Hi guys
I have recently been reconfiguring a small VMware implementation, which had been set up extremely badly. The implementation
consisted of two ESX 4.1 hosts (not ESXi), and a virtualized vCenter server, and 7 VM's all Windows server 2003 (both 32 and 64
bit).
I started by taking both ESX hosts comletely down, since we needed to change the physical cabling of the ESX host on the switch
side. I then started the two ESX hosts up, reconfigureded the network on them, and when finished I started up all the virtual
machines including vCenter. Service Console, vMotion, and the VM Network was not moved on the switch, only the back end cabling
which connected to two iSCSI targets (openfiler) were changed to a new vlan.
The vmnic0 to vmnic5 ports were changed and the vswitch configuration changed plus an additional vswitch was added. The first
implementation put for example two physical NIC ports from the same dual port NIC, on the VMnetwork. It was changed so instead of
using both physical ports from the same NIC, to take one port from each two different NIC's, to build some redundancy into the
design, so even if one NIC went down, all services (vm network, service console, and iSCSI port groups) would still on one port on
another NIC.
When we first started all VM's up after the reconfiguration, all seem to be working fine. But then we started to get a few VM's
who couldn't communicate with anything, and soon we had around 50% of the 7 VM's who couldn't connect to anything, including
vCenter. We got some of them fixed by shutting down the affected VM's, removing the virtual NIC adapter, and then adding another
type, for example shifting from E1000 to VMXNET3. A couple of the server we couldn't fix and had to rebuild them. vCenter crashed
down with a bang twice, when we were trying to vMotion another server from one host to the other. It did so twice, and after that
we couldn't even log into it, it keep on standing on "Applying computer setting" for a very long time and didn't get any further,
so we had to rebuild it.
I have just tried to install a new Windows Server 2003 Std x64, which we wanted to use as a template, in case we would have to
re-build more of the VM's. Just made a fresh install of the operating system, then installed VMware tools, and guess what, that
can't communicate at all either.
The weird thing is, other VM's can communicate perfectly alright running on the same host, and the NIC seems to be installed and
working alright on the VM's, the status in Windows says "Connected", they just can't talk to anything.
The problem seems to be on the actual Windows VM's and not the hosts, but what can make the NIC's freak out like that in the
Windows 2003 hosts ?. Can it be something I have done ?, I can't figure out what is going on, and the network is highly
unpredictable right now, we never now when the next VM, suddenly won't communicate any more.
Any help would be very much appreciated.
Kind regards