I am attempting to profile my code, and I'm unable to interpret the results. So first, the meta-question is, what's the best forum to use to request help with such things?
The code I'm profiling is the server side of a moderately complex webapp. For simplicity, I've wrapped it with net/http/cgi and am calling it from the command line as a CGI "script". As such, it is handling a single request with no actual network traffic. Also, I've disabled the garbage collector. I'm calling pprof.StartCPUProfile at the top of ServeHTTP. The pprof output is at https://scholacantorum.org/pprof.svg It shows three different call chains. The one that actually includes my code, starting at runtime.main, is only 6% of the total CPU samples. There are two other call chains that are entirely within the runtime package, one of them taking 17% of the time and the other taking a whopping 70% of the time, and I can't figure out what they're doing. I would appreciate any guidance. The chain that is taking 70% of the time is gcBgMarkWorker systemstack gcBgMarkWorker.func2 gcDrain markroot markroot.func1 suspendG Underneath suspendG it splits into nanotime, osyield, procyield, and preemptM, the latter of which has signalM, tgkill, and getpid children. getpid alone is shown as taking 4% of the CPU of the entire program! What the heck is this call chain doing, with nearly three quarters of my time? The chain that is taking 17% of the time is mcall park_m schedule findrunnable stopm mPark notesleep futexsleep futex Again I have no clue what this is doing. Any guidance is welcome, including pointers to other forums better suited for such questions. Regards, Steve -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/golang-nuts/CAAnpqKF4QXiL6fhOswgyg3xrE99HLnqQTSW8z3u70mes15suSw%40mail.gmail.com.
