Maybe someone can answer this. With A VMWare DVSwitch uplink ports spread across VMWare Hosts how can a Guest on Host A communicate through the uplink on Host B during a failure on Host A Uplink?
VMWare DVSwitch Failover
Also is there no Topic of VMWare to post under?
I'm not a VMware expert, but I did manage to pass the VMware Datacenter 6 cert. I'll give answering this a stab. So, in my experience dvswitches really are just a way to make network changes across all of your hosts in a unified way. Without a dvswitch you have to be careful about making VMkernel and Port Group names exactly the same across all hosts in a cluster. It's interesting, if you've never worked with dvswitches before, when you go to make changes it actually asks you through the wizard to select the hosts and network adapters to make the change to. Then it runs out and touches all those hosts for you. This again keeps you from making a typo on a port group and thus breaking vMotion in some cases.
So, in the example of a 2 node host high availability (HA) cluster. Let's say we have one dvswitch and each host is connected with 2 physical NICs. Well, the port groups that each of your VMs (guests) use are present on each of the hosts because you deployed the port group using the dvswitch wizard. Let's say you have a standard port group, "LAN". If host "A" was to go offline, the HA mechanism in VMware would simply vMotion the VM to the functional host and boot the VM. (HA in VMware doesn't keep the VM up through a failure like this) It would then just use the same port group it was using on the host that failed automatically.
Now if Host "A" was to just lose all network connectivity for the VMs port group, that VM effectively can't communicate with the network. It would have to be moved (vMotion) to a different host with a connection. As far as I know, one host can't use NICs from another through a dvswitch. Someone correct me if I am wrong.
I may not be answering your question to your satisfaction here because I'm not sure if you asking a "How Do I do this?" or "How does it work?" So I will try to answer the later because that I what I believe the question points to.
Remember that when you create a Distributed Switch, you essentially are abstracting the hardware hosts' nics--and functions a single switch across all the hosts. If you want to configure failover you'll have to start with loadbalancing and choosing the failover option.
You can configure load balancing and choose the
Use explicit failover orderoption and configure it.
Active uplinks(First choice for failover order)
Standby uplinks(if both Active uplinks have failed, you set up the order here to try next)
Unused uplinks(not used in failover).
With no additional posts, I'm marking this topic as solved. ~RW