Hi all,
I'm doing some tracing and I'm willing to accept higher DT overhead for more detail in my traces.
There are 1000's of classes that I could add custom sensors for...but I'm not yet willing to go down that path, and besides, it sounds so 90's, so Wily IntroScope-ish.
Specifically, I'm trying to find out the name of the java file in our package space (not hibernate or java.sql) that executes a particular SQL statement.
There is a sql statement in yellow with response time of 4457.72ms at the bottom of this screenshot.
It seems like the closest this trace will get me is “com.fisglobal.phx.svc.PhoenixExecutionContext.getService()”.
This getService() method is a generic entrypoint method….its nowhere near the code that executes SQL.
I tried the following setting, and I dialed in the underlined setting, but couldn't see any difference.
Edit System Profile / Agent Mapping / Advanced / Auto Sensors / Add info in the HIGHEST resolution
How do you all approach this?
Thanks,
--Erik
PS, here is the doc for the Auto Sensor setting:
"Adjust the resolution of the additional information that is stored on PurePaths. A higher resolution allows more detailed analysis of PurePaths at the cost of additional data being transferred between Agents and the Server."
Answer by Erik O. ·
Thank Kevin,
It doesn't sound to me that IO activity (100% IO, per the screenshot) will show up on the radar of a CPU profiler....but heck I'll give it a try.
Also, what can I do to verify that the "HIGHEST resolution" setting did anything?
Thanks,
--Erik
Answer by Kevin J. ·
Erik-
Have you tried using CPU Sampling to search for the method you are interested in? Even at the highest granularity, the auto sensors still might not pick up the method you are looking for, if it is not significant enough to overall execution time. The CPU Sampling will show a more comprehensive look into the call stack.
HTH,
Kevin
JANUARY 15, 3:00 PM GMT / 10:00 AM ET