For a Purepath of a method X, response time is 28 min and in its API distribution one of the APIs say JDBC is taking 400 min and others are taking around 20 to 30 min each. The running processes are multi-threaded. This method is apart of BatchJob. A similar 30 methods will be executed for one BatchJob. For that particular day, BatchJob execution too 6.0 hrs. And one of the method's JDBC execution time is 400Min=6.6hrs.
Is this a dyntrace error? Or is it okay to have such a huge difference between response time and API time?
The Response Time of a PurePath and the Execution Time of an API are slightly different metrics.
Response Time is the time it takes to execute transactions and respond the result back to the calling instance. This is the total execution time for the PurePath in milliseconds, excluding asynchronous paths that may take longer independently.
Execution Time also includes any asynchronous executions. So - if a PurePath Entry Point method executes in 1s it would be a 1s Response Time. But - if this PurePath then makes asynchronous calls, e.g: calls to background worker threads - the Execution Time can actually be longer than the Response Time.
Does this apply to your application?
Thanks for ua reply.
But my concern is, the process mention in question is apart of BatchJob. A similar 30 processes will be executed for one BatchJob. For that particular day BatchJob execution too 6.0 hrs. And one of the proccesses JDBC execution time is 400Min=6.6hrs. How is this possible?? Isn't this a dyntrace error?
Since it is non-conventional request (batch job request), I would spend more time understanding the flow of transaction and how the DB connections are made per job and how many queries are executed per job and how the connections are reused across multiple requests.
It will require more rigorous calculations and analysis to reach conclusion if the reported agent time is correct or not.
If you find some anomaly you are sure about, you can open support case.