Date: Thu, 5 Sep 2002 02:14:49 +0400 (MSD) From: Dmitry Morozovsky <marck@rinet.ru> To: Dominic Marks <dominic_marks@btinternet.com> Cc: Mike Hoskins <mike@adept.org>, Charles Sprickman <spork@inch.com>, <stable@FreeBSD.ORG> Subject: Re: interpreting netstat -m output Message-ID: <20020905021010.C56943-100000@woozle.rinet.ru> In-Reply-To: <20020905014934.J56943-100000@woozle.rinet.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, 5 Sep 2002, Dmitry Morozovsky wrote: DM> AFAIK according to /usr/src/sys/*/param.h, mbuf size if 256 (at least for DM> i386, see /usr/src/sys/i386/include/param.h, and is not defined for DM> Alphas); DM> DM> and mbcluster size defaults to 2k (I suppose the smallest 2^x to DM> cover standard Ethernet frame) DM> DM> however, in /usr/src/sys/i386/boot/dosboot/param.h, MCLBYTES defined as DM> 2^12 (4k). Don't know whether this value is really used. Returning to the original question, I'd never seen on our servers mbuf usage more that 2 times nmbclusters usage, so in every instance where I've explicitly set nmbclusters, I'd also set mbufs as 2*mnbclusters (this may slightly decrease kernel memory reservations comparing to setting nmbclusters only) This applies mainly to TCP servers, where most data flow is described as "Most packets are of maximum MTU". YMMV, you know ;-) Sincerely, D.Marck [DM5020, DM268-RIPE, DM3-RIPN] ------------------------------------------------------------------------ *** Dmitry Morozovsky --- D.Marck --- Wild Woozle --- marck@rinet.ru *** ------------------------------------------------------------------------ To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20020905021010.C56943-100000>