lilo removal in squeeze (or, "please test grub2")
in [Debian]
|
From: Jon Dowland on 2 Jun 2010 12:20 On 28/05/2010 17:39, Roger Leigh wrote: > One obvious solution not already mentioned is to back up the bootloader > *in Linux* as a normal file, so the backup software can then just back > it up like any other file. It's a simple enough workaround to the > deficiencies in your backup software. > > dd if=dev/hda of=/boot/bootsector-backup bs=512 count=nnn > > Stick it in as a daily cron job and you're done. When it comes to > restoring, you can just dd it back and you're in business. > I don't think this is necessary. It doesn't solve Stephen's problem (a machine restored from backup is not immediately bootable), and if you boot the machine via some other method, you can just as easily run "update-grub" (or whatever it's called) to re-create the boot. The solution is to have an external boot system to hand. A set of suitably configured USB keys would do it. They could even boot into a special init environment that just ran "update-grub" so that the macine was immediately recovered. Just how often is a total restore-from-backup required, I wonder? -- Jon Dowland
From: Tom H on 2 Jun 2010 12:40 On Wed, Jun 2, 2010 at 10:05 AM, Stephen Powell <zlinuxman(a)wowway.com> wrote: > On Wed, 02 Jun 2010 09:55:58 -0400 (EDT), Tom H wrote: >> On Wed, Jun 2, 2010 at 9:30 AM, Stephen Powell <zlinuxman(a)wowway.com> wrote: >>> On Wed, 02 Jun 2010 09:06:56 -0400 (EDT), Tom H wrote: >>>> >>>> Don't you think that lilo will be left in the repos but not available >>>> at install time? You could then install lilo post-OS-install or >>>> through pre-seeding. >>> >>> Not without an active upstream maintainer. That's the critical need now. >> >> I meant to say the Lenny repos (although I am curious to see whether >> it will really disappear from the Squeeze repos once Squeeze is >> released). > > If they pull it, they will probably pull it *before* squeeze becomes the > current stable release. Once a release becomes the stable release, it's > almost impossible to pull a package from it. I know. I chose the release as a check-point because I don't know at what previous milestone such a move would take place. -- To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org Archive: http://lists.debian.org/AANLkTiniPIhiiVoosAFdKzoIGhwxeMC_pPdhYztuVdKz(a)mail.gmail.com
From: Stephen Powell on 2 Jun 2010 21:30 On Tue, 01 Jun 2010 12:14:42 -0400 (EDT), John Hasler wrote: > Stephen Powell writes: >> Actually, I've been tempted to volunteer to become the upstream >> maintainer for lilo myself. > > Please do so. >> >> However, although the SAPL is written in assembly language, it is >> written in s390 assembly language, which is totally different from x86 >> assembly language. I don't know x86 assembly language at all. > > Set up a Web site (I suggest tuxfamily.org for hosting) and ask for > help. You'll find that there are people who can and will do the coding > but who don't want the hassle and responsibility of running the project. I don't know. I would at least want to be able to review the work of my assistants and have some hope of being able to tell if it makes sense. I appreciate the encouragement, but I'm just not qualified for something like this yet. If lilo is worth saving, surely someone who *is* qualified will step up to the plate. I sure wish I knew what in the world happened to John Coffman though. He was doing a great job. -- .''`. Stephen Powell : :' : `. `'` `- -- To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org Archive: http://lists.debian.org/770226811.246509.1275528114439.JavaMail.root(a)md01.wow.synacor.com
From: Stephen Powell on 3 Jun 2010 23:20 On Wed, 02 Jun 2010 12:02:15 -0400 (EDT), Jon Dowland wrote: > > Just how often is a total restore-from-backup required, I wonder? A total restore from backup could be for one of two purposes: (1) To restore a machine in case of a hard drive failure. Replace the bad drive with a good drive and restore. (2) To clone a new machine similar to an existing machine, changing a few things after restore, such as IP address, MAC address, hostname, etc. Fortunately, its use for the first purpose is rare. It's use for the second purpose is more common. It's faster than a new install. This is done routinely for a Windows desktop machine. It's the quickest way to get rid of a virus/worm/Trojan, etc., provided the backup is not contaminated also. And it is done for new employess to give them a standard image. It's done less often for a back-end Linux server. -- .''`. Stephen Powell : :' : `. `'` `- -- To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org Archive: http://lists.debian.org/652461449.281245.1275621488000.JavaMail.root(a)md01.wow.synacor.com
From: Stephen Powell on 4 Jun 2010 12:10
On Wed, 02 Jun 2010 10:23:54 -0400 (EDT), John Hasler wrote: > Tom H writes: >> I meant to say the Lenny repos (although I am curious to see whether >> [Lilo] will really disappear from the Squeeze repos once Squeeze is >> released). > > If it has an RC bug at the time of the release it will be removed. As of right now, lilo has no release-critical bugs. Bug number 505609 is marked as *affecting* lilo, and it is marked as release-critical, but it is actually assigned to linux-2.6. And I have now conclusively proved that this is a bug in the kernel maintainer scripts. (See my recent posts to the bug log.) The thing that amazes me is that this bug has been open since November 13, 2008. It doesn't look like anyone even tried to diagnose the problem. I was able to diagnose the problem simply by reading the symptoms in the bug report. And I was able to fix it by simply comparing the two maintainer scripts side by side. And I don't even know perl! The idea that modern kernels are too big for lilo to load is a myth: a myth whose origins are apparently tied to this bug report. I am currently running a machine that is using a recent stock Debian kernel, 2.6.32-3-686, using lilo, *without* the large-memory option, and it loads and runs just fine! (In fairness, I must also add that I use MODULES=dep in /etc/initramfs-tools/conf.d/driver-policy. I see no reason to put stuff in my initial RAM file system that I don't need.) -- .''`. Stephen Powell : :' : `. `'` `- -- To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org Archive: http://lists.debian.org/1181465194.292533.1275667466658.JavaMail.root(a)md01.wow.synacor.com |