On Thu, Mar 24, 2016 at 9:43 PM, David Lang <[email protected]> wrote: > it looks like adding dyn_stats calls makes a small, but noticable > difference in my throughput. The documentation for dyn_stats says that it > tries to get a lock, but if it can't it doesn't wait and retry, instead it > just throws away the update. > > This makes me thing that the dyn_stats stuff (and probably pstats as well) > is built on the wrong model. > > Instead of having a master variable that gets updated from multiple > threads, requiring a lock, we should have thread-local variables that can > be updated without locks, and when pstats runs, it pulls the data (flagging > it to be cleared if appropriate). > > This adds more overhead when spitting out the stats, but it reduces the > overhead in the much more common fast path while we are processing messages.
FWIW, that seems like a good approach. -peter > > > David Lang > _______________________________________________ > rsyslog mailing list > http://lists.adiscon.net/mailman/listinfo/rsyslog > http://www.rsyslog.com/professional-services/ > What's up with rsyslog? Follow https://twitter.com/rgerhards > NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad > of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you > DON'T LIKE THAT. > _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog http://www.rsyslog.com/professional-services/ What's up with rsyslog? Follow https://twitter.com/rgerhards NOTE WELL: This is a PUBLIC mailing list, posts are ARCHIVED by a myriad of sites beyond our control. PLEASE UNSUBSCRIBE and DO NOT POST if you DON'T LIKE THAT.
