Date: Tue, 18 Jul 2006 12:03:34 -0700 From: Doug Barton <dougb@FreeBSD.org> To: John Von Essen <john@essenz.com> Cc: freebsd-hackers@freebsd.org Subject: Re: odd behavior in apache 2.0.58 today Message-ID: <44BD3086.1010906@FreeBSD.org> In-Reply-To: <20060718143105.B64880@beck.quonix.net> References: <20060718140354.V64880@beck.quonix.net> <44BD2783.1000609@FreeBSD.org> <20060718143105.B64880@beck.quonix.net>
next in thread | previous in thread | raw e-mail | index | archive | help
John Von Essen wrote:
> Doug,
>
> Did some googling and I did find a connection between excessive
> CLOSED_WAITS, and hanging apache, and webbots. Some of the IP's I saw in
> my netstat were bots too. The problem has something to do with the bot
> no longer accepting data, but apache will continue to send it back since
> the bot didn't close the connection.
Yep, that sounds right to me.
> Because my MaxClients was set to 150, my server got so bogged down, that
> it actually crashed and I had to power cycle. I moved that down to 75,
> which is still more than enough for me, so if it happens again, my
> system should remain up in order to restart apache.
>
> Any ideas if I can tweak apache to prevent this from happening? Maybe
> turn-off keepalive?
Reducing MaxClients to a sane level is the best bet here. Maybe throw in a
monitoring script for system load, and/or apache going insane with too many
children. Keepalives are generally a good idea, and for legit clients do
more good than harm, so I wouldn't turn them off personally.
Doug
--
This .signature sanitized for your protection
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?44BD3086.1010906>