deploying VM from a template causes ESX host to disconnect from VirtualCenter

[Thibault]

This is a new development in our virtual envrionment. It seems when we deploy a VM from a template it causes the ESX server to disconnect from VC. We occasionally can clear the problem via "service mgmt-vmware restart" but not all of the time. We did not put realize the deplyments were a problem until it occured a couple of times. Now we have the problem on a production ESX server with running VMs. We can't vmotion the vm's off to reboot the server to see if that clears the problem. Does anyone have any ideas? Has anyone seen these symptoms before? Thanks

[depping]

I haven't seen these symptoms yet... What kind of storage are you using?

Duncan

VMware Communities User Moderator

-

---

Blogging:

Twitter:

If you find this information useful, please award points for "correct" or "helpful".

[Chamon]

We had this problem when deploying from a template and customizing it during the deployment. It happened often. Can you restart the mgmt-vmware? Or will it now not stop? We had this a few times as well. We had to kill the PID for the hostd and then restart it. We had the issue where we could not kill the hostd PID and had to delete it first. so what point are you in now? In the end we copied a new copy of sysprep in to the VC. If you are having this issue you may want to ensure that the sysprep package is for the correct version (SP1 or SP2) for your template.

if this is happening when you boot the new vm you will have to shut it down first. If the vmware-cmd <path/to/config/file> stop does not work then you will need to kill this PID first

to find the PID for this VM

ps -auxwww |grep <nameOFvm>

then kill the PID

Message was edited by: Chamon

[Thibault]

We are using EMC storage - a combination of clarion and symmetrix.

[Thibault]

We occasionally can restart the agent with mgmt-vmware restart but not always. In one case we were able to kill the hostd pid and restart the agent but not 100% of the time. In those situtations we needed to reboot ESX which meant VM's needed to come down. The problem is occuring with Windows 2003 and Windows 2008 for which the sysprep version does seem to be correct and has been working for a few months. This seems to be a new development and we can't seem to pinpoint what is different. It is very strange. Do you know why an incorrect version of sysprep might hang the VC agent? Thanks

[jbowman12345]

How many NICs are installed in the ESX server? And how are they configured?

[Thibault]

The servers are Dell R900's with 4 internal Broadcomm nics and two pci-e intel dual port nics. The first 2 internal nics are configured for console use and the other 2 for VMotion. We then use one port from each of the intel dual port nics for VM traffic.

[Thibault]

In our environment we have several Virtual Center clusters. In one of our clusters we have a virtualSwitch (for VM's) with a unique name (different than all other clusters). All of the other clusters use virtualswitches all having the same name. It seems when we try to deploy a VM from the template that is local to that odd VirtualSwitch we end up disconnecting the target ESX host from Virtual Center. When we copy the template into the target cluster we want to deploy to and then deploy it seems to pick up the correct VS name and life is good. Is this a bug? Does anyone know how to have a single template that will work regardless of the Virtual Switch names in the cluster? ...thanks