From owner-freebsd-net Wed Oct 31 5:27:35 2001 Delivered-To: freebsd-net@freebsd.org Received: from www.microelectronics.com (www.microelectronics.com [4.18.26.20]) by hub.freebsd.org (Postfix) with ESMTP id 3986437B403 for ; Wed, 31 Oct 2001 05:27:32 -0800 (PST) Received: from zul.microcenter.com (zul.microcenter.com [4.18.26.10]) by www.microelectronics.com (Pro-8.9.3/Pro-8.9.3) with SMTP id IAA28247 for ; Wed, 31 Oct 2001 08:27:33 -0500 Received: from sysadm.microcenter.com by zul.microcenter.com via smtpd (for www.microelectronics.com [4.18.26.20]) with SMTP; 31 Oct 2001 13:24:22 UT Received: from mail.microcenter.com (anbhpc.microcenter.com [10.10.29.94]) by sysadm.microcenter.com (8.8.5/8.8.5) with ESMTP id IAA10826 for ; Wed, 31 Oct 2001 08:27:15 -0500 (EST) Message-ID: <3BDFFCEC.32FF9CC@mail.microcenter.com> Date: Wed, 31 Oct 2001 08:30:20 -0500 From: Aaron Bush X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.2-2 i686) X-Accept-Language: en MIME-Version: 1.0 To: freebsd-net@FreeBSD.ORG Subject: tuning network limits? Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Sender: owner-freebsd-net@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org Hello, I have a FreeBSD 4.3-RELEASE box that appears to be dropping network connections. The system is using ipf to track state and only allow web (http, https) connections. Once per day i index one of the virtual domains by using htdig. Sometimes htdig is not able to index the entire site (itself, via route through loopback) and other times it can. The problem is not isolated to htdig though. A PHP application has also reported that it was unable to open a socket. The /var/log/messages show no packets being dropped by the ipf rules (default is to block and log). I am assuming that the system must be hitting some kernel limit and not allowing any more sockets to be created. In the handbook i see mention of tuning the NMBCLUSTERS options and assume that the following command shows that i have indeed hit the limit (mbuf clusters?). Also the requests for memory denied seems to be high, would failed attempts to create sockets cause this to increase? tosh# netstat -m 623/2304/4096 mbufs in use (current/peak/max): 263 mbufs allocated to data 360 mbufs allocated to packet headers 239/1024/1024 mbuf clusters in use (current/peak/max) 2624 Kbytes allocated to network (85% of mb_map in use) 9152 requests for memory denied 3 requests for memory delayed 0 calls to protocol drain routines What other tools can i use to determine where the problem may be. Thanks, -ab To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-net" in the body of the message