From owner-freebsd-questions Fri Feb 23 9:25:43 2001 Delivered-To: freebsd-questions@freebsd.org Received: from eltanin.draconis.mw.lg.supercluster.net (eltanin.draconis.mw.lg.supercluster.net [212.68.64.165]) by hub.freebsd.org (Postfix) with ESMTP id 6BB2A37B491 for ; Fri, 23 Feb 2001 09:25:39 -0800 (PST) (envelope-from sven@crashme.org) Received: from localhost (sven@localhost) by eltanin.draconis.mw.lg.supercluster.net (8.11.2/8.9.3) with ESMTP id f1NHPXK04647 for ; Fri, 23 Feb 2001 17:25:34 GMT (envelope-from sven@crashme.org) X-Authentication-Warning: eltanin.draconis.mw.lg.supercluster.net: sven owned process doing -bs Date: Fri, 23 Feb 2001 17:25:33 +0000 (GMT) From: Sven Brandenburg X-Sender: sven@eltanin.draconis.mw.lg.supercluster.net To: questions@freebsd.org Subject: 4.2-STABLE weird behaviour without swap Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-questions@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG Good evening :-) Is it possible that FreeBSD has some very strange problems when the system is configured without any swap partitions? The Box I`m using ran smoothly for half a year, doing its regular cvsups and make worlds, basically a rock solid system. 3 weeks ago, i threw another DIMM into that box, having 1G of memory now. At the same time I decided that this amount of mem should be sufficient and disabled swap. The main service on this host eats huge ammount of memory during its operation but for this application its better to get an "out of memory" and a following application crash (and restart) than having the performance penalty of a swapping system. On two occasions, the system behaved rather odd, not a real crash but definately in unusable state. The first one was a completely halted userland, with the ip stack continuing to work. The system was pingable and would establish a TCP connection on port numbers used by daemons before (sshd,..) None of these daemons were working anymore, only a TCP session was established and after the regular timeouts, closed. The second was a completey dead network, with thousands of syslog messages like this: Feb 22 20:07:25 eltanin /kernel: xl0: no memory for rx list -- packet dropped! Feb 22 20:07:25 eltanin last message repeated 202 times Feb 22 20:07:55 eltanin last message repeated 295 times The userland was still working, i could log in and do a safe reboot. My questions are: Does anybody have similar experiencies? Is it generally a stupid idea to have no swap at all? It looks very strange that a network device driver has problems allocating memory, but userland processes (the starting shell after login) have enough memory to operate. Would this scenario have been different with configured swap? The system is a dual p3-750, 1G mem, aha29160,3c905, SMP kernel. I can provide more details if required. thanks for the time reading this message *g* regards, Sven Brandenburg To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message