VMware ESX NIC Failover help requested

[EclipseAgent]

So,

Tomorrow I am going to do some fail over testing and just came across that maybe I'm not sure when a standby nic actually takes precedence over the active nic. Here is my setup

vSwitch

Nic 1 - Active - Switch 1

Nic 2 - Active - Switch 1

Nic 3 - Active - Switch 1

Nic 4 - Standby - Switch 2

Nic 5 - Standby - Switch 2

Nic 6 - Standby - Switch 2

Switch 1 is Active

Switch 2 is a Standby switch

Now, if switch 1 goes down the beacon should see that Nic 1 - 3 are down, and move over to standby correct?

Now what if only Nic 1 goes down? Will it try to just share load on Nic 2 - 3 and if not, will it try to go over Nic 4 which is on a standby switch, and shouldn't be used? Could this potentially cause some arp issues if the Nic 1 goes to Nic 4 which again is on the standby switch? Is this the correct configuration for this type of setup and failover environment?

Basiclly what I want is:

Nic 1 Nic 2 Nic 3 (to fail together within their group of 3) but if all fail (then all fail to nic 4 / 5 / 6).

Thanks any help would be quite helpful

[Texiwill]

Hello,

You have a single vSwitch with 6 pNICs? associated with it?

What happens depends on how the vSwitch is configured, etc. This is not a very good topology as you will be mixing security zones on the same vSwitch. However, can you give us more on how the vSwitch is configured, portgroups, load balancing, beacon monitoring, etc.

Best regards,
Edward L. Haletky
VMware Communities User Moderator
====
Author of the book 'VMWare ESX Server in the Enterprise: Planning and Securing Virtualization Servers', Copyright 2008 Pearson Education.
Blue Gears and SearchVMware Pro Blogs -- Top Virtualization Security Links -- Virtualization Security Round Table Podcast

[EclipseAgent]

Ok.. here's the exact configuration

Single vSwitch w/ 4 pNICs.

2 pNICs Active (on Physical Switch 1)

2 pNICs Standby (on Physical Switch 2)

Network Failover Detection is on Link Status only.

Load Balancing is on Route based on ip hash.

Failback set to Yes

Spanning Tree is configured on the switches.

[RenaudL]

Now, if switch 1 goes down the beacon should see that Nic 1 - 3 are down, and move over to standby correct?

Correct.

Now what if only Nic 1 goes down? Will it try to just share load on Nic 2 - 3 and if not, will it try to go over Nic 4 which is on a standby switch, and shouldn't be used?

You have configured 3 active nics, and ESX will try to keep it that way. This means that if 1 active nic fails, it will pick up the first standby nic to balance traffic (in the order of preference you specified).

Could this potentially cause some arp issues if the Nic 1 goes to Nic 4 which again is on the standby switch? Is this the correct configuration for this type of setup and failover environment?

Whenever a failover occurs, ESX fires gratuitous RARPs which update all physical switches on your broadcast domain with the new topology.

[EclipseAgent]

So one of the stand-by nic's is on a stand-by switch. I'll have to get with my network team to see how the links between the 2 switches are configured, to make sure it'll route while it's in standby mode.

I'm doing some pretty intense testing today and will find out how it works in our environment.. I'll let you know.