https://issues.apache.org/bugzilla/show_bug.cgi?id=51047
--- Comment #4 from Bartek <[email protected]> 2011-04-11 10:18:16 EDT --- We observed this issue when doing performance tests with 80-160 tps (and plenty of threads running in parallel). The app was later profiled and lots of threads were waiting on this synchronization. Changing these mechanism helped as the only synchronization that is now on the path is in the appender. As we have more than one appender, log4j now handles the number of threads that we use better - improved thread parallelism is observed :-). AppenderSkeleton doAppend() is synchronized. Removing the old synchronization mechanism from Category will not cause appenders to malfunction no matter how concrete appenders are implemented. That's what source code suggests and that's also what we observed in our tests :). I also logged an improvement related to synchronization in AsyncAppender: https://issues.apache.org/bugzilla/show_bug.cgi?id=51048 -- Configure bugmail: https://issues.apache.org/bugzilla/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
