I figured it out. When I removed %F and %L from log4j.appender.R.layout.ConversionPattern in my log4j configuration file, the system no longer calls LocationInfo - I guess that makes sense.
Still, it's surpsising that PrintStackTrace on Solaris would take that long - we didn't see anything like that while running on Win2K. Ilya At 11:15 AM 1/11/2002 -0500, you wrote: >Hi, >I am running into a performance problem running log4j 1.1.3 on Sun Solaris >2.280 (SunOS 5.8). I tried it with two VMs: >Solaris VM (build Solaris_JDK_1.2.2_05a, native threads, sunwjit ) and >Classic VM (build JDK-1.2.2_006, green threads, sunwjit) > >What I am seeing is that my category.info() calls take a long time - from >1/8 second to 1/2 second each. I am attaching a couple of screenshots from >a performance analyzer. Looking at the code, >LoggingEvent: 133 calls locationInfo = new LocationInfo(new >Throwable(), fqnOfCategoryClass); >and LocationInfo then calls t.printStackTrace(pw); > > From what I can tell by looking at the performance analyzer, the last > printStackTrace is what's taking up all the time. I suppose it's really > the JVM's fault. > > >I'd appreciate any pointers on how to fix it or a way to get around it. > >Thank you, > >Ilya Tsymbal > > > > > > >-- >To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> >For additional commands, e-mail: <mailto:[EMAIL PROTECTED]> -- To unsubscribe, e-mail: <mailto:[EMAIL PROTECTED]> For additional commands, e-mail: <mailto:[EMAIL PROTECTED]>