Date: Wed, 18 Apr 2001 03:38:21 +1000 (EST) From: Rowan Crowe <rowan@sensation.net.au> To: Chet Hosey <chosey@nidhog.com> Cc: freebsd-isp@FreeBSD.ORG Subject: Re: boa small/fast web server Message-ID: <Pine.BSF.4.21.0104180331570.90106-100000@velvet.sensation.net.au> In-Reply-To: <Pine.BSF.4.31.0104171306390.69834-100000@web1.nidhog.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 17 Apr 2001, Chet Hosey wrote: > I don't know if this is relevant, but when listen() is called on a socket > (as it likely is by boa), the application can tell the kernel how many > queued connections to keep. Although I may be entirely wrong, I would > guess that perhaps boa isn't letting the kernel queue enough connections. > > Unfortunately, there seems to be no simple way to override the > application's requests. A cap may be set on backlog length using > kern.ipc.somaxconn; depending on what boa requests, increasing this value > (default 128, it seems), may help. Chet, Thanks for the tip. Boa is requesting a backlog of 250, while kern.ipc.somaxconn was set to 128. These numbers seem rather large - at peak Boa is completing about 25 to 30 images per second - but I tried setting somaxconn to 250, and it seems to have worked! The number of overflows reported hasn't changed since I restarted Boa. I haven't yet been able to determine with a great amount of confidence whether this has cured the broken image problem, as I'm still getting a lot of connection reset by peer errors in Boa's error log, along with the dodgy "-H605" timezone (it should be something like "+1000") Thanks again. :) Cheers. -- Rowan Crowe http://www.rowan.sensation.net.au/ Sensation Internet Services http://info.sensation.net.au/ Melbourne, Australia Phone: +61-3-9388-9260 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-isp" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.21.0104180331570.90106-100000>