From owner-freebsd-stable Sat Jul 27 17:21:46 2002 Delivered-To: freebsd-stable@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 8C87537B400 for ; Sat, 27 Jul 2002 17:21:41 -0700 (PDT) Received: from postal3.es.net (postal3.es.net [198.128.3.207]) by mx1.FreeBSD.org (Postfix) with ESMTP id D1B2743E5E for ; Sat, 27 Jul 2002 17:21:40 -0700 (PDT) (envelope-from oberman@es.net) Received: from ptavv.es.net ([198.128.4.29]) by postal3.es.net (Postal Node 3) with ESMTP id GQF37091; Sat, 27 Jul 2002 17:21:39 -0700 Received: from ptavv (localhost [127.0.0.1]) by ptavv.es.net (Postfix) with ESMTP id AACA05D03; Sat, 27 Jul 2002 17:21:36 -0700 (PDT) To: Chris Pepper Cc: Paulo Fragoso , Alexandr Kovalenko , Gavin Atkinson , Michiel Boland , freebsd-stable@FreeBSD.ORG Subject: Re: 4.6-RELEASE -> 4.5-RELENG In-reply-to: Your message of "Thu, 25 Jul 2002 23:54:13 EDT." Date: Sat, 27 Jul 2002 17:21:36 -0700 From: "Kevin Oberman" Message-Id: <20020728002136.AACA05D03@ptavv.es.net> Sender: owner-freebsd-stable@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.ORG > Date: Thu, 25 Jul 2002 23:54:13 -0400 > From: Chris Pepper > > At 10:56 AM -0700 2002/07/10, Kevin Oberman wrote: > > > Date: Wed, 19 Jun 2002 15:14:00 -0300 (BRT) > >> From: Paulo Fragoso > >> Sender: owner-freebsd-stable@FreeBSD.ORG > >> > >> On Wed, 19 Jun 2002, Alexandr Kovalenko wrote: > >> > >> > You also should update your MAKEDEV script with mergemaster. > >> > And to sh MAKEDEV all again. Device minor/major numbers was changed with > >> > ata(4) MFC. > >> > >> After we redid all following this steps: > >> > >> make buildworld > >> make installworld > >> make buildkernel KERNCONF=KERNEL6 > >> make installkernel KERNCONF=KERNEL6 > >> mergemaster -a > >> cd /dev && sh MAKEDEV all > >> > >> and restoring some files from backup (group, rc.firewall, etc), all > >> is working fine. > > > >I keep posting this because people keep building new systems in a > >dangerous way. > > > >Doing the installworld before building and booting the new kernel is > >very risky. It will almost always make no difference, but, on the > >rare occasions when it does, it will leave you with a VERY sick > >system and the only way out maybe to re-install FreeBSD from scratch. > > > >While there are a few niggles, the proper sequence is: > >mergemaster -p (This is NEW!) > >make buildworld > >make kernel KERNCONF=your_kernel_conf_name > > Why not 'make installkernel' per > ? > Is there a problem with the Handbook? Just to save keystrokes when you will be installing the new kernel immediately after building. The 'kernel' make object is literally 'buildkernel installkernel'. There is no real difference and, if you don't want to install the new kernel immediately, buildkernel is the proper thing to do. > >REBOOT into single-user mode! (This does not mean drop to single user.) > > Why? Is this to test installkernel -- which I don't see in > your list? I've been using "shutdown now" successfully for a while > now; much easier not to have to remount & set the TZ before > installkernel, then rebooting (multi) to test the kernel, then > "shutdown now" to installworld. It is to confirm that the newly built (and installed) kernel works. IF you find the new kernel does not work, no big. Just boot kernel.old and figure out why it failed. But, if you have already done an installworld, you may not be able to do that. This may mean re-installing the system from scratch with significant down-time. A bad kernel is easy to recover from. A kernel that will not run with the installed userland is almost impossible to fix. I am confused as to what is do difficult about "mount -a -t ufs"? You don't re-boot multi until the installworld is completed. Once again: make kernel (or installkernel) KERNCONF=your_conf_file reboot hit a space while the loader is counting down and: boot -s fsck -p (not really required, but I am paranoid) mount -a -t ufs cd /usr/src make installworld mergemaster -i (or you favorite mergemaster invocation) exit (or reboot) If you exit to multi-user mode, this requires exactly one reboot and the mount command. I fail to see why this is a hardship unless you don't have console access. R. Kevin Oberman, Network Engineer Energy Sciences Network (ESnet) Ernest O. Lawrence Berkeley National Laboratory (Berkeley Lab) E-mail: oberman@es.net Phone: +1 510 486-8634 To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-stable" in the body of the message