Our web application (Java, Spring, Postgres SQL, extJS, etc) is using Dynatrace and recently we had a task to verify the cause in the degradation of the APDEX score.
We understood that the score is being calculated based on the following:
-Load Actions
-XHR Actions
-Errors (JavaScript/HTTP)
XHR Actions
We noticed that since March 2021 there has been some inconsistencies in the XHR graph’s reading. Kindly see screenshot below which shows the XHR graph from November 2019 till now.
Based on the chart is looks like Dynatrace does not detect your user actions as the load dropped since March 2021. You must first find out if you are really capturing the XHR user actions. Look at your sessions if you are really receiving them as expected.
For analyzing Apdex values, check first the errors captured and adjust error detection rules in the application if necessary (some errors do not affect users, so it’s required to ignore them from Apdex calculations). Also, adjust performance metric thresholds for actions where you don’t want the default value.
After going through the recorded user sessions. We’ve found that there were sessions that capture XHR actions and many others that have no XHR action capture.
Could you please confirm that the fact that we are missing those XHR data might as well be a cause to the bad APDEX score? Noting that we have key actions defined and all of them are XHR actions