From owner-freebsd-hackers Thu Jul 20 16:37:07 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.cdrom.com (8.6.11/8.6.6) id QAA29211 for hackers-outgoing; Thu, 20 Jul 1995 16:37:07 -0700 Received: from vinkku.hut.fi (vinkku.hut.fi [130.233.245.1]) by freefall.cdrom.com (8.6.11/8.6.6) with ESMTP id QAA29204 for ; Thu, 20 Jul 1995 16:37:05 -0700 Received: from lk-hp-20.hut.fi (lk-hp-20.hut.fi [130.233.247.32]) by vinkku.hut.fi (8.6.11/8.6.7) with ESMTP id CAA04093 for ; Fri, 21 Jul 1995 02:36:55 +0300 From: Juha Inkari Received: (inkari@localhost) by lk-hp-20.hut.fi (8.6.11/8.6.7) id CAA14979 for freebsd-hackers@freefall.cdrom.com; Fri, 21 Jul 1995 02:36:55 +0300 Message-Id: <199507202336.CAA14979@lk-hp-20.hut.fi> Subject: uptimes (was Re: What people are doing with FBSD) To: freebsd-hackers@freefall.cdrom.com Date: Fri, 21 Jul 1995 02:36:55 +0200 (EETDST) In-Reply-To: <199507190657.IAA09555@labinfo.iet.unipi.it> from "Luigi Rizzo" at Jul 19, 95 08:57:10 am X-Mailer: ELM [version 2.4 PL22] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1049 Sender: hackers-owner@FreeBSD.org Precedence: bulk > > non-existant source management, and probably some history. FreeBSD gets > > hurt badly by unstability, not much change to "sell" FreeBSD to anyone as > > long as the longest uptimes are weeks. > I see consistently uptimes of 50+ days on all our server/workstations. > These system only crash when there is a power outage (no UPSs here). I see a 138 days uptime on a NetBSD 1.0 box, up since the last blackout. It has done better than a somewhat similar setup on same sort of jobs that runs Dell SVR4 (a few crashes). I also run NetBSD and FreeBSD current versions, that do fairly well too, but ofcourse get booted after new kernel builds. The guilty part for unstability could be flakey or buggy hardware, too. How does one measure OS stability anyway ? If we compare uptime values, the system must operate under a load that can be reproduced on several test runs. And stability would be better expressed with a figure that depends on the number of tasks performed, than with a figure that gets better each second the system sits idle.