Date: Fri, 24 Feb 2017 16:13:49 +0300 From: Slawa Olhovchenkov <slw@zxy.spb.ru> To: Sepherosa Ziehau <sepherosa@gmail.com> Cc: rrs@freebsd.org, hiren@freebsd.org, Jason Eggleston <jeggleston@llnw.com>, Julien Charbon <jch@freebsd.org>, "freebsd-net@freebsd.org" <net@freebsd.org>, jtl@freebsd.org Subject: Re: listening sockets as non sockets Message-ID: <20170224131349.GJ15630@zxy.spb.ru> In-Reply-To: <20170223153636.GI15630@zxy.spb.ru> References: <20170127005251.GM2611@FreeBSD.org> <20170210063024.GE1973@FreeBSD.org> <20170216184903.GF58829@FreeBSD.org> <0858647a-ec3c-1a78-053f-d04397a82d8a@freebsd.org> <20170222232704.GJ8899@FreeBSD.org> <CAMOc5cw7fkrS6LVCwo1hXY459O9p5VdgyfcE4Apkc-kAgQWu7A@mail.gmail.com> <20170223153636.GI15630@zxy.spb.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Feb 23, 2017 at 06:36:36PM +0300, Slawa Olhovchenkov wrote: > On Thu, Feb 23, 2017 at 11:40:21AM +0800, Sepherosa Ziehau wrote: > > > On Thu, Feb 23, 2017 at 7:27 AM, Gleb Smirnoff <glebius@freebsd.org> wrote: > > > On Tue, Feb 21, 2017 at 09:58:45AM +0100, Julien Charbon wrote: > > > J> > T> 1) The patch worked pretty okay, but the idea of separate file type is > > > J> > T> abandoned. With current filedescriptor code it is almost impossible > > > J> > T> to racelessly switch fileops and f_data. > > > J> > T> For more details read the commit message, URL below. > > > J> > T> > > > J> > T> 2) I moved the work to git. It allows for easier code sharing. > > > J> > T> > > > J> > T> https://github.com/glebius/FreeBSD/commit/3a52df429889ea9c6e61013f6913aad95939f159 > > > J> > > > > J> > The current 'solisten' branch at https://github.com/glebius/FreeBSD has > > > J> > been running successfully for 48 hours at Netflix. But as mentioned before, > > > J> > our connection rate is pretty low. > > > J> > > > > J> > So, testing from Julien and Hiren will be much appreciated. > > > J> > > > J> Sound good, I will have spare time to get performance results on our > > > J> side w/ and w/o your change before the end of this week. > > > > > > Thanks, Julien! Looking forward for your results. Meanwhile, the code has > > > been running for a week on a single Netflix box. And for 24 hours on a > > > set of boxes as A/B test against non-patched code. Since our connection > > > rate is low, and we never seen accept_mtx to be a major problem in our > > > profiling, the A/B test didn't show any statistically significant > > > improvements. But my biased eye of course sees some improvements in the > > > graphs :) > > > > You can use this to force clients to generate 1 request/connection > > (--connreqs 1) to test the accept queue contention for HTTP workload: > > https://github.com/sepherosa/wrk > > > > Checkout the sephe/wrk branch. pkg install wrk. Then gmake in the > > checked out directory. > > pkg install gcc gmake, may be? > > PANIC: unprotected error in call to Lua API (27) > PANIC: unprotected error in call to Lua API (13) Ah, sorry, wrong branch. Can I use per-connection Lua script (for delays and url generation)? Not per-thread. I am need to emulate many client connections (about 10K-40K). Currently I am have some trouble w/ >20K connections.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20170224131349.GJ15630>