From owner-freebsd-questions@FreeBSD.ORG Tue Sep 19 22:21:51 2006 Return-Path: X-Original-To: freebsd-questions@freebsd.org Delivered-To: freebsd-questions@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 7738616A412 for ; Tue, 19 Sep 2006 22:21:51 +0000 (UTC) (envelope-from pobox@verysmall.org) Received: from thing.verysmall.org (thing.verysmall.org [212.100.226.116]) by mx1.FreeBSD.org (Postfix) with ESMTP id E23EF43D5F for ; Tue, 19 Sep 2006 22:21:49 +0000 (GMT) (envelope-from pobox@verysmall.org) Received: from [192.168.2.4] (chello084114134105.3.15.vie.surfer.at [84.114.134.105]) by thing.verysmall.org (Postfix) with ESMTP id 53E5E2282D; Tue, 19 Sep 2006 23:20:42 +0100 (BST) Message-ID: <45106D76.9020701@verysmall.org> Date: Wed, 20 Sep 2006 00:21:42 +0200 From: "pobox@verysmall.org" User-Agent: Thunderbird 1.5.0.7 (Windows/20060909) MIME-Version: 1.0 To: freeBSD References: <36F4A97F-BD2A-4602-A87F-881F59EA72E4@cbpratt.prohosting.com> <44odtbh9cn.fsf@be-well.ilk.org> In-Reply-To: <44odtbh9cn.fsf@be-well.ilk.org> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: Chris Subject: Re: cvsup between levels, single user issue X-BeenThere: freebsd-questions@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: User questions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 19 Sep 2006 22:21:51 -0000 Lowell Gilbert wrote: > Chris writes: > >> I was going to attempt an in-place upgrade of a light load server >> running 5.3 to bring it to 6.1 RELEASE P6 (RELENG_6_1). I read in >> UPDATING that going between 5.x and 6.x requires single user mode >> which isn't an option on a remote machine (though I can stop >> sendmail, bind, popper etc manually but not that which keeps me on >> like sshd). I've not taken a machine to single user yet during my >> builds but have always had the box in my possession and on my private >> net when I've done that. I saw threads even today that left the issue >> somewhat up in the air. I still need to protect the machine. Would a >> doable jump be to move the machine to RELENG_5_5 remotely or would >> that most likely be unsuccessful without going single user? There is >> no exotic hardware on the >> box. _______________________________________________ > > Look at it this way. Yes, in most cases, an update can be done > without using single-user mode, but what are you going to do if > your new kernel doesn't boot? [If you've already installed the > userland before that, you are well and truly stuck. So at least > make sure you reboot with the new kernel before starting the > installworld.] I was the origin of one of the recent questions touching the problem with updating a remote server. There were quite detailed answers that sum up to three ways - 1. Somebody at the remote site does the last 3-4 steps. This requires someone who knows what to do as mergemaster is interactive. 2. Some kind of back door through a serial cable/modem. 3. The 'do not worry' approach - finish without dropping to single user mode, just stop all services except ssh. Several people said they have done it successfully for years and years. May be there were some more or variations which I miss. What I did not quite get is - what damage can be done if update is not done in single user mode? Let's say, as Lowell says - we reboot just to test the new kernel. If it boots, we stop all services except ssh (does 'all services' mean all listed in rc.conf), install world, mergemaster and reboot. Iv. --