Adding a VLAN ID causes an outage: C7000, HP 5800-AF-48XG. What's wrong?

[Woter]

Hi,

Firstly, I am very new to blades (and HP switches for that) and have been learning on the job, so please me gental with me.

My aim is to add a new ESX Virtual Machine Port Group with a dedicated VLAN. Outside the world of Blades, this is straight forward in ESX and on the switch. I have to add the VLAN ID to the trunk port (Cisco world).

With Blades thown into the mix, I appear to have to add the new VLAN ID to various places on the Interconnect (HP VC FlexFabric 10Gb/24-Port Module). I have followed the "pattern" for other VLANs and I think its added correctly.

From the Virtual Connect Manager > Connections > Ethernet Networks, I have added two networks:

• CX_NLB_A (SUS1_A)
• CX_NLB_B (SUS1_B)

Both have been assigned VLAN ID: 335.

I have done this for both Interconnect bays.

Under Connections > Server Profiles > {Blade} > Ethernet Adapter Connections, I have added my VLAN to the existing "Multiple Networks" vNet mapping.

On the switch side, we have four interfaces connecting to the VC ports:

• ff-cxenc1-1_X5
• ff-cxenc1-2_X5
• ff-cxenc1-1_X6
• ff-cxenc1-2_X6

I went through each interface expanding the VLAN range to include 335. When I changes ff-cxenc1-2_X6, disaster struck. ESX through a major wobbly, essentially bringing down all guests. Whilst some ESX host and their guests in the second chassis recovered, all those affect by my change did not. As soon as I removed removed 335, service was restored.

On further investigation, in Virtual Connect Manager, under Connections > Ethernet Networks > External Connections > CX_NLB_x > Uplink Ports, I see:

• CX_NLB_A, Port: Bay 1 X5 (cxenc1), Status: LINKED-STANDBY
• CX_NLB_A, Port: Bay 1 X6 (cxenc1), Status: LINKED-STANDBY
• CX_NLB_B, Port: Bay 2 X5 (cxenc1), Status: LINKED-STANDBY
• CX_NLB_B, Port: Bay 2 X6 (cxenc1), Status: LINKED-ACTIVE

So my guess is there's a config error somewhere and as soon as I added the VLAN to the switch interface that was connected to the ACTIVE port, all hell broke loose.

So, my questions are:

1. Is it possible to force failover an Uplink Port for just CX_NLB_x. If so how?

2. How does one add a VLAN without an outage, or what have I not done / done incorrectly?

I'm very grateful for any help or pointers anyone can give.

W.

[Server-Support]

Yes,

From memory, adding / introducing new VLAN configured on the Virtual Connect to the HP Blade servers would cause an outage, because the Server profile will be updated with the new VLAN information from Virtual Connect.

have you set redundant uplinks for the ESXi servers ?

[Woter]

Hi,

Thanks for your reply.

For us it was because we (I) was trying to add the new VLAN to the individual switchports, as opposed to the Agrigate.

Once we added the VLAN to the Agrigate, all was well and it worked without any outage.

Thanks.

W.

[Server-Support]

Ah I see...

So in this case, to avoid outage the VLAN information should only be entered on the Uplink sets only where the server profile is attached ?

is that correct ?

Thanks for sharing.