Have to E-Fuse Servers after VC Upgrade

[gcj2021]

Greetings - we have a consistent issue on our C7000s that when we upgrade our VCs using VCSU ( i do oe manual and of manual and power off/on VC manually), the ugprade is successful, the VC says it is updated and online, but all of the servers do not see the VC network connection. Sometimes this happens on Bay 2 (B side) and sometimes on Bay 1 (A side). It's just 1 or the other so I typically power off/on B side first, efuse, and then do the remaining VCs.

An e-fuse of every server in the chassis makes this issue resolved, but is a huge issue for us. Anyone have a less intrusive fix?

Blade models are G9s on latest SPP

OA on 4.96 and VCs going to v.85

Enclsoure: BladeSystem c7000 Enclosure G3

Interconnect Bays:

Bay 1: HP VC FlexFabric 10Gb/24-Port Module

Bay 2: HP VC FlexFabric 10Gb/24-Port Module

Bay 5: 5. HP VC Flex-10/10D Module

Bay 6: 5. HP VC Flex-10/10D Module

[Ihaqueit]

After update complete reboot or not?

Is OA see interconnect bays with latest updated version?

Can you check the server profile on VCflex its online or not?

Is VCflex set as redundant which one you first update and which one you first restart?

[gcj2021]

After update complete reboot or not?

I power off the VC and power it back on so it will complete the upgrade - I have them set to manual so I have control of the power off/power on

Is OA see interconnect bays with latest update?

Yes - OA sees the new version. 

Can you check the server profile on VCflex its online or not?

Yes it shows it is online and happy, but at ESXi and Windows it is showing down.

Is VCflex set as redundant which one you first update and which one you first restart?

VCSU points to the OA and pushes the firmware itself - I always reboot the Bay 2 B side VC first. Not sure if that answers your question. 

[Ihaqueit]

When you upgrade or update VCflex, before reboot your all blade shutdown manually or not?

[gcj2021]

All servers are online and operations prior and during the VC upgrade - The way it should work is 1 side is always online so I shouldn't have an outage - So while I reboot VC in bay 2, everything is single legged on VC in bay 1, and then the other way when I rebooted VC in bay 1. End goal is to have everything updated without having to e-fuse anything.

[Ihaqueit]

The first thing, as per your statement vcflex server profile online , but not be able to access network.

In this situation if you have backup of vclex of previous, try to restore and check once.

If restore be careful to shutdown all blades.

If you attach any error or log then will help you more. 

[gcj2021]

No need for a restore - after e-fuse the blades are working fine and my VCs are updated - I want to be able to update the VCs without having to e-fuse each blade after the update. 

[Ihaqueit]

e-fuse you want to reset ?

[gcj2021]

No - I don't want to have to e-fuse my servers for them to see both sides of the network - I believe I should be able to reboot each VC 1 at a time and when it comes back up - the server will see both networks.

[Torsten.]

Not enough details ...

(previous version, driver versions, exact update steps, etc ..)

In case of doubt contact HPE support and provide them a support dump..