Invalid configuration for device '4'. ?

[timteller]

I upgraded my ESX 3.0 servers to ESX 3.01 and my VC to 2.0.1 this week. And now when I create a new Virtual Machine. VC errors and gives me a dialog box that reads... Invalid Configuration for device '4'. It doesn't create the VM, or any logs, as though I can tell. The Recent tasks pane in the VC just says the same thing. Did I do something wrong? How I upgraded, I put in the ESX 3.0.1 Cd and rebooted. I chose Upgrade and rebooted when it was finished.

Does anyone know what this means?

Thanks,

Tim

[rthomsen]

Hi Tim

Is the host configured in a HA cluster ? try to disconnect the host from the cluster make a VM and reconf. into the HA cluster.

It should then be possible to make other VM's

Thanks,

Rasmus

[bvig]

I also received the message \[Invalid configuration for device '4']. In our environment, upgraded ESX 3.0 to 3.01, upgraded VC to 2.0.1.

In this instance, we had two esx hosts 3.0.1 in cluster but using locally attached storage (dev, not san attached). I set DRS to "manual" and I could then create new guest machines.

Hope this helps, Brad

[brugh]

i have the same issue... is there a known cause for this? is there a known solution? i can 't imagine the solution is to remove hosts from a cluster everytime you want to add new VMs.

[oskar_berg]

I got the same error. This was because the LUN that I tried to put the virtual machine on wasn't published to all of the ESX-servers (3.01) in the DRS cluster.

Regards

Oskar

[brugh]

aah that would make sence.. i wonder why the errormessage doesn't read "Selected storage not available on all participating hosts" in stead of the "invalid configuration for device 4" error.

than, how do i create new VMs on the host local storage when the host is a drs member?

[hullt]

The only way I have found so far is to deploy from a template. I am new with this as I just installed our first 3 servers this week and our SAN is still in the box, so I am just using local storage for now to test and learn.

Message was edited by:

hullt

[timteller]

What I had to do was remove my clusters and rebuild them from scratch. I think the database got corrupted when upgrading it from 2.0 to 2.0.1.

Did anyone find a better solution?

Tim

[fzumi]

Setting DRS to manual also helped in my case, though this is quite an odd solution. I'm also looking forward to a better one, i.e. the next upgrade or patch.

Message was edited by:

fzumi

[fzumi]

No, sorry, the behaviour is actually not so odd, since locally attached Storage cannot be handled properly by DRS.

[virtech]

This is a known bug (128689) , should be fixed in 3.0.2 early next year.

[daniel_uk]

Great...So i cant use HA until 3.02!

[mikehop]

Same issue as above. Recently upgraded to enterprise and enabled HA and DRS. First VM to be created since. I had created a LUN and presented to all Servers in the Cluster. I had, on only one of the cluster hosts refreshed the storage adaptors and added the new storage device. I immediately tried to add a new VM and got the obscure error message "invalid configuration for device 4" after reading this post i went back and refreshed all the storage adapters on each of the other hosts in the cluster and i was then able to add a new VM on the new LUN installed.

Makes sense why it fails in this case, It would make sense also as brugh[/b] posted above to have more realistic sensible error messages. This seams to happen a lot with VMware.

Also would be handy to allow you to set a task to update the configuration options on all host in a cluster at once rather than having to do each one individually especially for thing like storage etc

etc

[ascari]

I have same problem, but in my case i have two different cpu system in my HA lab environment for MSCS testing. When i stop DRS, i can chose my destination physical esx system manually (localstore) and problem is fixed. In a few days i'm waiting a new system identically for cpu at my first esx server. I will try again.

Alberto

[ahoogerhuis]

Just ot add some of my experiences to the pool of knowledge in here; I had a VM that had syntax issues in the .vmx file after an upgrade from 2.5.x and decided to just drop it from inventory, remove all but the .vmdk files and just build a new machine and manually slide in the previous .vmdk files after having a blank machine.

As it turns out, I was unable to add VMs to the DRS cluster in any way from VC, giving the venerable error in the topic. But using the VI client and attach directly to the host I wanted to create the machine on (as I wanted to use local storage) allowed me to create an empty VM and then modify the .vmx files afterward to replace with the previous .vmdk files.

Just my two cents.

-A

[atessiot]

I have had the same trouble. To solve it I find only three ways :

\- Putting my ESX out of the cluster and then create my VMs and then put my ESX into the cluster back.

\- Only create VMs with disks into storage shared by all the ESXs

\- Create my VMs only on the storage owned by the first ESX of my cluster

[ABK]

I had the same problem in our 3.01 cluster but in my case it was caused by some inconsistencies in the database. To solve the problem I had to select the hosts one by one in Virtual Center and reconfigure them for HA. After that I was able to create new virtual machines without any problems.

/Anders

[icampbell79]

As per others experiences, I had this issue when trying to create a vm no my second host using it's local storage. I had to temporarily disaable DRS so that I could create it.

[alex2801]

same issue when installing a couple of fresh 3.0.2 esx servers. I am migrating physical machines to VM's and i have only 1 connection left on my fibre switch until the physical machines are migrated. So only 1 esx server is connected to the SAN. Setting DRS to manual solved my problem.

[spencerralph]

I have a similar issue with 3.0.2 fresh install but it Device '1' not 4, but i don't have DRS or HA running at present.

Any suggestions.

Spence.