We have a simple PowerShell process count monitor setup and polling successfully. When we view the data via Orion it displays accurate values.
However if we add that metric to Perfstack we display much higher values, until we drill into a more detailed view.
Here is the 12 hour view of the data, where we are going to zoom in on a subset:
Notice the Process Count Monitor values are around 201 for one server and 186 for the other... now look at the zoomed in values:
Notice here the Peak CPU and Peak Memory are still in line, but the process count values from the Powershell Monitor statistic are now down around 67 and 62...
We are trying to understand why this is happening, and whether there is a setting for this monitor to correct it?
This is how the monitor is configured:
Any insights into how to address this are welcome!
Solved! Go to Solution.
Another followup on this, the fine folks at SolarWinds are still looking into this, it may possibly be fixed in a newer release than we are currently on, we have that upgrade planned for this fall, were waiting for an upgrade to our vSphere environment to 6.5+ which just happened. Another note on this (we will call it a bug..) is that it produces skewed results as your monitoring the Perfstack that are even higher, but when you refresh they are normalized:
and this is after refreshing the browser page or re-opening the same data:
I am assuming this is related to the same issue that is driving these counts so much higher in this view...
For reference a support case #00349266 Perfstack Statistic Count was opened in response to this question. In the end the answer was that the formulas mentioned above are static with no option to select which one. Since this technically makes display of many types of Powershell Monitor statistics appear inflated, I am not sure if this is something to file as a bug or submit as a feature request?
It seems like the configuration screen for the powershell monitor where the powershell script is being defined would need to be modified to allow selection of what type of a statistic is being captured?
bmrad is there any way you can review this and give a quick opinion on which side of the line this falls on? Bug or Feature Request? (Thanks!)
Following up on this... our admins opened a support case on this. The initial response is that the behavior is believed to be tied to Transform Algorithms applied to the metrics when adding them to Perfstack. They provided an interesting word document with the different Transform Algorithm Descriptions used by Perfstack, which includes the Mathematical Formulas used. These include: Change Point,Difference, Linear Regression, Normalized, Percentile Standardized, Smoothing, Smoothing and Standardized, Standardized. The response also indicated that they thought which algorithm is applied could be configured, but we are waiting on more details.
SolarWinds solutions are rooted in our deep connection to our user base in the THWACK® online community. More than 150,000 members are here to solve problems, share technology and best practices, and directly contribute to our product development process.