• Print
  • Share
  • Dark
    Light

Traces

  • Updated on 15 Jun 2018
  • 1 minute to read
  • Contributors

One of the ways we strive to deliver relevant insights is to only collect traces around interesting requests. Some examples of these request types that are interesting would be faster than normal, slower than normal, new web requests, requests introducing new SQL, requests introducing new exceptions, or abnormal satisfaction scores. Under performing requests will be profiled more often than fast requests. As long as the app is getting steady traffic Retrace should be saving a trace every 5 to 15 minutes.

The idea behind a Trace is to highlight problem areas in your application while still giving a general summary on the performance of the average requests of that kind.

Example Trace

Both Prefix and Retrace provide the same type of detailed trace view. This screenshot below is from Prefix.

image.png

Getting more from your Traces

Occasionally you may come across a Trace that doesn't display the information you are looking for or will show "Untracked Application Code" in the trace structure.

Both of these can occur because of how we optimize profiling, but this can be potentially changed by customizing the profiler instrumentation for your language.

Custom instrumentation will allow you to specify what Namespaces and Methods you would like to always be profiled. That way when a web request or transaction occurs, you will have move insight into the methods being called and what your code is doing.