Java (Java Agent, Installation, JVM, and Controller Installation)

Showing results for 
Search instead for 
Did you mean: 

AppD Metric browser results differs(low) compared with Server logs




There was a Load test in the last week,  There are 4 lines in the attached graph, two from each of 2 different JVMs.  One line of each pair is incoming requests and the other is outgoing requests to RISE, an internal DataPower installation.  They should be in about a 1:1 ratio.  According to Tomcat’s access log, there were about 900-1100 incoming calls per minute on each server, which is what I expect based on the load we were throwing at it.  In addition, the DataPower guy said he saw 60k total requests, which sounds about right (2 servers @1000 requests per minute * about 30 minutes.)


However the numbers in AppD are pretty different.  The incoming calls per minute are too low and the outbound (to DataPower) calls are too high.  At first this caused the developers quite a bit of consternation because they couldn’t understand why there were so many backend calls.  However we think the AppD numbers are incorrect. Can anyone let me know the root cause for this?


Please see the attached metric browser screenshot for reference and any responses are highly appreciated.





Thank you,


AppD Metric browser results differs(low) compared with Server logs
AppDynamics Team

Hi Santosh,


Could you please send us details as listed here


1) Zipped agent logs in debug mode. Please find attached screenshot to capture agent logs in debug mode. Please make sure that there is a traffic on the application when debug mode is ON.

2) Access logs to validate what you see with respect to what debug agent logs shows.





Hi Rajesh,


My apolozies for late response, as requested please check the attached most recent agent logs for both the servers and access logs at the timeframe the issue occured. I wonder if I can pull the Debug agent logs for the exact issue time frame but it gives me the most recent? please let me know if you need anything.