Starting dotTrace affects service being profiled

Hi,
Using dotTrace 6.2 to proflile a windows Service.  The service will from time to time consume 100% CPU.  We have so far been unable to determine why our service is performing this way.
When starting either a Sampling profiler, or a Timeline profiler on the service, the service immediately drops to 0% CPU, and stays that way!  Every time.
So, unfortunately we cannot profile the service while its at 100% CPU.

My question is, does anyone know what would cause this behavior?  By virtue of attaching jetbrains to the process, what would cause our service to change?

3 comments

Hello Chris,

If you use the ‘Sampling’ profiling mode, then dotTrace stops all threads of your service one by one once in a while in order to take measurements. If your service performs some CPU intensive tasks, during the profiling it has to perform these tasks with pauses and the real CPU usage is therefore lower.


Kind regards,
Maria

0

Thank you for the insight Maria.  That would likely explain what I saw with the Sampling profile.
However, I see the exact same behaviour using the Timeline profile mode too.  Does the Timeline type behave in the same fashion you described too?
If so, is there a profiler type I should be using to not interrupt and capture the service to determine what exactly is causing the high CPU usage?

Thanks!

chris

0

Hello Chris,

How many CPUs (cores) does your computer have? Could you please profile your service using the 'Timeline' profiling method, then save and send us this snapshot? Or upload to our ftp server: ftp://ftp.intellij.net/.uploads/  


Kind regards,
Maria

0

Please sign in to leave a comment.