2016-03-23 19:03 GMT+01:00 David Lang <[email protected]>: > On Fri, 18 Mar 2016, Rainer Gerhards wrote: > >> >> One thing that triggers my mind is the change of hash function. In >> 8.17, we use a different hash function inside libfastjson. This has >> prooven to be much faster in experiments we carried out, but maybe >> it's not the case for you. You can comment out line 1640 >> >> https://github.com/rsyslog/rsyslog/blob/master/tools/rsyslogd.c#L1640 >> >> here and see what happens. > > > that does not appear to have made any noticable difference
OK, that's more along the lines of what I thought. Could we try to identify the bottleneck, first? I think performance profiler data would be great. Could you run $ valgrind --tool=cachegrind rsyslogd -n ... on a test machine (or quickly in a terminal session on the prod machine). A short run handling around 100k...200k messages should be fully sufficient. It'll write a cachegrind.out.<pid> file, which you can read for example with the kcachegrind gui tool. That will tell where the hotspots are. Can you send me the cachegrind file plus the git head hash from that you build (so that I get correct source listings). That won't get me to the point where I know sufficient details, but it'll get me started to think about other steps. I think the next thing to do is most probably do a testrun with json-c instead of libfastjson, just to verify. But let's decide this once we see the baseline profile data. Thanks, Rainer PS: in Germany, a long easter holiday starts tomorrow end stops next Tuesday (so fr to mon is holiday). I'll be traveling with family, but I'll have access to sufficiently good equipment to look into this issue. _______________________________________________ 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.
