Wired Intelligent Edge (Campus Switching and Routing)

Reply
New Contributor

2930M Redundant Fault Tolerant Configuration

I am hoping someone can give me some direction... I have 2 TOR Switches that I am wanting to leverage for FT NICs in ESXI6.7 per vswitch. Problem I am having is finding solid configuration information to do this successfully. I am not looking to stack switches only have NIC1 go to one switch and NIC 2 to go to the other and both switched are cabled to a single core switch. I've tried trunking the Core Switch ports but didn't get the result I was looking for down stream and lost NIC connectivity when attempting to simulate a switch port failure. I've been through the switch configuration guide for the 2930M and vSphere networking guide, but am coming up short in nailing down redundant switch specifics. Thanks.

simple-2930M-Redundant.JPG

Super Contributor I

Re: 2930M Redundant Fault Tolerant Configuration

Do you have a network topology diagram?

There are a few options


* Simplest: Use the default settings. No teaming just add two NIC's to the vSwitch and connect the cables to a standard port (not trunk). In case of a failover ESX will send a gARP to make sure the switches will learn the MAC addresses at the new ports
* Static link aggregation: Connect the two ports to a single logical switch and configure the ports as a trunk port (without LACP). Within ESX you have to configure the vSwitch with the correct Load Balancing configuration. See this (old) article but the basics are the same in 6.7. In this article VMware called it LACP but there is no LACP involved with this config https://kb.vmware.com/s/article/1004048?
* Same as the previous option but with LACP. For this option you must run distributed vSwitches

I would say use the first option

Willem Bargeman ACMX#935 | ACCX #822

Please give me kudos if my post was useful!
If your issue is solved mark the post as solution!
Super Contributor I

Re: 2930M Redundant Fault Tolerant Configuration

Sorry, didn't see the diagram. In this setup just use the default settings. Don't use any teaming options


Willem Bargeman ACMX#935 | ACCX #822

Please give me kudos if my post was useful!
If your issue is solved mark the post as solution!
New Contributor

Re: 2930M Redundant Fault Tolerant Configuration

Thanks Willem - Option 1 is the path I originally went down, but with the loss of the uplink from TOR1 to CoreSW (cable got unplugged) we lost all access which I didn't expect. The path through TOR2 never assumed the traffic restoring the cluster to connectivity. So that's why I started looking at how others were doing it and that's been a convoluded mess. If there is nothing I need to do on the switches then I'll have to work backward through the host networking and see if there is something amiss there. Thanks again.

Highlighted
Super Contributor I

Re: 2930M Redundant Fault Tolerant Configuration

Okey that is correct. If the uplink is down of the TOR than some traffic is dropped.

 

What you can do:

  • Make sure the TOR's have a redundant uplink
  • use the feature uplink failure detection

Uplink failure detection example

uplink-failure-detection
uplink-failure-detection track 1 links-to-disable <esx ports> links-to-monitor <uplink>  delay 30

Willem Bargeman ACMX#935 | ACCX #822

Please give me kudos if my post was useful!
If your issue is solved mark the post as solution!
New Contributor

Re: 2930M Redundant Fault Tolerant Configuration

Ah, I've been behind a desk to long... I think you hit the nail on the head with UFD. UFD does require NICs Teamed NFT correct? or will ESXi manage the link state and failover without designating the uplinks as a Team?

New Contributor

Re: 2930M Redundant Fault Tolerant Configuration

If I am reading correctly I can just accept the defaults in the vSwitch for NIC Teaming they appear to be correct for what I need. Thanks for the assist Willem.

Search Airheads
cancel
Showing results for 
Search instead for 
Did you mean: