From owner-freebsd-current Tue Oct 15 22:12:31 1996 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.5/8.7.3) id WAA09000 for current-outgoing; Tue, 15 Oct 1996 22:12:31 -0700 (PDT) Received: from phoenix.csie.nctu.edu.tw (phoenix.csie.nctu.edu.tw [140.113.17.171]) by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id WAA08993 for ; Tue, 15 Oct 1996 22:12:28 -0700 (PDT) Received: from FreeBSD.csie.NCTU.edu.tw (freebsd.csie.nctu.edu.tw [140.113.235.250]) by phoenix.csie.nctu.edu.tw (8.7.5/8.7.5) with ESMTP id NAA17616 for ; Wed, 16 Oct 1996 13:09:40 +0800 (CST) Received: (from jdli@localhost) by FreeBSD.csie.NCTU.edu.tw (8.7.6/8.7.3) id NAA19045 for freebsd-current@freebsd.org; Wed, 16 Oct 1996 13:11:07 +0800 (CST) From: Jian-Da Li Message-Id: <199610160511.NAA19045@FreeBSD.csie.NCTU.edu.tw> Subject: mbuf full problem To: freebsd-current@freebsd.org Date: Wed, 16 Oct 1996 13:11:06 +0800 (CST) X-Mailer: ELM [version 2.4ME+ PL28 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 8bit Sender: owner-current@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Hi : Our FreeBSD server is running 2.2-961006-SNAP with the latest current kernel, and there is a serious problem about mbuf useage (since long ago). I found that whenever the mbuf clusters useage reached (or over) the limit defined by NMBCLUSTERS, it seems will never decrease. The mbuf full problem is usually caused by too many mirror processes running, or the search engine updating it's gathered database. The mbuf useage is normally around 800 on this server, but when too-many-mirrors are running or search engine is gathering, it will increase to 1200 (current kernel limit) even to 1800 (old kernel limit). And it will cause network almost down except ping. I tried to kill those processes, but the mbuf useage just keep staying around 1200 (was 1800) no matter how many network-related processess I killed. I have to reboot the machine to solve this problem. Now I increase NMBCLUSTERS back to 1800 to make the bad thing happen slower, but I don't think it's the right way to go. Is it a mbuf-leaking ? Which information else can I provide ? Any idea ? ====== netstat -m, maxuser=48, DRAM=32MB ====== 1430 mbufs in use: 1366 mbufs allocated to data 29 mbufs allocated to packet headers 23 mbufs allocated to protocol control blocks 12 mbufs allocated to socket names and addresses 1280/1280 mbuf clusters in use 2738 Kbytes allocated to network (100% in use) 0 requests for memory denied 0 requests for memory delayed 0 calls to protocol drain routines =============================================== Thanks for your help. -- 李 建 達 (Jian-Da Li) 交 大 資 工 E-Mail : http://www.csie.nctu.edu.tw/~jdli