We are doing load testing, pushing the system to its limits. The load exceded what the Dynatrace Agent could support and there was a spike in es „Unrecorded Pure Paths“ and „Early
Discarded PurePaths“ which is understandable.
My question is: Data was recorded in the interval when the agent was overloaded. Can this data be relied on? Is it valid? There are values for:
The data is reliable. There is a specific point where the appMon server decides it is going to start discarding purepaths to save itself from crashing. The early discarded purepath measure keeps track of how many purepaths it drops in that time period. You can chart the number of early discarded purepaths, skipped purepaths, and unrecorded purepaths to keep an eye on your lost purepaths. But, for your collected data, there shouldn't be any issue. We also find it helpful to look at purepath length. that tells us if there is anything in particular being dropped.
I would recommend, if it is possible, to dedicate more RAM and CPU if this continues to be an issue.