrade of a dual homed FEX cisco recommned the following procedure , Do you know why is necessary pre-provisioning the FEX before start? this imply lost of config?

Step 1

 Configure FEX module pre-provisioning for all the FEXs connected to both the switches (vPC primary and vPC secondary switches).Upgrade the vPC primary switch with the new image using the install all kickstart image system image command. During the upgrade process, the switch is reloaded. When the switch is reloaded, only singled-homed FEXs connected to the switch are reloaded and dual-homed FEXs are not reloaded. Servers connected to the dual-homed FEXs retain network connectivity through the vPC secondary switch.

Step 2

 Verify that the upgrade of the vPC primary switch is completed successfully. At the completion of the upgrade, the vPC primary switch restores vPC peering. However, dual-homed FEXs are connected only to the secondary vPC switch.

Step 3

 Reload the dual-homed FEXs using the reload fex command from the vPC secondary switch. Reload the FEXs one-by-one or in a bunch of two or three FEXs. The servers connected to the dual-homed FEXs will lose connectivity.

Step 4

 Wait for the FEXs to reload. After the reload, the FEXs connect to the upgraded switch (vPC primary switch)

Step 5

 Upgrade the vPC secondary switch with the new image using the install all kickstart image system image command. During the upgrade process, the switch is reloaded. When the switch is reloaded, only singled-homed FEXs connected to the switch are reloaded and dual-homed FEXs are not reloaded.

Step 6

 Verify that the upgrade of the vPC secondary switch is completed successfully. At the completion of the upgrade, the vPC secondary switch restores vPC peering. Dual-homed FEXs connect to both the peer switches and start forwarding traffic.

Thanks.