Still learning the ins and out of ESX and am having a hard time understand the performance metrics when viewing them through virtual center. When I view one of my hosts running in esx, the cpu usage on the box is low 0-3%, but when I view the CPU usage in virtual center, it shows like 45% usage, the box has 2 virtual cpu's. Any docs you could point me too, or any suggestions i would appreciate.
Is it a single CPU or SMP and if it is SMP was it converted from a non-smp hardware install?
NB. Screen savers will consume a lot of CPU because the VM emulates a video device and a screen saver keeps it busy.
Well... I had this issue on my 3.02 hosts, many of the 2000 vm where eating a lot of cpu while idle... I installed the app and it solve my problem...
I've previously tried to upgrade and reinstall the VMwareTools with no results... always a lot of cpu... this small service ruled...
It may be not relevant... but it worked
Well @ this point I am willing to give it a shot. I currently have a ticket open with VMWARE and have not heard of any other good suggestions. I will attempt this off hours and see how things react on Wednesday. Its really annoying right now since it impact whats DRS thinks it should do in some cases.
I think the answer for this one on ESX 3.5 will be the monitor.idleLoopSpinUS value in the vmx file. monitor.idleLoopSpinUS = 1000
This requires down time as the VM will not use the value utiil it reboots.There are others that have had success with this setting.
So i added the following to the vmx file:
monitor.idleLoopSpinUS = 1000
Still did not appear to fix the problem, but the vmware idler is working perfect. Can I please confirm from somone that the positon of the monitor.idleLoopSpinUS = 1000 in the vmx file matter? Right now I just threw it in @ the bottom of the file.
It's good to hear the the idler is working thanks go to . On another post there was a discussion about the value of the setting, VMware indicates a recommended range of 100 - 2000. The only other possiblity is to lower the value to 100 or possibly slightly lower. But you have the issues resolved already and unless you need to keep 2000 around for a longer term I think I would go with the Idler program.
Going further would be of benifit to the community if you have time to try it.
I will for sure try changing the value, I would like to use that fix instead of the vmware idler program, but for now will stick with vmware idler until I can find a value that fixes the problem. Also just wanted to verify that there isnt a problem with the position of the parameter in the esx file.
Is a nagios agent running on this server?
Last time I had this problem, it was because Nagios client was eating all the processor.
This happened on a 2k3 server but still, uninstalling/reinstalling the agent solved the problem