Virtual machines on a particular subnet/Vswitch not responding

[ajaypr]

The configuration:

We have a remote ESXi server with two VM's running.One VM is running server 2008 Domain controller and the second VM is an XP box.We also have two virtual switches on the ESXi server. The First Vswitch0 is on a 10.X.X.X network which is our data network.This particular Vswitch also has the ESXI managment network connected with a 10.X.X.X ip address.The second Vswitch1 is on a different subnet 192.168.X.X that we use for voice communications.The XP Box has two virtual Nic's one connected to the data network switch Vswitch0 and the other nic to the voice network Vswitch1.

The domain controller is connected to the vswitch0 only (data network).

The problem:The VM's are running properly but after few days, the VM's ( Both Server 2008 & XP box) vNic on the data network stop responding and are unable to communicate via ping etc.Strangely the XP box is still able to communicate only through the voice network.This means that the problem is not on the VM side but more on Vswitch0 configuration.The only way I can fix the issue temporarily is by removing the vNic from the VM setting and readding a new vNic on the VM's. It immedately starts responding once I remove and re add the VM nic.

Not sure what may be causing the issue, but it seems to me that the physical NIC cards the vswitch0 is connected is currently on a trunk port on the physical switch. I am not sure if the voice traffic is flooding the vswitch0. I currently do not see any errors in the windows event log in regard to the NIc problem.

Also I had another question - the Vswitch1 on the voice network just has the virtual machine port configured for that particular voice VLAN. It does not have a VMkernel port assigned with an ip address on the vswitch1. Vswitch1 is directly connected to two different physical Nics. I was wondering do I need to add a VMkernel port for the voice network with a ipaddress of the voice network? I was thinking the VMkernel won't be needed as it is only for managment port/ISCSI/NFS and Vmotion. But I am confused whether this configuration is right?

Would appreciate if you could kindly advice or provide any recommendations for the above issue

Thanking you in advance.

[a_p_]

When this happens, are the VMs - Windows 2008 and Windows XP - able to communicate with each other over the vSwitch? Which virtual NIC did you assign to the VMs (E1000, VMXNET, ...)?

Regarding the vSwitches. You don't assign any IP address to a vSwitch. Only the virtual machines as well as the VMkernel port groups which are connected to the vSwitch need IP addresses for their dedicated use (Management, iSCSI, vMotion, ...).

André

[ajaypr]

Thanks for your help Andre !!! I did confirm that when this problem occurs, the two VM's are able to ping each other's ip address and even the ip address of the Esxi managment port on the same vswitch0, but I cannot ping any device on the 10.X.X.X outside the Esxi server.

If I right click the Server 2008 VM settings - I see the network adapter within the Vmware Vcenter environment as E1000. Within the Windows Server 2008 VM I see the network adapter as Intel pro 1000 MT

On the XP box, within Vmware Vcenter environment with edit settings selected on the VM, it shows Flexible for both Nic's.But Within the XP OS the network adapter for both the Nic's is Vmware accelerated AMD Pcnet adapter.

Thanks once again.

[jamesbowling]

What you are seeing inside the VMs in reference to the NIC's naming is correct based on the adapter types you have assigned.

[a_p_]

From what you say, the VM's as well as the virtual network configuration look ok to me. Did you already double check the physical switch's port configuration. E.g. port security disabled, spanning-tree portfast, ...

André

[ajaypr]

Thanks Andre, I am not an networking guru, but I was able to check on our Adtran Netvanta switch for the settings

The port security is greyed out for those ports as I think the port is currently set on Vlan Trunk. I think the switch allows you to only change the settings for ports where the switch port mode is  set for access.

Also the spanning tree for those particular ports were set as    spanning-tree edgeport. I have a feeling that I would have to change the spanning-tree to spanning-tree portfast trunk.

But again I am not 100% sure, just my thoughts.

[a_p_]

Yes, spanning-tree portfast trunk is the setting you should use.

See http://kb.vmware.com/kb/1004074 for a sample configuration.

André