atom feed192 messages in org.freebsd.freebsd-archRe: FreeBSD problems and preliminary ...
FromSent OnAttachments
Vadim GoncharovAug 17, 2011 4:09 pm 
Baptiste DaroussinAug 18, 2011 12:49 am 
Robert WatsonAug 18, 2011 8:41 am 
Maxim IgnatenkoAug 18, 2011 11:09 am 
Lev SerebryakovAug 18, 2011 12:49 pm 
Lev SerebryakovAug 18, 2011 2:23 pm 
Vadim GoncharovAug 18, 2011 2:45 pm 
Hans Petter SelaskyAug 18, 2011 2:45 pm 
Vadim GoncharovAug 18, 2011 2:59 pm 
Vadim GoncharovAug 18, 2011 3:17 pm 
Julian ElischerAug 18, 2011 7:53 pm 
Michael V. BuzuverovAug 18, 2011 8:48 pm 
Lars EngelsAug 18, 2011 10:55 pm 
Lars EngelsAug 19, 2011 12:37 am 
Robert WatsonAug 19, 2011 1:23 am 
Lev SerebryakovAug 19, 2011 1:36 am 
Robert WatsonAug 19, 2011 1:41 am 
Lev SerebryakovAug 19, 2011 1:49 am 
Robert WatsonAug 19, 2011 1:55 am 
Slawa OlhovchenkovAug 19, 2011 2:05 am 
Lev SerebryakovAug 19, 2011 2:05 am 
Adrian ChaddAug 19, 2011 2:05 am 
Slawa OlhovchenkovAug 19, 2011 2:22 am 
Lev SerebryakovAug 19, 2011 2:25 am 
Lev SerebryakovAug 19, 2011 2:27 am 
Robert N. M. WatsonAug 19, 2011 2:36 am 
Slawa OlhovchenkovAug 19, 2011 2:36 am 
Robert N. M. WatsonAug 19, 2011 2:37 am 
Slawa OlhovchenkovAug 19, 2011 2:50 am 
Bob BishopAug 19, 2011 2:51 am 
Pieter de GoejeAug 19, 2011 5:01 am 
Kris MooreAug 19, 2011 5:20 am 
Lev SerebryakovAug 19, 2011 6:52 am 
Adrian ChaddAug 19, 2011 8:26 am 
Gary PalmerAug 19, 2011 10:22 am 
Robert WatsonAug 19, 2011 11:49 am 
Nathan WhitehornAug 19, 2011 1:17 pm 
Pedro F. GiffuniAug 19, 2011 1:29 pm 
Gary PalmerAug 19, 2011 1:30 pm 
Kostik BelousovAug 19, 2011 2:26 pm 
Rick MacklemAug 19, 2011 3:38 pm 
Robert WatsonAug 19, 2011 3:57 pm 
Pedro F. GiffuniAug 19, 2011 5:57 pm 
Lev SerebryakovAug 19, 2011 11:14 pm 
Robert WatsonAug 20, 2011 4:37 am 
Luigi RizzoAug 20, 2011 6:42 am 
Poul-Henning KampAug 20, 2011 7:10 am 
Robert N. M. WatsonAug 20, 2011 7:21 am 
Lev SerebryakovAug 20, 2011 2:09 pm 
Luigi RizzoAug 20, 2011 2:36 pm 
Peter JeremyAug 21, 2011 4:04 am 
Slawa OlhovchenkovAug 21, 2011 6:51 am 
Marcin WisnickiAug 22, 2011 10:04 am 
Marcin WisnickiAug 22, 2011 11:33 am 
selvenAug 22, 2011 11:06 pm 
Adrian ChaddAug 22, 2011 11:49 pm 
Garrett CooperAug 22, 2011 11:54 pm 
Diane BruceAug 23, 2011 7:01 pm 
Vadim GoncharovAug 24, 2011 3:08 pm 
Vadim GoncharovAug 24, 2011 3:15 pm 
Vadim GoncharovAug 24, 2011 3:20 pm 
Vadim GoncharovAug 24, 2011 3:29 pm 
Milo HysonAug 24, 2011 7:42 pm 
gnuy...@gmail.comAug 24, 2011 8:33 pm 
Alex GoncharovAug 24, 2011 8:47 pm 
Milo HysonAug 24, 2011 10:04 pm 
Adrian ChaddAug 24, 2011 10:24 pm 
Vadim GoncharovAug 25, 2011 8:20 am 
Vadim GoncharovAug 25, 2011 8:33 am 
Vadim GoncharovAug 25, 2011 8:41 am 
122 later messages
Subject:Re: FreeBSD problems and preliminary ways to solve
From:Slawa Olhovchenkov (sl@zxy.spb.ru)
Date:Aug 19, 2011 2:05:13 am
List:org.freebsd.freebsd-arch

On Fri, Aug 19, 2011 at 09:56:02AM +0100, Robert Watson wrote:

On Fri, 19 Aug 2011, Lev Serebryakov wrote:

Our network stack is actually pretty parallel as such things go, and there are a number of changes in FreeBSD 9.x that extend this work. Most of the performance work is being done on edge nodes rather than middle nodes -- i.e., maxing out multiple 10gbps links serving content, etc, rather than in routing configurations, though. We also have a strong and growing collection of 10gbps drivers. You'll find our drivers lifted for many other systems, including Solaris :-).

I need to bribe our admins (OPs) on my paid work to try FreeBSD instead of Solaris on new servers :) We are processing huge amount of multicast streams (up to 2.5-3Gbit/s with 500-1000 bytes packets) and have difficulties not to lose any packets on Solaris :) They tried Linux without success, but FreeBSD is unknown to them.

One problem for FreeBSD is that our applications are Java-based... Problems are not in Java, but in Intel drivers (igb / ixgb in FreeBSD terms), which sometimes lose packets with "buffer is not available" diagnostics when consumer is heavily-multithreaded.

Is the issue here that FreeBSD is dropping more packes, or just that FreeBSD is reporting that it drops packets? Historically, we've returned ENOBUFS from datagram sockets when the interface queue is overflowed, but some other systems (most noticeably Linux) simply return success when they drop a packet on an outgoing interface queue. You can debate which is the better model, but one impact is that sometimes people report errors on FreeBSD that they don't see on Linux -- when actually, the same failure is present, we just allow the application to learn about it.

Historically, Linux on datagram (UDP) socket allow use select, FreeBSD -- don't allow. FreeBSD always report 'UDP socket ready to transmit'. And after try to send packet -- 'oops, ENOBUFS'.