Date: Tue, 27 Apr 2004 14:04:50 +0200 From: Ivan Voras <ivoras@fer.hr> To: Luigi Rizzo <rizzo@icir.org> Cc: current@freebsd.org Subject: Re: Dummynet low bandwidth simulation Message-ID: <408E4C62.9090207@fer.hr> In-Reply-To: <20040427043253.B50266@xorpc.icir.org> References: <408D85FD.10809@fer.hr> <20040426153919.A74609@xorpc.icir.org> <408DA87E.3000603@fer.hr> <20040427043253.B50266@xorpc.icir.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Luigi Rizzo wrote:
> considering your msg about rst etc -- there are a number of potential
> failure modes including the most likely one below:
>
> + because of the extra delay, the initial syn handshake takes too
> long to complete causing way too many entries in the listen queue,
> so this saturates the server side.
I dont' understand this - the messages ("limiting RST response freq.") are
not showing on the server side (where the web server is), but on the client.
(Apache's default backlog is 511 and I raised kern.ipc.somaxconn to 1000 on
the server).
But I've discovered a possible cause: apache httpd processes have been
dyeing on the server, and occasionally I see: "collecting pv entries,
suggest increasing PMAP_SHPGPERPROC". What is that?
--
Every sufficiently advanced magic is indistinguishable from technology
- Arthur C Anticlarke
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?408E4C62.9090207>