Date: 12 Aug 1999 12:43:13 +0200 From: Dag-Erling Smorgrav <des@flood.ping.uio.no> To: Doug <Doug@gorean.org> Cc: ndear@areti.net, freebsd-isp@FreeBSD.ORG Subject: Re: mbuf clusters. Message-ID: <xzphfm5th32.fsf@flood.ping.uio.no> In-Reply-To: Doug's message of "Wed, 11 Aug 1999 10:25:02 -0700" References: <199908111513.QAA21355@post.mail.areti.net> <37B1B1EE.4DCF214F@gorean.org>
next in thread | previous in thread | raw e-mail | index | archive | help
Doug <Doug@gorean.org> writes: > It's impossible to answer this question intelligently without knowing your > version of the OS and your current kernel maxusers setting, not to mention > memory, processor, etc. However in general for a short term crisis > situation like this you can go to 512 maxusers and 15000 NMBCLUSTERS on a > 2.2.8 system, assuming that you have enough ram. No. Increasing MAXUSERS beyond 128 on a pre-3.2 system is very risky. The problem is not RAM, but kernel virtual memory space. BTW, there's no point in increasing MAXUSERS if you set NMBCLUSTERS separately. The only other parameters controlled by MAXUSERS are NPROC (20 + 16 * MAXUSERS) and MAXFILES (NPROC * 2), but MAXFILES can be tuned at run time by setting the kern.maxfiles sysctl variable. The default for NMBCLUSTERS is (512 + MAXUSERS * 16). > I'd say 256M ought to do > it. If you have a 3.2 system you can go higher on these numbers, but > chances are you won't need to. netstat [-s | -sr | -m] will give you a > better idea of hot to tune it when the crisis is over. If you have a 3.2 system, you can simply add the following line to /boot/loader.conf: kern.ipc.nmbclusters="8192" No need to build a new kernel. DES -- Dag-Erling Smorgrav - des@flood.ping.uio.no 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?xzphfm5th32.fsf>