From owner-freebsd-bugs Sun Apr 7 19:23:25 1996 Return-Path: owner-bugs Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id TAA20400 for bugs-outgoing; Sun, 7 Apr 1996 19:23:25 -0700 (PDT) Received: from ter2.fl.net.au (root@ter2.fl.net.au [203.63.198.3]) by freefall.freebsd.org (8.7.3/8.7.3) with ESMTP id TAA20315 for ; Sun, 7 Apr 1996 19:22:00 -0700 (PDT) Received: from cafu.fl.net.au (adf@cafu.fl.net.au [203.63.198.10]) by ter2.fl.net.au (2.0/adf) with SMTP id MAA10258; Mon, 8 Apr 1996 12:24:55 +1000 (EST) Date: Mon, 8 Apr 1996 12:22:45 +0000 () From: Andrew Foster To: freebsd-bugs@freebsd.org cc: Rob Wise Subject: Machines locking Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-bugs@freebsd.org X-Loop: FreeBSD.org Precedence: bulk I've observed this on several machines : A lockup in which : - pings return - some programs will still work via opening a socket (e.g. telnetting to port 80 will sometimes work, but not always) - I can telnet in (no connection refused) but nothing will happen - I can switch vtys at the console - Pressing enter at the console login prompt will redisplay the banner - Entering a username and pressing enter will do nothing - the machine just sits there - The only way to fix it is to press reset - My syslog.conf mentions a file which doesn't exist - as a result when I press C-A-D I see a syslogd error - so the machine tries to respawn something (this is my guess) but can't. Now, this happened on 2 machines I had which were under a lot of load (16 PPP sessions running from them). One of these machines just does some simple routing, erpcd and mail and NFS and it rarely locks up now. It's a DX4-100 w/16MB. One main machine used to do : - Apache 1.0.2 - INN - erpcd - Harvest proxy cache - shell, mail, DNS, etc And locked up a lot. I moved INN, erpcd and harvest to another machine. Both lock up now, but the P90 32MB machine running INN, erpcd and harvest locks up much less than the DX4-100 w/16MB running only some mail, DNS, apache and shell. It seems that the more that the machine is doing (not necessarily the load average - I've had it lockup when the load av was 1.10 just before it happened) the higher the chance of a lock. One machine (which does it the most) is now running stable. It's happened in release (the P90 w/32MB runs release) and also I've tried 960303-SNAP. I'm running the stable supp'ed from sup.au.undernet.org.au on Saturday 6 April evening. Thanks, Andrew Foster ---------- Andrew Foster adf@fl.net.au