What data is an alert that uses Volume -> "Disk Seconds Per Transfer" triggering from?

tigger2

Background:

Using NPM v12.1

- We have an alert that triggers off of this condition:

     Volume -> "Disk Seconds Per Transfer" >= 0.20

- In the Orion UI under Settings -> Orion Polling settings it has these two settings for polling Volume data:

    "DEFAULT VOLUME POLL INTERVAL"

     "DEFAULT VOLUME STATISTICS POLL INTERVAL"

Question: When the alerting rule runs and uses this metric to alert on, is the Volume -> "Disk Seconds Per Transfer" triggering off of the data collected during the Default polling interval (the "most up to date" data) , or the statistics interval ("summarized/calculated" data)? 

Note: In the Orion UI, there is a resource for "Avg Disk Seconds per transfer" which I believe is the summarized data, and I can't seem to find a way to chart the raw data collected, so it seems to reason the alert can *only* trigger off of summarized data as this may only exist as a summarized metric (it seems to be a summation of reads/writes, but I'm not sure if Orion is doing this calculation or if it's occurring on the Windows system being polled and Orion pulls the metric at the polling interval and not the statistics interval),  however, in my alerting rule I have it set to report on the value it alerted on via this text: "${NodeName} Disk Transfer Latency is ${DiskTransfer}", and it seems that this variable may be pointing to non-summarized data (or maybe there is rounding occurring somewhere)?

mesverrum

The first value you referenced, the volume poll interval, is basically just and up down status check and happens every 2 min by default.  No other metrics are collected at that time.

The statistics polls are every 15 min by default and this is where you would be getting things like the disk space used and disk sec / transfer.  This is a single, moment in time value, based on what WMI is reporting at when you poll it.  The statistic is not summarized or calculated and Orion does not do any clean up on it, it's exactly what you would find in perfmon

What are you looking at that leads you to believe you have rounding/summarization happening?  If it is a chart on the volume details page then that very well could be the case since the charts summarize their displayed data based on the intervals and time ranges you set up.  To have maximum granularity you should have it like this:

Increasing the zoom range to more than 24 hours will cause it to summarize because the default chart can't fit that many datapoints in, and if you load more than a month worth of data the smallest sample interval will be an hourly summary.  Just for demonstration's sake I put the variable from your alert into the subtitle and you can see it loaded the current value of .000678.  When I hover over my datapoints it shows the 15 minute time period and it rounds the mouseover values to 2 digits (.00 seconds) but the chart still reflects the actual relative change of the values even though they are ridiculously small in this case.

-Marc Netterfield

    Loop1 Systems: SolarWinds Training and Professional Services

•                 LinkedIN: Loop1 Systems
•                 Facebook: Loop1 Systems
•                 Twitter: @Loop1Systems