Date: Mon, 2 Apr 2007 08:11:39 -0700 From: Barrett Lyon <blyon@blyon.com> To: Andre Oppermann <andre@freebsd.org> Cc: current@freebsd.org Subject: Re: TCP sessions hanging in FIN_WAIT_1? Message-ID: <EAEDF08B-55EC-4A66-9D26-6B1DFB619414@blyon.com> In-Reply-To: <4610CD68.2060706@freebsd.org> References: <20070401005604.051A473039@freebsd-current.sentex.ca> <F557156A-76D0-4A6D-B4E0-2940A7089968@blyon.com> <4610CD68.2060706@freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Generic kernel, I had some stuff in there related to the http accept filter but I removed all that when I noticed the problem. I moved the traffic over to if_em for testing and the problem went away, when I brought it back to mxge the FIN_WAIT_1 issue came back. This was without rebooting or changing anything other than the interface. -Barrett On Apr 2, 2007, at 2:31 AM, Andre Oppermann wrote: > Barrett Lyon wrote: >> I've been testing CURRENT for a bit and with a thousands of short >> lived HTTP sessions, CURRENT hangs most of the sessions in >> FIN_WAIT_1 status until the machine runs out of resources and >> causes the network stack to stop performing. The HTTP daemon I >> am playing with is lighttpd-1.4.13, and I don't see this >> happening on any of the 6.2 test machines I run. >> There's nothing unusual about the build other than the interface >> card running the mxge driver. Has anyone else seen reports of >> this? I have seen this behavior on all source up to today's tree. > > Are you running a pure GENERIC kernel or do you have any other > modifications? > What about firewall packages (ipfw, pf, ipfilter)? > > -- > Andre >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?EAEDF08B-55EC-4A66-9D26-6B1DFB619414>