22 Feb 2022
04:06 AM
- last edited on
23 Feb 2022
01:12 AM
by
MaciejNeumann
Hi
Today around 12:00 CET I received concerns regarding the display of data on several dashboards.
For example some single value displays of builtin:tech.jvm.memory.gc.suspensionTime was returning very bad numbers, but also builtin:synthetic.http.availability.location.totalWoMaintenanceWindow displayed percentages in the thousands.
When I looked up the JVMs in the process-group alerting high suspension time, nothing of concern was found. I checked the settings for the values displayed on the dashboard in Data Explorer and noticed that Fold Transformation was set to auto, and it seemed to choose count as the method. I correcred the fold to last value, and received proper numbers reflecting the current state of the JVM GC.
The same was the case with the HTTP monitors availability where I corrected Fold Transformation from auto to average.
Have anyone experienced anything like this, that Fold Transformation change with no input and mess with dashboards display of data?
I spoke with the chat about the issue, who agreed that it did sound weird.
Edit:
I seem to be able to recreate the issue. When I edit displayed data in Data Explorer, the initial Fold Transformation setting is default (I assume) set to Auto. When I then change the Fold Transformation from auto to another option, and then back to auto it mess up the (expected) displayed value.
Best regards
Hello @rsmsdk ,
in recent versions the :fold transformation accepts also desired aggregation. Without it, I guess it uses the default transformation for the metric and you are getting bad numbers.
Hi
Thank you for the input. Chat support just confirmed that there is an internal ticket registered on this issue.
I seem to be able to recreate the issue. When I edit displayed data in Data Explorer, the initial fold transformation setting is default set to auto (I assume). If I then change the fold transformation from auto to another option, and then back to auto it mess up the (expected) displayed value.