VMware Cloud Community
TimRG
Contributor
Contributor

Very high ESX CPU usage with very low guest CPU usage

Good Morning,

I recently P2V'd a Windows 2000 physical server to an ESX 3.5u2 (110268) host using the VIClient built in enterprise converter. The conversion worked fine and the guest is now running on the esx host. However, I have run into a problem regarding the CPU usage on the ESX host... let me explain.

If you logon to the guest console then you can see the system is idling with 1-10% cpu usage as far as Windows is concerned, yet in the VIClient I can see that ESX us churning away on this guest at about 4-5ghz! The guest doesn't run slow and it still does everything it always has (file server), CPU ready time on the guest is averaging around 100ms also. There is only 1 other guest on this ESX host (it's a new box) and it is hardly used.

The host is a HP DL385G1 with 16gb of ram and opteron 2.6ghz CPU's, the old guest physical machine was a HP DL380G3 running Windows 2000. The new guest has been provisioned 2 CPU's (down from 4) and 2GB of RAM (up from 1.5). I have disabled all old HP services etc from the system but like I said, from what I can tell it doesn't appear as though the guest is actually working hard. Latest version of VMTools has been installed and appears to be running fine to my knowledge.

Can anyone point me in the right direction? Anyone experienced this before? I'm guessing this is something to do with the way windows 2000 handles its idle cpu time, but I'm not sure how to approach this.

Thanks in advance!

Regards,

Tim

Reply
0 Kudos
4 Replies
malaysiavm
Expert
Expert

What you can try is go to your VM guest -> right click -> Edit setting -> go to Resources tab -> CPU resources alocation - > Set SHARES to "HIGH".

Malaysia VMware Communities -

Craig vExpert 2009 & 2010 Netapp NCIE, NCDA 8.0.1 Malaysia VMware Communities - http://www.malaysiavm.com
Reply
0 Kudos
TimRG
Contributor
Contributor

Thanks, but changing the CPU resource shares isn't going to help this issue. The shares are only useful when resources are in contention and currently the host still has plenty of capacity left on both CPU and RAM.

Reply
0 Kudos
Named_Jason
Contributor
Contributor

This sounds like a classic example of a HAL mismatch - can you tell us if you downgraded it from 2 physical CPUs to 1 during the P2V process? If so, the following procedure should resolve your issue (with a couple of VM reboots):

Log into the VM as an administrator and do the following (this is from memory, so some of these steps may be a little vague; my apologies):

1) Open up device manager

2) Expand the "Computer" object

3) Right click on "ACPI Multiprocessor PC" and select "Properties"

4) Select the "Driver" tab

5) Click on the "update driver" button

6) Step through the wizard; at every opportunity where windows tries to automate the process, select the most manual option available (ie. select the driver from a list) until you come to a list of drivers

7) Ensure the checkbox for displaying all drivers is enabled and, in the right panel should be an option to use "ACPI Uniprocessor PC". Select this driver and complete the wizard.

😎 You will need to reboot Windows, go ahead and do so.

9) Windows will come back up and need to reboot again. Go ahead and do so.

This will fix the HAL mismatch, which is, in my experience, the most common culprit for the symptoms that you have described.

Reply
0 Kudos
fand
Contributor
Contributor

Hello,

I have the same symptoms with several Windows 2000 vms (25, 50 or even 60% host CPU usage when idle), but they all seem to be configured "right":

- 2 or 4 vCPUs

- ACPI Multiprocessor HAL

- Created using P2V on multiprocessor hardware

On the same ESX farm we have Windows 2003 and Linux multiprocessor machines with normal host CPU usage (~5% when idle).

By "idle" I mean that the Windosw task managar shows <5% cpu usage...

Anyone has a clue ?

Thanks !

Reply
0 Kudos