So I have this host who's stress badge is 315. Now that is high if you ask me ![]()
It may have something to do with me experimenting with profiles but still I really see nothing that contributes to it.
There are no undersized machines on this host. CPU Usage and demand is 1/3 of the total CPU.The effective demand is 72% and the overcommitted allocation is 163%.
My ratio vCPU:pCPU ratio is may be a wee bit high but I don't seem to have high ready times.
So how does it get to this value?
Cheers
from the help-
"The Stress score indicates the historic workload of the selected object. While the Workload score shows a snapshot of the current resource usage, the Stress score analyses the resource usage data for a longer period. The Stress score is calculated as a ratio between the demand for resources and the usable capacity for a certain period.
The Stress score helps you identify hosts and virtual machines that do not have enough resources allocated, or hosts that are running too many virtual machines. A high Stress score does not imply a current performance problem, but highlights potential for future performance problems."
check your related settings (item 4c in "Configuration" -> "Edit Policy"), see if those are off (make sure to do this on the correct policy, which is applied to that host - this can be see in the Dashboard when the host is selected)
Once/if those are changed, it takes a while till you see the change - at least a day or more.
Value does not seem to alter after leaving it for several days. I think stress is not clearing for some reason as I have moved several VM off the host and I think the capacity remained unchanged too.
I also have a VM that had a runaway process which caused to stress the CPU. The process was killed several days ago but stress value remains the same and I am not able to kill alarm either.
Cheers
Hi,
I wrote this a while back on Stress :
http://www.virtualclouds.co.za/?p=544
http://www.virtualclouds.co.za/?p=510
Hope that helps you out.
Great articles Hugo.
I have a few more questions though. I must be getting something wrong with my settings.
Can you please correct my assumptions?
Under 3A - Capacity and Time remaining.
Demand or Allocation by Compute allows me to specify whether I want to use demand, allocation or both.
I have unchecked "by Allocation" all together so I assume my numbers are purely on demand.
It is referring to 3C Usage Calculation. Am I correct in assuming that the Allocation overcommit ratios are irrelevant as long as I have "by allocation" disabled in 3A?
It is also referring to 4C Underuse and Stress. Under "containers have stress" it will calculate based on the option you selected under 3A, thus either demand, allocation or both. Correct?
Also under this section there is a checkbox "Consider allocation in stress allocation" which is checked in my case. Does this actually take allocation into account when "by allocation" is unchecked under 3A?
Somehow I have a feeling that my stress result takes allocation into account even though I am trying not to. I have a host which is giving me a 107% Memory allocation stress. Yes it is overcommitted but based on my 4C Underuse and Stress "containers have stress" setting , and assuming that it only takes demand into account, there should be no memory stress.
Is my assumption correct that if one would specify "by allocation" settings you would see stress as soon as there is any overcommitment at all?
