BladeSystem Virtual Connect
cancel
Showing results for 
Search instead for 
Did you mean: 

Virtual Connect Active\Active and network loops

chuckk281
Trusted Contributor

Virtual Connect Active\Active and network loops

Mike had a customer quesiton:

 

***************

 

Customer just setup VC Flex10 with an Active/Active NIC setup under vSphere 4.1 u1  (Blade Hardware & VC are all up to date)

 

On the ESX hosts, it appears that a loop may have been created on the mgmt network (vmKernel) of the esx hosts since the hosts were dropping out of vCenter on a pretty regular basis (every couple minutes).  I had the customer disable the active/active NIC setup and backed off to only a single active NIC.  This resolved the problem, and the ESX hosts stabilized.

 

In VC Server Profile, we’ve had define two NIC’s for the management network

 

NIC1 – Mgmt1  - Connects to the uplinks off Flex-10 in IC bay 1

NIC2 – Mgmt2 - Connects to the uplinks off Flex-10 in IC bay 2

 

The goal is to have both NIC’s active in ESX with redundancy at the NIC’s only.  We’ve configured both NIC’s as part of the same vSwitch in ESX using a MAC hash for redundancy with both NIC’s active.

 

The Cisco admin here is “new” so I think the issue may be on the network side but I’m not sure what we can check/ ask.  We’re connecting to two older Cisco switches, but I don’t have the models.   

 

Any suggestions would be greatly appreciated.

 

*********************

 

Robert replied:

 

****************

 

There does not appear to be anything wrong with the active/active config you describe.  NIC1 assigned to Mgmt1, and NIC2 assigned to Mgmt2.  I have never seen ESX cause a loop, but the load balancing in the vswitch should be configured for route based on originating port ID, not IP hash.

 

***************

 

Any other help for Mike?

1 REPLY
grzesiczek1
Occasional Advisor

Re: Virtual Connect Active\Active and network loops

Hi,

 

I have very similar environment in my company and its working fine.

 

I would suggest you some steps that might help.

 

0. You enabled aggregation at the VC Uplinks, don't you...??

 

1. Check if all your VC are 3.18 FW

2. Check if you have latest bnx2x and be2net drivers at ESX hosts ->there were a lot of issues with CNA cards and FLEX-10 modules.

3. Make sure your VC modules are connected to different physical switches!!!

4. Change load balancing option to "Port ID"

 

Review VMWare KB 1004048 -> general info about aggregation.

 

Hope it helps

 

BR

Greg