This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.
You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

Custom report, Historical Average response time not correct

I create a custom report, showing Average response time for nodes last 14 days, filtering the results to only show nodes with 100ms+ responsetime. I discovered that when Im altering the filter, i get som strange results.

For example, setting it to 100ms shows that the node with the highest average responsetime last 14 days has a responsetime of 119ms.

If i set the filter to only show nodes with responsetime over 400ms I get nodes with average responsetime of 2045ms. The same node show an average response time between 1,5ms and 4ms in Orion Web, and that seems right.. Maybe I do something wrong, please help.

 

The custom report has the following settings:

Select fields

Field: Average response time sort: descending function average

Field: Node Name sort: * funktion: *

 Filter Results

Records where Average response time is greater than 100

  • FormerMember
    0 FormerMember

    quick question on this there is another tab where you select how many of these records you will see.


    Do you have it set to show all or just a certain number


    if you set it to a certain number you could be getting truncated responses and that is why the data seems strange.


    As it will start running and find the first <number> that meet the criteria and then stop.


     Might try using top XX or show all feature if this is the case

  •  I have set it to show all records, so I dont think that is the problem..

    Im kind of stuck... 

  • Field: Average response time sort: descending function average


    The filter is applied to the field Average Response Time rather than the output


    Only occasions when the avg response time was measured as above your filter figure are included in the second averaging function

  • FormerMember
    0 FormerMember

    I actually found this same issue when trying to limit the field for CPU_Load


    the problem appears any time you try to filter on an average it seems.


    in my case i was trying to filter for CPU_Load >= 50% and got bogus information just as he has outlined above.


    The report i made was based off the CPU_Load historical report that comes with Orion.  Just add a filter like above and you will see how this causes the error as he is describing.


  • The sql created by the filter is working a little differently than what you might intuitively think.  In your example of cpu load >= 50%, you would expect the query to average all of the cpu load records for a node and then only return nodes where the average cpu load was >= 50%.  What the filter does is select the records for a node where the cpu load measured greater than or equal to 50% and then averages those records together.  

  • Yes, for me its not intuitiv but... Anyway, removed the "show only nodes with 100ms responstime, and I kind of get the results I want, so Im 90% happy.