Date: Sat, 13 Nov 2010 00:18:29 +1100 From: Lawrence Stewart <lstewart@freebsd.org> To: Kostik Belousov <kostikbel@gmail.com> Cc: freebsd-net@freebsd.org, Igor Sysoev <is@rambler-co.ru>, Andre Oppermann <andre@freebsd.org>, Maxim Dounin <mdounin@mdounin.ru> Subject: Re: net.inet.tcp.slowstart_flightsize in 8-STABLE Message-ID: <4CDD3EA5.7020101@freebsd.org> In-Reply-To: <20101112104109.GY2392@deviant.kiev.zoral.com.ua> References: <20100713140051.GV99657@mdounin.ru> <4C5BCE48.5070504@freebsd.org> <20100913172453.GG99657@mdounin.ru> <20100927081217.GM44164@mdounin.ru> <4CA09987.8020205@freebsd.org> <20101102001134.GP44164@mdounin.ru> <4CD090EF.4020402@freebsd.org> <4CD64814.2000906@freebsd.org> <4CDD0C7D.1070102@freebsd.org> <4CDD1529.8070504@freebsd.org> <20101112104109.GY2392@deviant.kiev.zoral.com.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
On 11/12/10 21:41, Kostik Belousov wrote: > On Fri, Nov 12, 2010 at 09:21:29PM +1100, Lawrence Stewart wrote: >> On 11/12/10 20:44, Lawrence Stewart wrote: >>> On 11/07/10 17:32, Lawrence Stewart wrote: >>>> On 11/03/10 09:30, Andre Oppermann wrote: >>>>> On 02.11.2010 01:11, Maxim Dounin wrote: >>>>>> Hello! >>>>>> >>>>>> On Mon, Sep 27, 2010 at 03:17:59PM +0200, Andre Oppermann wrote: >>>>>> >>>>>>> On 27.09.2010 10:12, Maxim Dounin wrote: >>>>>>>> Hello! >>>>>>>> >>>>>>>> On Mon, Sep 13, 2010 at 09:24:53PM +0400, Maxim Dounin wrote: >>>>>>>> >>>>>>>> [...] >>>>>>>> >>>>>>>>> Andre, could you please take a look at one more patch as well? >>>>>>>>> >>>>>>>>> Igor reported that it still sees 100ms delays with rfc3465 turned >>>>>>>>> on, and it turns out to be similar issue (setting cwnd to 1*MSS) >>>>>>>>> for hosts found in hostcache. >>>>>>>>> >>>>>>>>> The problem with setting cwnd from hostcache was already reported >>>>>>>>> here: >>>>>>>>> >>>>>>>>> http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/92690 >>>>>>>>> http://lists.freebsd.org/pipermail/freebsd-net/2007-July/014780.html >>>>>>>>> >>>>>>>>> As using larger cwnd from hostcache may cause problems on >>>>>>>>> congested links (see second thread) I changed code to only use >>>>>>>>> cached cwnd as an upper bound for cwnd (instead of fixing current >>>>>>>>> code). This is also in-line with what we do on connection >>>>>>>>> restarts. >>>>>>>>> >>>>>>>>> We may later consider re-adding usage of larger cwnd from >>>>>>>>> hostcache. But I believe it should be done carefully and >>>>>>>>> probably behind sysctl, off by default. >>>>>>>> >>>>>>>> Ping. >>>>>>> >>>>>>> Thanks for the reminder. I'll disable priming CWND from hostcache. >>>>>> >>>>>> Ping again. >>>>>> >>>>>> I would like to see the patch in question[1] committed and MFC'ed >>>>>> somewhere before 8.2. >>>>>> >>>>>> Andre, if you are just too busy to do it yourself - please say so, >>>>>> I'll ask somebody else to commit it. >>>>> >>>>> Thanks for the reminder. After EuroBSDCon Developer Summit I was >>>>> busy with other work and most of my tcp work was stalled due to >>>>> review bandwidth issues. >>>> >>>> Sorry for the review bandwidth issues. I'm only just managing to keep my >>>> head above water at the moment and have had to reduce much of my FreeBSD >>>> work to a trickle. >>>> >>>>> Lawrence, could you please spare a few seconds and take a quick look >>>>> at the attached patch? >>>> >>>> The patch looks fine, please commit. I hope to give the hostcache some >>>> TLC at some point, but in the meantime it's fine to ignore cached cwnd. >>> >>> FYI Andre, this patch got rolled into r215166 because I had to move all >>> the code around which the patch was based on. Is that going to cause any >>> issues here? I should have checked prior to committing but completely >>> forgot I had pulled the patch into my dev tree before creating the mod >>> CC patch. >> >> Gah, I think I've answered my own question. MFCing the patch standalone >> in time for 8.2 as Maxim requested is not going to be possible. I think >> I'll have to back out r215166, commit your patch and then recommit the >> modular CC patch. >> >> Unless someone can think of a way that doesn't involve all the mucking >> about? > Well, if the change is indeed already in HEAD, you can do a direct commit > to stable/8 after proper MFC period, noting that this is a cherry-pick of > the part of corresponding revision. hmm, not ideal but would get the job done. When I eventually MFC the modular CC patch everything will be back in sync mergeinfo wise as well. Would anyone object if I follow Kostik's suggestion and direct commit the TCP_METRICS_CWND patch to stable/8 in, say, a week's time? If I don't hear anything, I'll aim to do the commit next weekend. Cheers, Lawrence
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4CDD3EA5.7020101>