Powering on single VM slows down entire VM Network

Kasher_Khan

Limp Gawd
Joined
Nov 18, 2005
Messages
297
It's hard for me to describe where to start with this issue. But we have a small setup of two Dell Hosts each running Vmware ESXi 4.1. There is a physical dell server that is the vcenter server for this environment. It is running Essentials Plus bundle. We no longer have support license from vmware for this environment. The servers have local storage as well as a Dell Powervault MD3000i 5TB SMB grade SAN. This is a production enthronement that was setup long ago but has not been touched in a while.

The servers each have two NICS.One nic is assinged to iSCSI network and the other is where all the vmkernel and virtual machines reside.

Now for the issue at hand. When I power on a speacific VM (it is a file server running server 2003) it crashes the whole vmnetwork and brings it to a crawl. There is no one using the vm so it's not a load issue. The ping times go from 5ms to 2500+ms to any of the vms on the same network. I have tried removing the vnic from the server. It still does the same thing. As long as the vm in question is powered on, it sucks up all the bandwith on the network regardless of whether or net the Guest OS has a network connection. I have tried to restore from my backups but still the same issue so its not a virus or something like that. Definetly seems like a vmware related problem. There is some sort of broadcast or MAC issue that I can't figure out.

When the vm is powered off, everything is fine. I can probably rebuild the vm in question but i am trying to get to the bottom of this to find out of there is a bigger problem here.

I'd appericiate any pointers.
 
Just one NIC in that VM? Sure it's not two that are bridging?

I am not sure I understand the point. Could you please explain. The problem vm has one vnic just like all my vms. all the vms on the vlan have access to one physical nic on the server and the same vm port group. there is a physical nic on the server dedicated to iscsi traffic.
 
What's the load on the array show during the boot (look at esxtop, hit 'u' for the disk sections).
 
What's the load on the array show during the boot (look at esxtop, hit 'u' for the disk sections).

Not very good at making sense of esxtop but here is the output. Again the after powered on shots really are when I enable networking within the guest OS i.e when things super slow down.

Here is capture for before vm powered on



This is after vm is powered on



Network stats after vm powered on

 
Try re-installing VMWare tools. (If you can.) What is the hardware version?
 
I've seen this happen on non-VM systems when the OS was infected. Turning on the infected machine essentially shut every other machine off the network. Just a thought.
 
Back
Top