Date: Tue, 23 Sep 2008 19:11:16 +0300 From: Chris <chris@cretaforce.gr> To: Daniel Gerzo <danger@rulez.sk> Cc: Yury Michurin <yury.michurin@gmail.com>, Mel <fbsd.questions@rachie.is-a-geek.net>, freebsd-questions@freebsd.org Subject: Re: Uplading file via Lighttpd - system hangs Message-ID: <1222186276.11548.2.camel@desktop.lan> In-Reply-To: <e70bc55c7418aa0dcadc96b39633d3f0@services.rulez.sk> References: <705829.85873.qm@web110509.mail.gq1.yahoo.com> <692c9a9f0809220911o54c8a7eey41df730aa10f2c9f@mail.gmail.com> <200809221904.42753.fbsd.questions@rachie.is-a-geek.net> <e70bc55c7418aa0dcadc96b39633d3f0@services.rulez.sk>
next in thread | previous in thread | raw e-mail | index | archive | help
If you check top command, do you see lighttpd using 99% CPU ? Also does both php and static content "hang" ? If you have these symptoms the problem is the same. I use strace and found that lighttpd doesn't accept more connections when this thing happens. > Hello guys, > > I have noticed today that I have another problem possibly related to this > issue. I am running lighttpd+php-cgi in a jail on a 7.0-STABLE system as of > Thu Aug 21 10:59:02 CEST 2008. The problem seems to be that lighttpd stops > serving the pages, however I haven't had time to dig into this any more > closer, but I have found out that there's a lot of connections to the > lighty when it seems to be "dead". May be it does not close the connections > properly?? > The only common thing here is that that this jail is serving a page where a > lot of image uploading happens. > > On Mon, 22 Sep 2008 19:04:41 +0200, Mel > <fbsd.questions@rachie.is-a-geek.net> wrote: > > On Monday 22 September 2008 18:11:05 Yury Michurin wrote: > >> Well, I'm working now on creating memory dump. and send it forward for > > more > >> knowledgeable people, > >> however, as you might notice, different people, with different hardware, > >> and even different version 7.0 and 7.1, > >> have the same problem. > >> > >> Even if lighttpd / php / some script / whatever misbehaves, system > > should > >> not be halted by such userland proccess. > > > > I don't think it's halted, I think it's cluttered by invalid syscalls. > > Secondly, any userland process can make the system unresponsive, by bad > > coding. Just write /tmp and /var/tmp full. It's not so hard. > > > > I don't think that's the case here though. Any of you guys logging > netstat > > -m > > output every 500ms? Maybe you can see mbufs being drained just before the > > > system stops servicing syscalls. > > > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1222186276.11548.2.camel>