UEFI bootloaders (was: Re: reasons to ditch LILO before upgrading to jessie?)
There has been some mention of booting UEFI systems in this thread. This appears to be a comprehensive resource: http://www.rodsbooks.com/efi-bootloaders/index.html Many bootloaders are covered, and the author also mentions his own project, rEFInd http://www.rodsbooks.com/refind/ cheers, -- Joel Roth
Re: reasons to ditch LILO before upgrading to jessie?
On Sun 10 Jul 2016 at 08:38:15 -0500, Richard Owlett wrote: > On 7/9/2016 4:00 PM, Stephen Powell wrote: > >[snip] > >>What I'd like to find which I've had no luck with so far, is finding a > >>Debian > >>installer cmdline option to skip the waste of time that is installation of > >>any bootloader. My disks get generic MBR code and Grub installed by me > >>before > >>any OS gets installed. Thus, I have no need to see warnings about blocklists > >>and unreliability from installers trying to do what I don't want or need > >>done > > > >If you do the installation in expert mode, you can skip the step to install a > >boot loader. But that's in interactive mode. I've never done an automated > >installation, so I don't know what can and cannot be done in that > >environment. > > > > Preseed.cfg apparently can do everything except "walk the dog". > [Well really customized partitioning is annoying. But that is documented.] > I tend to do multiple installs as I am experimenting with having Debian to > do things in *MY* idiosyncratic way. I started with using EXPERT mode but > found repetitive typing annoying. An appropriately edited preseed.cfg allows > me to automate all options except the particular feature(s) of interest. > > My only problem is finding a *SINGLE* references which lists *ALL* the > choices which can be preseeded. > > I find Debian documentation to frequently resemble the early _CPM-80 > Manual_. It's all there but finding it can me daunting. The template files in the udebs are the definitive source for preseed options. Some of them interact with each, requiring care and testing in the writing of a reference guide. What Felix Miata and Stephen Powell want is d-i grub-installer/skip boolean true and d-i lilo-installer/skip boolean true in a preseed.cfg. It can also be done on the kernel line when booting the installer; the Manual has details for this.
Re: reasons to ditch LILO before upgrading to jessie?
On 7/9/2016 4:00 PM, Stephen Powell wrote: [snip] What I'd like to find which I've had no luck with so far, is finding a Debian installer cmdline option to skip the waste of time that is installation of any bootloader. My disks get generic MBR code and Grub installed by me before any OS gets installed. Thus, I have no need to see warnings about blocklists and unreliability from installers trying to do what I don't want or need done If you do the installation in expert mode, you can skip the step to install a boot loader. But that's in interactive mode. I've never done an automated installation, so I don't know what can and cannot be done in that environment. Preseed.cfg apparently can do everything except "walk the dog". [Well really customized partitioning is annoying. But that is documented.] I tend to do multiple installs as I am experimenting with having Debian to do things in *MY* idiosyncratic way. I started with using EXPERT mode but found repetitive typing annoying. An appropriately edited preseed.cfg allows me to automate all options except the particular feature(s) of interest. My only problem is finding a *SINGLE* references which lists *ALL* the choices which can be preseeded. I find Debian documentation to frequently resemble the early _CPM-80 Manual_. It's all there but finding it can me daunting.
Re: reasons to ditch LILO before upgrading to jessie?
On Sun, Jul 10, 2016, at 03:31, Pascal Hambourg wrote: > > AFAICS, elilo is not available any more in stretch and sid. > I'm sorry to hear that. I don't have any UEFI-based systems right now, so it's not an issue for me -- yet. But it may be someday. On the other hand, CSM-less UEFI systems may fail in the marketplace. There have been attempts in the past to eliminate 16-bit support which failed. We'll just have to wait and see. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
On Sun 10 Jul 2016 at 10:31:38 +0200, to...@tuxteam.de wrote: > On Sat, Jul 09, 2016 at 11:15:08PM +0100, Brian wrote: > > [...] > > > This is a common misconception. Debian is about providing the best free > > operating system possible. > > In your humble opinion. > > (sorry, I know I'm feeding it, but I couldn't resist). Not just my opinion. Please see http://upsilon.cc/~zack/talks/2012/20121113-orange.pdf On page 1: Debian: a Free Operating System --- and its Relationships with the Outer World -- Stefano Zacchiroli Debian Project Leader 13 November 2012 Paris, France Orange — Open Source Group 2nd annual seminar On page 5: Debian: once upon a time 1st major distro developed “openly in the spirit of GNU” On page 6: 1/3 of Debian: the operating system --- flagship product: Debian stable Finally, on page 9: 1/3 of Debian: the Project -- Common goal: Create the best, Free operating system. Is this authoritative enough?
Re: reasons to ditch LILO before upgrading to jessie?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Sat, Jul 09, 2016 at 11:15:08PM +0100, Brian wrote: [...] > This is a common misconception. Debian is about providing the best free > operating system possible. In your humble opinion. (sorry, I know I'm feeding it, but I couldn't resist). - -- t -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) iEYEARECAAYFAleCB+oACgkQBcgs9XrR2kYwjgCffaje/B2wKNujDBfGmh8N7Zay heUAn0yq5lqiJiSq19VDhj08Gwzhybda =cj6k -END PGP SIGNATURE-
Re: reasons to ditch LILO before upgrading to jessie?
Le 09/07/2016 à 22:41, Stephen Powell a écrit : So I'm not concerned about it's maintenance status. As long as there are PCs with a BIOS, or a CSM, lilo will remain usable. If the BIOS/CSM goes, lilo goes with it. lilo can't function without a BIOS/CSM. But for UEFI-only systems, there's elilo as a grub alternative. AFAICS, elilo is not available any more in stretch and sid.
Re: reasons to ditch LILO before upgrading to jessie?
Le 09/07/2016 à 22:00, Brian a écrit : All well and good but the installer inexplicably offers a choice between GRUB and LILO. The installer manual is unhelpful on which to choose. A newcomer wouldn't have a clue. We do them no service with this retrograde offering. Get rid of it. What is the point of a choice? Maybe the choice is because GRUB won't install on some very specific setups while LILO will. A boot loader is not just like any other program that you can install at anytime. It must be present for the whole system to start. So it makes sense to include it (and choice for it) in the installation sequence.
Re: reasons to ditch LILO before upgrading to jessie?
Stephen Powell wrote: > As far as LILO being unmaintained is concerned, I wouldn't be too concerned > about that. I've been thinking about offering to maintain it myself. I > haven't > heard from Joachim lately. Maybe I'll drop him another line. I think LILO is an important part of Linux infrastructure. Please consider asking for contributions if you do take over maintenance. You also mentioned elilo. This project appears to be abandoned. https://sourceforge.net/projects/elilo/ Are you aware if there is much in common? (E)LILO for GPT and UEFI would be helpful to Linux minimalists dealing with modern hardware. Thanks for your LILO reference pages! ~Joel > The three main things I like about LILO are (1) I understand how it works, > (2) it is easy to configure, and (3) it doesn't use any unallocated sectors. > > The main limitations to LILO's future viability, as I see it, are UEFI > and GPT. LILO is heavily dependent on the traditional BIOS. Most UEFI > systems have a CSM to provide a BIOS for forward compatibility, though it > is often disabled by default. But some of the newest systems are starting > to ship with no CSM. I won't buy one of those as long as I have a choice. > And GPT is needed to break the 2 TiB disk size limit. But none of my disks > are anywhere near that big. > > If your system has a BIOS and a traditional DOS-style partition table, > there's no reason not to use LILO, unless you just don't want to. > > -- > .''`. Stephen Powell> : :' : > `. `'` >`- > -- Joel Roth
Re: reasons to ditch LILO before upgrading to jessie?
On Sat, Jul 9, 2016, at 18:25, Brian wrote: > On Sat 09 Jul 2016 at 16:41:24 -0400, Stephen Powell wrote: >> >> Long live choice! > > For choice to exist it does not have to be presented as such in the > installer. > Your point is well taken. The installer does not offer choice in everything, just the big things. A choice of desktop, for example. However, even desktop choice is not presented in the installation steps. It's hidden in installer options. For example, I generally install with expert desktop=xfce and then, if I select "Desktop Environment" in tasksel, I get an xfce desktop instead of a gnome3 desktop. Perhaps the boot loader choice could be handled in a similar fashion. Something like expert desktop=xfce bootloader=lilo with the defaults being gnome3 and grub2, respectively. (Dare I suggest adding initsystem=sysvinit as an option too? No, I better not. I don't want to get that flame war going again.) Peace. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
Brian composed on 2016-07-09 21:00 (UTC+0100): ...the installer inexplicably offers a choice between GRUB and LILO. The installer manual is unhelpful on which to choose. A newcomer wouldn't have a clue. We do them no service with this retrograde offering. Get rid of it. Probably a Bad idea. Apparently there are RAID and/or LVM configurations that are incompatible with booting via an inappropriate choice of bootloader. Details I cannot provide, as I use both RAID and Grub2 little, Lilo and LVM not at al, and don't remember of which I've repeatedly read in various help forums. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
On Sat 09 Jul 2016 at 16:41:24 -0400, Stephen Powell wrote: > On Sat, Jul 9, 2016, at 16:00, Brian wrote: > > > > All well and good but the installer inexplicably offers a choice between > > GRUB and LILO. The installer manual is unhelpful on which to choose. A > > newcomer wouldn't have a clue. We do them no service with this retrograde > > offering. Get rid of it. > > > > What is the point of a choice? Just offer GRUB; it is the bootloader for > > Debian and has many advantages over LILO in todayss Linux ecosystem. > > People who have a great desire to use LILO can search it out. > > > > Unmaintained in Debian, The bit-rot starts here. > > I am not a member of the Debian installer team, and I am not authorized to > speak for them. However, I will make the observation that LILO used to be > the default boot loader, indeed the only boot loader at one point, in the > Debian installer for i386/amd64. I suspect that LILO has been retained as > an option in the Debian installer for that reason. Historical reasons for doing something aren't necessarily bad reasons but there comes a time when a reassessment of what goes into the installer has to be made. For example, Stretch drops quite a few of what in Jessie are Standard utilities. Not on technical grounds but because of their usefulness to most users. > The lilo package is maintained in Debian. It's maintainer is Joachim Wiedorn. > He is also the upstream maintainer. He has ceased active development of > lilo, but I believe he still accepts bug reports. And if he wants rid of it, > I know a couple of people who are interested in taking it over, myself > included. > So I'm not concerned about it's maintenance status. As long as there are > PCs with a BIOS, or a CSM, lilo will remain usable. If the BIOS/CSM goes, > lilo goes with it. lilo can't function without a BIOS/CSM. But for UEFI-only > systems, there's elilo as a grub alternative. That's a good reason for keeping LILO in Debian. I would hope it doesn't disappear. > Long live choice! For choice to exist it does not have to be presented as such in the installer.
Re: reasons to ditch LILO before upgrading to jessie?
On Sat 09 Jul 2016 at 22:05:45 +0200, Erwan David wrote: > Le 09/07/2016 à 22:00, Brian a écrit : > > > > What is the point of a choice? Just offer GRUB; it is the bootloader for > > Debian and has many advantages over LILO in todayss Linux ecosystem. > > People who have a great desire to use LILO can search it out. > > > > Unmaintained in Debian, The bit-rot starts here. > > > > What is the point of a choice, just use the windows provided with your PC... > > Linux and debian is just about choice given to the user to use what > he/she thinks is better for him/her. This is a common misconception. Debian is about providing the best free operating system possible. As for choice in the installer, which was the point I made and which you snipped: where is the choice of printing system, the choice of editor, the choice of standard utilites such as netcat etc. Choice doesn't exist. Ok, it does but you have to search it out and get it for yourself if you want it at installation time. On the other hand, LILO is presented as an alternative to the default, well-supported and competent GRUB. No reason given; it's just there, cluttering up the menu and most likely adding to the confusion of users who use the expert install. Let those users who want it seek it out.
Re: reasons to ditch LILO before upgrading to jessie?
On Sat, Jul 9, 2016, at 16:33, Felix Miata wrote: > > All that's well and good, but I see nothing there that equates to my > understanding of the meaning of "editing", which includes removal as well as > appending. Oh, I see what you're saying. Well, the Linux kernel generally does it's own overriding. For example, if the kernel command line contains conflicting options, such as "ro" and "rw", the last one supplied takes precedence. There are some exceptions. For example, if the "console" parameter is specified twice, both consoles are used. But, strictly speaking, you're right. LILO appends options supplied on the command line to options specified in the "append" configuration file statement, it does not replace them. > > What I'd like to find which I've had no luck with so far, is finding a Debian > installer cmdline option to skip the waste of time that is installation of > any bootloader. My disks get generic MBR code and Grub installed by me before > any OS gets installed. Thus, I have no need to see warnings about blocklists > and unreliability from installers trying to do what I don't want or need done If you do the installation in expert mode, you can skip the step to install a boot loader. But that's in interactive mode. I've never done an automated installation, so I don't know what can and cannot be done in that environment. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
On Sat, Jul 9, 2016, at 16:00, Brian wrote: > > All well and good but the installer inexplicably offers a choice between > GRUB and LILO. The installer manual is unhelpful on which to choose. A > newcomer wouldn't have a clue. We do them no service with this retrograde > offering. Get rid of it. > > What is the point of a choice? Just offer GRUB; it is the bootloader for > Debian and has many advantages over LILO in todayss Linux ecosystem. > People who have a great desire to use LILO can search it out. > > Unmaintained in Debian, The bit-rot starts here. > I am not a member of the Debian installer team, and I am not authorized to speak for them. However, I will make the observation that LILO used to be the default boot loader, indeed the only boot loader at one point, in the Debian installer for i386/amd64. I suspect that LILO has been retained as an option in the Debian installer for that reason. The lilo package is maintained in Debian. It's maintainer is Joachim Wiedorn. He is also the upstream maintainer. He has ceased active development of lilo, but I believe he still accepts bug reports. And if he wants rid of it, I know a couple of people who are interested in taking it over, myself included. So I'm not concerned about it's maintenance status. As long as there are PCs with a BIOS, or a CSM, lilo will remain usable. If the BIOS/CSM goes, lilo goes with it. lilo can't function without a BIOS/CSM. But for UEFI-only systems, there's elilo as a grub alternative. Long live choice! -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
Erwan David composed on 2016-07-09 22:05 (UTC+0200): Brian composed: What is the point of a choice? Just offer GRUB; it is the bootloader for Debian... What is the point of a choice, just use the windows provided with your PC... :-D Linux and debian is just about choice given to the user to use what he/she thinks is better for him/her. +++ -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
Stephen Powell composed on 2016-07-09 13:19 (UTC-0400): Felix Miata wrote: Stephen Powell composed on 2016-07-09 08:58 (UTC-0400): As for features, LILO has all the features that I need. One feature it never acquired AFAIK, which Grub shares with Syslinux, is the ^ ability to edit the kernel cmdline at boot time, before kernel load. With problematic hardware, problematic BIOS, and pre-release kernel and distro versions, that ability is a big troubleshooting convenience. It's one of the features that facilitated my decision to migrate from OS/2 to Linux as primary OS. Not true. I use the traditional text-mode interface of LILO (install=text). To supply kernel options during boot, press the Shift key (by itself) before the "delay" timer expires to get a boot prompt ("boot:"). Then type the label of the kernel you want followed by the desired boot parameters. For example, Linux single to boot the kernel in single-user mode. Or Linux forcepae to get a PAE-requiring kernel to boot on a Banias-class Pentium M or Celeron M processor, if you forgot to specify append="forcepae" in /etc/lilo.conf before running lilo. If you can't remember the names of your kernel labels, press the Tab key at a "boot:" prompt. LILO will display the names of your kernel labels followed by another "boot:" prompt. All that's well and good, but I see nothing there that equates to my understanding of the meaning of "editing", which includes removal as well as appending. I've never used the menu-based interface of LILO, but I'm sure that there is a way to supply kernel options at boot time with the menu-based interface as well. What I wrote probably assumed too much of the reader, as my use of Grub to boot virtually always begins in its Gfxboot incarnation, and on a multi-boot PC with various incarnations of pre-release OS installations. I keep Gfxboot configured in most cases to place the entirety of kernel appendages for the selected stanza in edit mode at the outset, so that I know exactly what to expect before proceeding. See my LILO web page at http://www.stevesdebianstuff.org/lilo.htm for more information. P.S. I used to use OS/2 as well. But I switched because OS/2, after Warp 4, was more or less abandoned by IBM. Besides, Linux is free. If I had known about Linux back then, I would probably have gone straight from DOS to Linux. More recent incarnations of OS/2 remain the best host for a DOS application I still depend on constantly. Some things never get improved upon, or if they do, the "improvements" don't add meaningful value for every user context, or add complexity that outweighs putative benefit. e.g. for some, 64 bit, which is killing off 32 bit in more and more distros. Many don't need more power than it takes to run apps that can do everything needed of them in as little as an 8 or 16 bit OS, much less 32 or 64. I'm glad to see Lilo remains available for those who remain content to use it. Debian's Grub (Legacy) is too broken for use on installations with EXT4 filesystems even without 64 bit nodes, while openSUSE's keeps on keeping on pleasing me, needing no scripts to maintain to my liking. What I'd like to find which I've had no luck with so far, is finding a Debian installer cmdline option to skip the waste of time that is installation of any bootloader. My disks get generic MBR code and Grub installed by me before any OS gets installed. Thus, I have no need to see warnings about blocklists and unreliability from installers trying to do what I don't want or need done in the first place. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
Le 09/07/2016 à 22:00, Brian a écrit : > > What is the point of a choice? Just offer GRUB; it is the bootloader for > Debian and has many advantages over LILO in todayss Linux ecosystem. > People who have a great desire to use LILO can search it out. > > Unmaintained in Debian, The bit-rot starts here. > What is the point of a choice, just use the windows provided with your PC... Linux and debian is just about choice given to the user to use what he/she thinks is better for him/her.
Re: reasons to ditch LILO before upgrading to jessie?
On Sat 09 Jul 2016 at 13:19:08 -0400, Stephen Powell wrote: > On Sat, Jul 9, 2016, at 10:53, Felix Miata wrote: > > Stephen Powell composed on 2016-07-09 08:58 (UTC-0400): > > > >> As for features, LILO has all the features that I need. > > > > One feature it never acquired AFAIK, which Grub shares with Syslinux, is > > the > > ability to edit the kernel cmdline at boot time, before kernel load. With > > problematic hardware, problematic BIOS, and pre-release kernel and distro > > versions, that ability is a big troubleshooting convenience. It's one of > > the > > features that facilitated my decision to migrate from OS/2 to Linux as > > primary OS. > > Not true. I use the traditional text-mode interface of LILO (install=text). > To supply kernel options during boot, press the Shift key (by itself) before > the "delay" timer expires to get a boot prompt ("boot:"). Then type the > label of the kernel you want followed by the desired boot parameters. > For example, > >Linux single > > to boot the kernel in single-user mode. Or > >Linux forcepae > > to get a PAE-requiring kernel to boot on a Banias-class Pentium M or > Celeron M processor, if you forgot to specify > >append="forcepae" > > in /etc/lilo.conf before running lilo. If you can't remember the names of > your kernel labels, press the Tab key at a "boot:" prompt. LILO will > display the names of your kernel labels followed by another "boot:" prompt. > > I've never used the menu-based interface of LILO, but I'm sure that there > is a way to supply kernel options at boot time with the menu-based interface > as well. > > See my LILO web page at > >http://www.stevesdebianstuff.org/lilo.htm > > for more information. All well and good but the installer inexplicably offers a choice between GRUB and LILO. The installer manual is unhelpful on which to choose. A newcomer wouldn't have a clue. We do them no service with this retrograde offering. Get rid of it. What is the point of a choice? Just offer GRUB; it is the bootloader for Debian and has many advantages over LILO in todayss Linux ecosystem. People who have a great desire to use LILO can search it out. Unmaintained in Debian, The bit-rot starts here.
Re: reasons to ditch LILO before upgrading to jessie?
On Sat, Jul 9, 2016, at 10:53, Felix Miata wrote: > Stephen Powell composed on 2016-07-09 08:58 (UTC-0400): > >> As for features, LILO has all the features that I need. > > One feature it never acquired AFAIK, which Grub shares with Syslinux, is the > ability to edit the kernel cmdline at boot time, before kernel load. With > problematic hardware, problematic BIOS, and pre-release kernel and distro > versions, that ability is a big troubleshooting convenience. It's one of the > features that facilitated my decision to migrate from OS/2 to Linux as > primary OS. Not true. I use the traditional text-mode interface of LILO (install=text). To supply kernel options during boot, press the Shift key (by itself) before the "delay" timer expires to get a boot prompt ("boot:"). Then type the label of the kernel you want followed by the desired boot parameters. For example, Linux single to boot the kernel in single-user mode. Or Linux forcepae to get a PAE-requiring kernel to boot on a Banias-class Pentium M or Celeron M processor, if you forgot to specify append="forcepae" in /etc/lilo.conf before running lilo. If you can't remember the names of your kernel labels, press the Tab key at a "boot:" prompt. LILO will display the names of your kernel labels followed by another "boot:" prompt. I've never used the menu-based interface of LILO, but I'm sure that there is a way to supply kernel options at boot time with the menu-based interface as well. See my LILO web page at http://www.stevesdebianstuff.org/lilo.htm for more information. P.S. I used to use OS/2 as well. But I switched because OS/2, after Warp 4, was more or less abandoned by IBM. Besides, Linux is free. If I had known about Linux back then, I would probably have gone straight from DOS to Linux. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
Stephen Powell composed on 2016-07-09 08:58 (UTC-0400): As for features, LILO has all the features that I need. One feature it never acquired AFAIK, which Grub shares with Syslinux, is the ability to edit the kernel cmdline at boot time, before kernel load. With problematic hardware, problematic BIOS, and pre-release kernel and distro versions, that ability is a big troubleshooting convenience. It's one of the features that facilitated my decision to migrate from OS/2 to Linux as primary OS. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
On Thu, Jul 7, 2016, at 20:53, Felix Miata wrote: > Stephen Powell composed on 2016-07-07 20:30 (UTC-0400): > > > If your system has a BIOS and a traditional DOS-style partition table, > > there's no reason not to use LILO, unless you just don't want to. > > Or, if you like to be able to boot without hunting down rescue media even > though you forgot to "rerun" some configuration utility after a kernel > upgrade. Once you understand how Grub's shell works, which includes command > history and tab completion much the same as *ash, it is simple (maybe not so > much in Grub2, which I don't use, as in Grub, which accounts for about 97% of > my bootloader installations). Grub's shell has built-in help. With Grub, > there's no reason to be scared when an expected boot menu doesn't show up. > ... I've been using LILO for about 16 years, and I've never had to use rescue media because I forgot to run lilo. Modern Debian systems make sure that lilo gets run when it needs to be run. As for features, LILO has all the features that I need. Chiefly, it has the feature of being able to load the Linux kernel into storage (and it's initial RAM file system image) and transfer control to it. It does one thing, and it does that one thing very well: it loads the Linux kernel. I believe in the KISS philosophy (Keep It Simple, Stupid). If you're happy with grub-legacy, great. I'm happy for you. Keep using it. I'm happy with LILO, and I intend to keep using it. And apparently, the OP is happy with LILO too; and there's no reason, at least at this point, why *he* shouldn't keep using it. Inside every large program is a small program struggling to get out -- Hoare's law of large programs. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
On 08/07/16 07:06 PM, Brian wrote: On Fri 08 Jul 2016 at 18:13:01 -0400, Gary Dale wrote: On 08/07/16 02:19 PM, Brian wrote: If you have some way of easily adjusting files in /etc/grub.d to the needs of a user I wish you would say. So that's the problem. You never took the time to RTFM. See https://help.ubuntu.com/community/Grub2/Setup#File_Structure That page, no matter how informative it may be, is not the manual. My comment was intended to elicit some specific technical advice on how to adjust an /etc/grub.d file(s) to perform a particular task. Something like the topic in https://lists.debian.org/debian-user/2016/07/msg00278.html Using labels in a hand written grub.cfg is a snap. Stopping it being overwritten is another snap. Perhaps altering file X in grub.d is just as easy and unstressful. Maybe one day I will read 'pinfo grub'. :) True enough, but it shows how much good information there is out there. As for altering the correct files, it is just as easy and unstressful. same change either way but the grub template file won't get clobbered so there is no need to run dpkg-divert. It's not the same change. ("grub template file". What is that? There isn't one as far as I can see). Again, RTFM. There are lots of files for you to tinker with in /etc/grub.d. You can even add your own and it's guaranteed not be tampered with by any updates. So there are. Users should stay away from them (40_custom excepted) unless they know what they are doing. Same is true for altering any file. Moreover, you don't need to update a system manual to note that you are diverting a package. You just need to note that a single file is not the package maintainers default - something you have to do either way. Diverting is a Debian thing. The GRUB manual would never mention it. No, but you do have to document each system that you are running. If you are doing anything that is not standard, it has to be recorded so that the next person (or you 3 months later) will know that you've done it. I gather from your comment that you aren't doing this. That's asking for trouble, especially if you are running custom configurations and protecting them with diversions. Complexity added onto complexity without documentation is a good recipe for disaster. I cannot argue with that.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 18:13:01 -0400, Gary Dale wrote: > >>On 08/07/16 02:19 PM, Brian wrote: > > > >If you have some way of easily adjusting files in /etc/grub.d to the > >needs of a user I wish you would say. > So that's the problem. You never took the time to RTFM. See > https://help.ubuntu.com/community/Grub2/Setup#File_Structure That page, no matter how informative it may be, is not the manual. My comment was intended to elicit some specific technical advice on how to adjust an /etc/grub.d file(s) to perform a particular task. Something like the topic in https://lists.debian.org/debian-user/2016/07/msg00278.html Using labels in a hand written grub.cfg is a snap. Stopping it being overwritten is another snap. Perhaps altering file X in grub.d is just as easy and unstressful. Maybe one day I will read 'pinfo grub'. :) > >>same change either way but the grub template file won't get clobbered so > >>there is no need to run dpkg-divert. > >It's not the same change. ("grub template file". What is that? There > >isn't one as far as I can see). > Again, RTFM. There are lots of files for you to tinker with in /etc/grub.d. > You can even add your own and it's guaranteed not be tampered with by any > updates. So there are. Users should stay away from them (40_custom excepted) unless they know what they are doing. > >>Moreover, you don't need to update a system manual to note that you are > >>diverting a package. You just need to note that a single file is not the > >>package maintainers default - something you have to do either way. > >Diverting is a Debian thing. The GRUB manual would never mention it. > > > No, but you do have to document each system that you are running. If you are > doing anything that is not standard, it has to be recorded so that the next > person (or you 3 months later) will know that you've done it. > > I gather from your comment that you aren't doing this. That's asking for > trouble, especially if you are running custom configurations and protecting > them with diversions. Complexity added onto complexity without documentation > is a good recipe for disaster. I cannot argue with that.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 16:57:30 -0500, David Wright wrote: > On Fri 08 Jul 2016 at 21:16:00 (+0100), Brian wrote: > > > Stop moaning. Do it or file file a bug, Then stop moaning and do it. > > I'm the person without a complaint about Grub2, not the one moaning. Apologies. I was intending to talk in general but it looks as though my remark was directed solely at you. That wasn't my intent but I can see how it looks that way. Time to avoid controversy and try to stick to technical points. :)
Re: reasons to ditch LILO before upgrading to jessie?
On 08/07/16 03:51 PM, Brian wrote: On Fri 08 Jul 2016 at 15:08:21 -0400, Gary Dale wrote: On 08/07/16 02:19 PM, Brian wrote: On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: On 07/07/16 05:12 PM, Brian wrote: On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: On 07/07/16 02:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do edit it, the changes will be overwritten the next time a debian upgrade automatically regenerates it. The only method for preserving your changes is to update the grub templates then run update-grub. No it's not. dpkg-divert. That's sufficient to search the list archives for something which has been mentioned a few times but has passed you by. A lot of trouble for something that can be avoided if you just edit the correct files in the first place. Let's see. You write your own grub.cfg or edit the existing one. You want to preserve your file from being changed so you use a *one line command* to ensure that. But this one line command is a lot of trouble. A one line command is onerous? It is much easier to edit the unspecified "correct files" to stop any changes to grub.cfg at a Debian upgrade which attempts to regenerate it? One lives and learns. Let's see then. You can edit a grub template file or grub.cfg. You make the No. You edit or devise a grub,cfg. GRUB template files are not involved. They are irrelevant. You can ignore them because you are doing you own thing. You have decided what *you* want, not what /etc/grub.d wants to give you. If you have some way of easily adjusting files in /etc/grub.d to the needs of a user I wish you would say. So that's the problem. You never took the time to RTFM. See https://help.ubuntu.com/community/Grub2/Setup#File_Structure same change either way but the grub template file won't get clobbered so there is no need to run dpkg-divert. It's not the same change. ("grub template file". What is that? There isn't one as far as I can see). Again, RTFM. There are lots of files for you to tinker with in /etc/grub.d. You can even add your own and it's guaranteed not be tampered with by any updates. Moreover, you don't need to update a system manual to note that you are diverting a package. You just need to note that a single file is not the package maintainers default - something you have to do either way. Diverting is a Debian thing. The GRUB manual would never mention it. No, but you do have to document each system that you are running. If you are doing anything that is not standard, it has to be recorded so that the next person (or you 3 months later) will know that you've done it. I gather from your comment that you aren't doing this. That's asking for trouble, especially if you are running custom configurations and protecting them with diversions. Complexity added onto complexity without documentation is a good recipe for disaster.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 21:16:00 (+0100), Brian wrote: > On Fri 08 Jul 2016 at 14:23:01 -0500, David Wright wrote: > > On Fri 08 Jul 2016 at 19:19:00 (+0100), Brian wrote: > > > On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: > > > > On 07/07/16 05:12 PM, Brian wrote: > > > > >On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > > > > >>On 07/07/16 02:55 PM, David Wright wrote: > > > > >>>On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > > > > The big selling feature of Grub over Lilo was that it didn't need to > > > > updated each time you changed something. That fell by the wayside > > > > with Grub 2. Now the big selling feature is that it works with more > > > > than just Linux. > > > > >>>I guess I don't know what you mean by "update". > > > > >>>If I change the contents of grub.cfg, the effect is immediate: > > > > >>>the changes will be seen at the next boot. I don't do anything more. > > > > >>However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If > > > > >>you do > > > > >>edit it, the changes will be overwritten the next time a debian > > > > >>upgrade > > > > >>automatically regenerates it. The only method for preserving your > > > > >>changes is > > > > >>to update the grub templates then run update-grub. > > > > >No it's not. dpkg-divert. That's sufficient to search the list archives > > > > >for something which has been mentioned a few times but has passed you > > > > >by. > > > > > > > > A lot of trouble for something that can be avoided if you just edit the > > > > correct files in the first place. > > > > > > Let's see. You write your own grub.cfg or edit the existing one. You > > > want to preserve your file from being changed so you use a *one line > > > command* to ensure that. But this one line command is a lot of trouble. > > > A one line command is onerous? > > > > > > It is much easier to edit the unspecified "correct files" to stop any > > > changes to grub.cfg at a Debian upgrade which attempts to regenerate > > > it? One lives and learns. > > > > I *think* I've worked out what this rant is all about: the replacement > > of /boot/grub/menu.lst in old Grub by /boot/grub/grub.cfg in Grub2. > > Was that a rant? If so, It has nothing whatsoever to do with GRUB > legacy. I was referring to the threads which started Thu, 7 Jul 2016 14:39:51 -0400 as quoted above. In them, I have argued against the view that something "fell by the wayside" in Grub2 wrt the legacy version. Sorry if you thought that I was accusing yourself of ranting. > If you want GRUB to do what you want you write your own grub,cfg. I do; unconventionally, but it suits me. > Stop moaning. Do it or file file a bug, Then stop moaning and do it. I'm the person without a complaint about Grub2, not the one moaning. Cheers, David.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 14:23:01 -0500, David Wright wrote: > On Fri 08 Jul 2016 at 19:19:00 (+0100), Brian wrote: > > On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: > > > On 07/07/16 05:12 PM, Brian wrote: > > > >On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > > > >>On 07/07/16 02:55 PM, David Wright wrote: > > > >>>On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > > > The big selling feature of Grub over Lilo was that it didn't need to > > > updated each time you changed something. That fell by the wayside > > > with Grub 2. Now the big selling feature is that it works with more > > > than just Linux. > > > >>>I guess I don't know what you mean by "update". > > > >>>If I change the contents of grub.cfg, the effect is immediate: > > > >>>the changes will be seen at the next boot. I don't do anything more. > > > >>However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If > > > >>you do > > > >>edit it, the changes will be overwritten the next time a debian upgrade > > > >>automatically regenerates it. The only method for preserving your > > > >>changes is > > > >>to update the grub templates then run update-grub. > > > >No it's not. dpkg-divert. That's sufficient to search the list archives > > > >for something which has been mentioned a few times but has passed you by. > > > > > > A lot of trouble for something that can be avoided if you just edit the > > > correct files in the first place. > > > > Let's see. You write your own grub.cfg or edit the existing one. You > > want to preserve your file from being changed so you use a *one line > > command* to ensure that. But this one line command is a lot of trouble. > > A one line command is onerous? > > > > It is much easier to edit the unspecified "correct files" to stop any > > changes to grub.cfg at a Debian upgrade which attempts to regenerate > > it? One lives and learns. > > I *think* I've worked out what this rant is all about: the replacement > of /boot/grub/menu.lst in old Grub by /boot/grub/grub.cfg in Grub2. Was that a rant? If so, It has nothing whatsoever to do with GRUB legacy. If you want GRUB to do what you want you write your own grub,cfg. Stop moaning. Do it or file file a bug, Then stop moaning and do it.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 15:08:21 -0400, Gary Dale wrote: > On 08/07/16 02:19 PM, Brian wrote: > >On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: > > > >>On 07/07/16 05:12 PM, Brian wrote: > >>>On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > >>> > On 07/07/16 02:55 PM, David Wright wrote: > >On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > >>The big selling feature of Grub over Lilo was that it didn't need to > >>updated each time you changed something. That fell by the wayside > >>with Grub 2. Now the big selling feature is that it works with more > >>than just Linux. > >I guess I don't know what you mean by "update". > >If I change the contents of grub.cfg, the effect is immediate: > >the changes will be seen at the next boot. I don't do anything more. > However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you > do > edit it, the changes will be overwritten the next time a debian upgrade > automatically regenerates it. The only method for preserving your changes > is > to update the grub templates then run update-grub. > >>>No it's not. dpkg-divert. That's sufficient to search the list archives > >>>for something which has been mentioned a few times but has passed you by. > >>A lot of trouble for something that can be avoided if you just edit the > >>correct files in the first place. > >Let's see. You write your own grub.cfg or edit the existing one. You > >want to preserve your file from being changed so you use a *one line > >command* to ensure that. But this one line command is a lot of trouble. > >A one line command is onerous? > > > >It is much easier to edit the unspecified "correct files" to stop any > >changes to grub.cfg at a Debian upgrade which attempts to regenerate > >it? One lives and learns. > > > Let's see then. You can edit a grub template file or grub.cfg. You make the No. You edit or devise a grub,cfg. GRUB template files are not involved. They are irrelevant. You can ignore them because you are doing you own thing. You have decided what *you* want, not what /etc/grub.d wants to give you. If you have some way of easily adjusting files in /etc/grub.d to the needs of a user I wish you would say. > same change either way but the grub template file won't get clobbered so > there is no need to run dpkg-divert. It's not the same change. ("grub template file". What is that? There isn't one as far as I can see). > Moreover, you don't need to update a system manual to note that you are > diverting a package. You just need to note that a single file is not the > package maintainers default - something you have to do either way. Diverting is a Debian thing. The GRUB manual would never mention it.
Re: reasons to ditch LILO before upgrading to jessie?
On Fri 08 Jul 2016 at 19:19:00 (+0100), Brian wrote: > On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: > > On 07/07/16 05:12 PM, Brian wrote: > > >On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > > >>On 07/07/16 02:55 PM, David Wright wrote: > > >>>On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > > The big selling feature of Grub over Lilo was that it didn't need to > > updated each time you changed something. That fell by the wayside > > with Grub 2. Now the big selling feature is that it works with more > > than just Linux. > > >>>I guess I don't know what you mean by "update". > > >>>If I change the contents of grub.cfg, the effect is immediate: > > >>>the changes will be seen at the next boot. I don't do anything more. > > >>However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you > > >>do > > >>edit it, the changes will be overwritten the next time a debian upgrade > > >>automatically regenerates it. The only method for preserving your changes > > >>is > > >>to update the grub templates then run update-grub. > > >No it's not. dpkg-divert. That's sufficient to search the list archives > > >for something which has been mentioned a few times but has passed you by. > > > > A lot of trouble for something that can be avoided if you just edit the > > correct files in the first place. > > Let's see. You write your own grub.cfg or edit the existing one. You > want to preserve your file from being changed so you use a *one line > command* to ensure that. But this one line command is a lot of trouble. > A one line command is onerous? > > It is much easier to edit the unspecified "correct files" to stop any > changes to grub.cfg at a Debian upgrade which attempts to regenerate > it? One lives and learns. I *think* I've worked out what this rant is all about: the replacement of /boot/grub/menu.lst in old Grub by /boot/grub/grub.cfg in Grub2. So under the old system, you generated or wrote menu.lst and then you could edit it to your heart's content. Under the new system, grub.cfg is always (re)built from a set of scripts and some preference data in /etc/default/grub, unless you've protected it with your one-line command. In case of the latter, it behaves just like menu.lst did: if you edit it, the effect is immediate (and unlike lilo). Somebody else then chimed in about how the partition numbering scheme has been screwed (more like corrected IMHO) which meant people were scratching their heads over "Is it (hd1,0) or (hd0,1)" when the world has moved on to using UUID/LABELs instead if worrying about whether the kernel happened to discover their disks in the right order. Cheers, David.
Re: reasons to ditch LILO before upgrading to jessie?
On 08/07/16 02:19 PM, Brian wrote: On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: On 07/07/16 05:12 PM, Brian wrote: On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: On 07/07/16 02:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do edit it, the changes will be overwritten the next time a debian upgrade automatically regenerates it. The only method for preserving your changes is to update the grub templates then run update-grub. No it's not. dpkg-divert. That's sufficient to search the list archives for something which has been mentioned a few times but has passed you by. A lot of trouble for something that can be avoided if you just edit the correct files in the first place. Let's see. You write your own grub.cfg or edit the existing one. You want to preserve your file from being changed so you use a *one line command* to ensure that. But this one line command is a lot of trouble. A one line command is onerous? It is much easier to edit the unspecified "correct files" to stop any changes to grub.cfg at a Debian upgrade which attempts to regenerate it? One lives and learns. Let's see then. You can edit a grub template file or grub.cfg. You make the same change either way but the grub template file won't get clobbered so there is no need to run dpkg-divert. Moreover, you don't need to update a system manual to note that you are diverting a package. You just need to note that a single file is not the package maintainers default - something you have to do either way.
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 23:34:11 -0400, Gary Dale wrote: > On 07/07/16 05:12 PM, Brian wrote: > >On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > > > >>On 07/07/16 02:55 PM, David Wright wrote: > >>>On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > The big selling feature of Grub over Lilo was that it didn't need to > updated each time you changed something. That fell by the wayside > with Grub 2. Now the big selling feature is that it works with more > than just Linux. > >>>I guess I don't know what you mean by "update". > >>>If I change the contents of grub.cfg, the effect is immediate: > >>>the changes will be seen at the next boot. I don't do anything more. > >>However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do > >>edit it, the changes will be overwritten the next time a debian upgrade > >>automatically regenerates it. The only method for preserving your changes is > >>to update the grub templates then run update-grub. > >No it's not. dpkg-divert. That's sufficient to search the list archives > >for something which has been mentioned a few times but has passed you by. > > A lot of trouble for something that can be avoided if you just edit the > correct files in the first place. Let's see. You write your own grub.cfg or edit the existing one. You want to preserve your file from being changed so you use a *one line command* to ensure that. But this one line command is a lot of trouble. A one line command is onerous? It is much easier to edit the unspecified "correct files" to stop any changes to grub.cfg at a Debian upgrade which attempts to regenerate it? One lives and learns.
Re: reasons to ditch LILO before upgrading to jessie?
Gary Dale composed on 2016-07-07 14:39 (UTC-0400): It also has a "rescue shell" that I've never been able to do anything useful with. When grub fails, I boot from a rescue cd instead. That way I get a real working environment. The Grub shell works the same whether in boot rescue mode or run from a normal boot, a lot like more familiar shells. A little practice under a normal boot should make it easy to understand what to do in rescue mode so that it shouldn't be too intimidating under the pressure of being in actual rescue mode. I find Grub's rescue mode nearly always easier to work through to a normal boot than to locate and boot appropriate rescue media, whose purpose often as not is just to work past a broken boot menu. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
On 07/07/16 05:12 PM, Brian wrote: On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: On 07/07/16 02:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do edit it, the changes will be overwritten the next time a debian upgrade automatically regenerates it. The only method for preserving your changes is to update the grub templates then run update-grub. No it's not. dpkg-divert. That's sufficient to search the list archives for something which has been mentioned a few times but has passed you by. A lot of trouble for something that can be avoided if you just edit the correct files in the first place.
Re: reasons to ditch LILO before upgrading to jessie?
Stephen Powell composed on 2016-07-07 20:30 (UTC-0400): If your system has a BIOS and a traditional DOS-style partition table, there's no reason not to use LILO, unless you just don't want to. Or, if you like to be able to boot without hunting down rescue media even though you forgot to "rerun" some configuration utility after a kernel upgrade. Once you understand how Grub's shell works, which includes command history and tab completion much the same as *ash, it is simple (maybe not so much in Grub2, which I don't use, as in Grub, which accounts for about 97% of my bootloader installations). Grub's shell has built-in help. With Grub, there's no reason to be scared when an expected boot menu doesn't show up. -- "The wise are known for their understanding, and pleasant words are persuasive." Proverbs 16:21 (New Living Translation) Team OS/2 ** Reg. Linux User #211409 ** a11y rocks! Felix Miata *** http://fm.no-ip.com/
Re: reasons to ditch LILO before upgrading to jessie?
On Thu, Jul 7, 2016, at 10:57, Giovanni Gigante wrote: > > At the end, I decided to try the upgrade to jessie with LiLo (24.1) in > place. I thought that the probability of hitting some bug caused by the > interaction between LiLo and the upgraded distribution was less than the > probabily of causing some damage by switching the bootloader (for > example, by messing up the RAID configuration) since I've never > configured Grub before. Also, the golden heuristic of "if it ain't > broke, don't fix it", and the fact that this particular Debian upgrade > is also switching from the init scripts to systemd, so I did not want to > introduce even more changes in the boot process for the moment. > > Apparently, it worked. > I never doubted that it would. I still use LILO with the latest jessie kernels. I maintain a web page for Debian users who use LILO, or who are thinking of switching from GRUB to LILO: http://www.stevesdebianstuff.org/lilo.htm I don't address the issue of software RAID though. I've never tried it. I have used LILO with hardware RAID. It works just fine. I'm using it right now as I write this. As far as LILO being unmaintained is concerned, I wouldn't be too concerned about that. I've been thinking about offering to maintain it myself. I haven't heard from Joachim lately. Maybe I'll drop him another line. The three main things I like about LILO are (1) I understand how it works, (2) it is easy to configure, and (3) it doesn't use any unallocated sectors. The main limitations to LILO's future viability, as I see it, are UEFI and GPT. LILO is heavily dependent on the traditional BIOS. Most UEFI systems have a CSM to provide a BIOS for forward compatibility, though it is often disabled by default. But some of the newest systems are starting to ship with no CSM. I won't buy one of those as long as I have a choice. And GPT is needed to break the 2 TiB disk size limit. But none of my disks are anywhere near that big. If your system has a BIOS and a traditional DOS-style partition table, there's no reason not to use LILO, unless you just don't want to. -- .''`. Stephen Powell: :' : `. `'` `-
Re: reasons to ditch LILO before upgrading to jessie?
On 07/07/2016 05:47 PM, rhkra...@gmail.com wrote: I'll take advantage of this thread to ask a question / express my frustration with grub: The thing that always frustrated me about grub is that, iirc, they counted disks / partitions different than lilo and the rest of Linux--they start counting at 1 (like Windows, iirc), and lilo and Linux start counting at 0--is that still the case? OTOH, I haven't touched either lilo or grub in a long time--I don't even know which I'm using--I installed whatever Debian 7 offered, and, since it is a single boot system, (and I haven't booted in ~85 days), I don't really have an issue. Yes, the partitions are counted differently, and that can be a little frustrating. However, with Grub, you can refer to the partitions by label or by UUID (Grub usually defaults to using UUID) rather than drive/partition numbers, which makes it a little easier. Lilo cannot refer to the partitions in this way, so you must know what drive/partition you are referring to. This is not a big deal in many installations, however, in situations where the drive/partition number may change (e.g. usb sticks that on one boot may be /dev/sdb on one boot and /dev/sdc on another depending on how/when the bios enumerates them) this comes in very handy, as you don't have to know which drive/partition will be the root partition, you only have to know either the UUID of the partition or a label you have assigned to it, and these things don't change regardless of how/where a usb device is enumerated. On Thursday, July 07, 2016 03:37:12 PM Michael Milliman wrote: On 07/07/2016 01:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: -- 73's Mike, WB5VQX
Re: reasons to ditch LILO before upgrading to jessie?
On 07/07/2016 05:47 PM, rhkra...@gmail.com wrote: I'll take advantage of this thread to ask a question / express my frustration with grub: The thing that always frustrated me about grub is that, iirc, they counted disks / partitions different than lilo and the rest of Linux--they start counting at 1 (like Windows, iirc), and lilo and Linux start counting at 0--is that still the case? OTOH, I haven't touched either lilo or grub in a long time--I don't even know which I'm using--I installed whatever Debian 7 offered, and, since it is a single boot system, (and I haven't booted in ~85 days), I don't really have an issue. Yes, the partitions are counted differently, and that can be a little frustrating. However, with Grub, you can refer to the partitions by label or by UUID (Grub usually defaults to using UUID) rather than drive/partition numbers, which makes it a little easier. Lilo cannot refer to the partitions in this way, so you must know what drive/partition you are referring to. This is not a big deal in many installations, however, in situations where the drive/partition number may change (e.g. usb sticks that on one boot may be /dev/sdb on one boot and /dev/sdc on another depending on how/when the bios enumerates them) this comes in very handy, as you don't have to know which drive/partition will be the root partition, you only have to know either the UUID of the partition or a label you have assigned to it, and these things don't change regardless of how/where a usb device is enumerated. On Thursday, July 07, 2016 03:37:12 PM Michael Milliman wrote: On 07/07/2016 01:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: -- 73's Mike, WB5VQX
Re: reasons to ditch LILO before upgrading to jessie?
I'll take advantage of this thread to ask a question / express my frustration with grub: The thing that always frustrated me about grub is that, iirc, they counted disks / partitions different than lilo and the rest of Linux--they start counting at 1 (like Windows, iirc), and lilo and Linux start counting at 0--is that still the case? OTOH, I haven't touched either lilo or grub in a long time--I don't even know which I'm using--I installed whatever Debian 7 offered, and, since it is a single boot system, (and I haven't booted in ~85 days), I don't really have an issue. On Thursday, July 07, 2016 03:37:12 PM Michael Milliman wrote: > On 07/07/2016 01:55 PM, David Wright wrote: > > On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote:
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 15:18:05 -0400, Gary Dale wrote: > On 07/07/16 02:55 PM, David Wright wrote: > >On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > >>The big selling feature of Grub over Lilo was that it didn't need to > >>updated each time you changed something. That fell by the wayside > >>with Grub 2. Now the big selling feature is that it works with more > >>than just Linux. > >I guess I don't know what you mean by "update". > >If I change the contents of grub.cfg, the effect is immediate: > >the changes will be seen at the next boot. I don't do anything more. > However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do > edit it, the changes will be overwritten the next time a debian upgrade > automatically regenerates it. The only method for preserving your changes is > to update the grub templates then run update-grub. No it's not. dpkg-divert. That's sufficient to search the list archives for something which has been mentioned a few times but has passed you by.
Re: reasons to ditch LILO before upgrading to jessie?
On 07/07/2016 01:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. With LILO any time you change the configuration, you have to re-install the LILO boot loader for those changes to take effect. With Grub, you make a change to the configuration file and it is immediately available on the next boot without having to re-install the Grub boot loader. Grub understands the file systems and so can read the configuration directly, whereas LILO does not understand the file system, and so the configuration must be provided as data directly in the loader itself (hence the re-install step). It also has a "rescue shell" that I've never been able to do anything useful with. When grub fails, I boot from a rescue cd instead. That way I get a real working environment. Horses for courses. Cheers, David. -- 73's Mike, WB5VQX
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 15:18:05 (-0400), Gary Dale wrote: > On 07/07/16 02:55 PM, David Wright wrote: > >On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > >>The big selling feature of Grub over Lilo was that it didn't need to > >>updated each time you changed something. That fell by the wayside > >>with Grub 2. Now the big selling feature is that it works with more > >>than just Linux. > >I guess I don't know what you mean by "update". > >If I change the contents of grub.cfg, the effect is immediate: > >the changes will be seen at the next boot. I don't do anything more. > However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If > you do edit it, the changes will be overwritten the next time a > debian upgrade automatically regenerates it. The only method for > preserving your changes is to update the grub templates then run > update-grub. Well, if you want to use templates to build your grub.cfg, then you have to build it with update-grub. That's hardly surprising. If you roll your own, you don't. The reason I wrote 'I don't know what you mean by "update"' is because AIUI if you changed /etc/lilo.conf, you had/have to remember to run /sbin/lilo to reinstall the loader again. With grub, you don't. AFAICT that's the big selling feature you mentioned above, was it not? If not, what was it, and what has changed between grub 1 and 2? Cheers, David.
Re: reasons to ditch LILO before upgrading to jessie?
On 07/07/16 02:55 PM, David Wright wrote: On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. However the second line of grub.cfg says "DO NOT EDIT THIS FILE". If you do edit it, the changes will be overwritten the next time a debian upgrade automatically regenerates it. The only method for preserving your changes is to update the grub templates then run update-grub. It also has a "rescue shell" that I've never been able to do anything useful with. When grub fails, I boot from a rescue cd instead. That way I get a real working environment. Horses for courses. Cheers, David. Glad you find the grub shell useful. I've yet to encounter a boot error that it could fix.
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 14:39:51 (-0400), Gary Dale wrote: > The big selling feature of Grub over Lilo was that it didn't need to > updated each time you changed something. That fell by the wayside > with Grub 2. Now the big selling feature is that it works with more > than just Linux. I guess I don't know what you mean by "update". If I change the contents of grub.cfg, the effect is immediate: the changes will be seen at the next boot. I don't do anything more. > It also has a "rescue shell" that I've never been able to do > anything useful with. When grub fails, I boot from a rescue cd > instead. That way I get a real working environment. Horses for courses. Cheers, David.
Re: reasons to ditch LILO before upgrading to jessie?
On 05/07/16 09:38 AM, Giovanni Gigante wrote: Hello, I am preparing my system for the upgrade from wheezy to jessie. Since ancient ages, this system has been using LILO as the bootloader, because, long ago, it was the only bootloader that was recommended for my setup: this machine has two SATA disks in a software RAID 1 & LVM; that is, in /etc/lilo.conf I have: boot=/dev/md0 root=/dev/mapper/vg00-rootlv raid-extra-boot = mbr My doubt is that I have read that LILO, besides being very old, is now unmantained. However, I see that the jessie installation manual still mentions it, so it does not seem deprecated yet. So the question is: is there any serious reason to switch the system to GRUB before upgrading, or can I just keep my current setup and proceed to jessie? Thanks Giovanni The big selling feature of Grub over Lilo was that it didn't need to updated each time you changed something. That fell by the wayside with Grub 2. Now the big selling feature is that it works with more than just Linux. It also has a "rescue shell" that I've never been able to do anything useful with. When grub fails, I boot from a rescue cd instead. That way I get a real working environment.
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 08:33:57 (+0200), to...@tuxteam.de wrote: > On Wed, Jul 06, 2016 at 02:23:39PM -0500, Don Armstrong wrote: > > On Wed, 06 Jul 2016, Jonathan Dowland wrote: > > > YMMV, I find it impenetrable. > > > > I'm assuming you mean the generated configuration? It's literally just > > some boilerplate [...] > > Let's make it impenetrable boilerplate, then. > > FWIW, I concur with Jonathan. While I actively worked on my lilo.conf, > since grub (espeecially -2), I try to avoid touching the conf. > > Wonderfully general, but (for me) wonderfully unusable. > > Now don't get me wrong: without Grub, my box wouldn't boot, and chances > are that the "legacy" emulation of whatever monster bios is in there > is so buggy that Lilo wouldn't cope: therefore I am still full of thanks > and praise for the Grub authors for their hard work. > > Still I do disagree on many of its design principles. Well, I think it's a shame that there isn't a GRUB_ENABLE_LINUX_LABEL=true along with the GRUB_DISABLE_LINUX_UUID parameter. I did take a look at the scripts with a view to hacking one into shape, but it struck me that it would be harder to maintain it unless it got incorporated upstream than what I eventually did. Even if I succeeded in writing it. So I took the easier course of writing a python script to read /run/udev/data/b* and replace the --fs-uuid UUIDs in /boot/grub/grub.cfg with --label LABELs. Along the way, it also prettifies the menu strings and adds an fsck entry which I can call with grub-reboot 'fsck>fsck' (Again, I don't know why the scripts don't do this themselves.) I keep a before/after copy so that, whenever grub or a kernel is updated, I can copy the 'after' file if the 'before' compares equal. I don't think the scripts are much harder to understand than those in /etc/init.d/ which regularly come up here when people try to hack their own. Cheers, David.
Re: reasons to ditch LILO before upgrading to jessie?
Brian wrote: Giovanni Gigante seems happy enough with LiLo and there appears to be no definite indication that it would fail to boot an upgraded machine. He could consider leaving it in place, reading the bug reports and having a plan to install GRUB should something go wrong afterwards. At the end, I decided to try the upgrade to jessie with LiLo (24.1) in place. I thought that the probability of hitting some bug caused by the interaction between LiLo and the upgraded distribution was less than the probabily of causing some damage by switching the bootloader (for example, by messing up the RAID configuration) since I've never configured Grub before. Also, the golden heuristic of "if it ain't broke, don't fix it", and the fact that this particular Debian upgrade is also switching from the init scripts to systemd, so I did not want to introduce even more changes in the boot process for the moment. Apparently, it worked. Perhaps, in the future, I'll also upgrade the bootloader, one rainy afternoon :-) gg
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 12:44:40 +0200, to...@tuxteam.de wrote: > On Thu, Jul 07, 2016 at 11:32:42AM +0100, Brian wrote: > > On Thu 07 Jul 2016 at 10:35:47 +0100, Lisi Reisz wrote: > > > > > On Thursday 07 July 2016 07:33:57 to...@tuxteam.de wrote: > > > > Let's make it (GRUB2) impenetrable boilerplate, then. > > > > > > :-) +1! > > :-) > > > It doesn't need to be penetrable, does it? > > De gustibus non est disputandum. > > To me, one of the attractions of the whole Free Software thing is its > "penetrability". I see that there's a force field among many poles, > functionality and "getting things done NOW" being two very valid ones > which sometimes are at odds with simplicity. And I readily admit that > preferences are extremely individual. > > > The generated grub.cfg just needs to boot the machine. In any case, > > it can easily be replaced by a more readable five or six line version > > if desired. > > Things get more complicated when you're in the position of "fixing" > it :-) Admittedly, a generated grub.cfg or most of the files in /etc/grub.d are not something I would recommend for bedtime reading. > And please, again: the fact that I disagree with some Grub design > decisions shouldn't detract from the fact that I have utmost > respect for the Grub developers and packagers. Their hard work, > their decisions. Giovanni Gigante seems happy enough with LiLo and there appears to be no definite indication that it would fail to boot an upgraded machine. He could consider leaving it in place, reading the bug reports and having a plan to install GRUB should something go wrong afterwards.
Re: reasons to ditch LILO before upgrading to jessie?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Thu, Jul 07, 2016 at 11:32:42AM +0100, Brian wrote: > On Thu 07 Jul 2016 at 10:35:47 +0100, Lisi Reisz wrote: > > > On Thursday 07 July 2016 07:33:57 to...@tuxteam.de wrote: > > > Let's make it (GRUB2) impenetrable boilerplate, then. > > > > :-) +1! :-) > It doesn't need to be penetrable, does it? De gustibus non est disputandum. To me, one of the attractions of the whole Free Software thing is its "penetrability". I see that there's a force field among many poles, functionality and "getting things done NOW" being two very valid ones which sometimes are at odds with simplicity. And I readily admit that preferences are extremely individual. > The generated grub.cfg just needs to boot the machine. In any case, > it can easily be replaced by a more readable five or six line version > if desired. Things get more complicated when you're in the position of "fixing" it :-) And please, again: the fact that I disagree with some Grub design decisions shouldn't detract from the fact that I have utmost respect for the Grub developers and packagers. Their hard work, their decisions. regards - -- tomás -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) iEUEARECAAYFAld+MpgACgkQBcgs9XrR2kYYXwCdELtvb8dcQ1Zwgy/FO/mbmjRJ riAAljXe1c+7A3aKL2crpeDQBot9hkw= =LS3G -END PGP SIGNATURE-
Re: reasons to ditch LILO before upgrading to jessie?
On Thu 07 Jul 2016 at 10:35:47 +0100, Lisi Reisz wrote: > On Thursday 07 July 2016 07:33:57 to...@tuxteam.de wrote: > > Let's make it (GRUB2) impenetrable boilerplate, then. > > :-) +1! It doesn't need to be penetrable, does it? The generated grub.cfg just needs to boot the machine. In any case, it can easily be replaced by a more readable five or six line version if desired.
Re: reasons to ditch LILO before upgrading to jessie?
On Thursday 07 July 2016 07:33:57 to...@tuxteam.de wrote: > Let's make it (GRUB2) impenetrable boilerplate, then. :-) +1! Lisi
Re: reasons to ditch LILO before upgrading to jessie?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On Wed, Jul 06, 2016 at 02:23:39PM -0500, Don Armstrong wrote: > On Wed, 06 Jul 2016, Jonathan Dowland wrote: > > YMMV, I find it impenetrable. > > I'm assuming you mean the generated configuration? It's literally just > some boilerplate [...] Let's make it impenetrable boilerplate, then. FWIW, I concur with Jonathan. While I actively worked on my lilo.conf, since grub (espeecially -2), I try to avoid touching the conf. Wonderfully general, but (for me) wonderfully unusable. Now don't get me wrong: without Grub, my box wouldn't boot, and chances are that the "legacy" emulation of whatever monster bios is in there is so buggy that Lilo wouldn't cope: therefore I am still full of thanks and praise for the Grub authors for their hard work. Still I do disagree on many of its design principles. Regards - -- tomás -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.12 (GNU/Linux) iEYEARECAAYFAld999QACgkQBcgs9XrR2ka/RQCeP8ml0YqrsOKn6vbbBcj4e0A/ jcQAn1y3dODdJ1u4uPDOm+zOp3MEA4wb =pS2e -END PGP SIGNATURE-
Re: reasons to ditch LILO before upgrading to jessie?
On Wed, 06 Jul 2016, Jonathan Dowland wrote: > YMMV, I find it impenetrable. I'm assuming you mean the generated configuration? It's literally just some boilerplate for fancy splash screens, and then menu entries. Each entry containing appropriate module loading, root configuration, kernel and initrd loading, and then boot commands. All of which is generated directly from /etc/grub.d/ and is pretty well documented[1]. It's the same set of commands you can use to load almost any kernel from almost any device almost anywhere on a system (or even the network), so it's pretty useful to learn if you maintain machines which occasionally need to do weird things. 1: https://www.gnu.org/software/grub/manual/html_node/Configuration.html -- Don Armstrong https://www.donarmstrong.com A Bill of Rights that means what the majority wants it to mean is worthless. -- U.S. Supreme Court Justice Antonin Scalia
Re: reasons to ditch LILO before upgrading to jessie?
On Tue, Jul 05, 2016 at 04:46:32PM -0500, Don Armstrong wrote: > On Tue, 05 Jul 2016, Marc Shapiro wrote: > > I finally switched to Jessie (but still using SysV Init) a few months > > ago. This box and its predecessors have uses lilo (and SysV Init) > > since Bo was a pup. I have yet to see any real reason to switch from > > lilo to grub. I have never had a problem with lilo and I like having a > > config that I can directly control and understand. > > Grub's configuration is pretty simple; /etc/default/grub is pretty > straight forward, and the resultant configuration is also pretty easy to > understand as well. YMMV, I find it impenetrable. > There's a reason why no one is interested in maintaining lilo anymore. And another why some look for alternatives to Grub :) signature.asc Description: Digital signature
Re: reasons to ditch LILO before upgrading to jessie?
On Tue, 05 Jul 2016, Marc Shapiro wrote: > I finally switched to Jessie (but still using SysV Init) a few months > ago. This box and its predecessors have uses lilo (and SysV Init) > since Bo was a pup. I have yet to see any real reason to switch from > lilo to grub. I have never had a problem with lilo and I like having a > config that I can directly control and understand. Grub's configuration is pretty simple; /etc/default/grub is pretty straight forward, and the resultant configuration is also pretty easy to understand as well. Grub also has a lot more features than lilo has, including the ability to boot from cd images (useful for updating bios), proper serial support, the ability to boot any kernel even if it isn't listed in the config, the ability to handle raid5 and raid6 filesystems, lvm, support for UEFI, encrypted /boot, network booting, fallback support, etc. There's a reason why no one is interested in maintaining lilo anymore. -- Don Armstrong https://www.donarmstrong.com He no longer wished to be dead. At the same time, it cannot be said that he was glad to be alive. But at least he did not resent it. He was alive, and the stubbornness of this fact had little by little begun to fascinate him -- as if he had managed to outlive himself, as if he were somehow living a posthumous life. -- Paul Auster _City of Glass_
Re: reasons to ditch LILO before upgrading to jessie?
On 07/05/2016 10:10 AM, Don Armstrong wrote: On Tue, 05 Jul 2016, Giovanni Gigante wrote: I am preparing my system for the upgrade from wheezy to jessie. Since ancient ages, this system has been using LILO as the bootloader, because, long ago, it was the only bootloader that was recommended for my setup: this machine has two SATA disks in a software RAID 1 & LVM; that is, in /etc/lilo.conf I have: boot=/dev/md0 root=/dev/mapper/vg00-rootlv raid-extra-boot = mbr My doubt is that I have read that LILO, besides being very old, is now unmantained. However, I see that the jessie installation manual still mentions it, so it does not seem deprecated yet. So the question is: is there any serious reason to switch the system to GRUB before upgrading, or can I just keep my current setup and proceed to jessie? There's always the possibility that you'll discover a new bug with lilo and newer kernels which no one else has seen, but that's probably fairly unlikely. In my experience, grub now works way more reliably than lilo ever did, and it's worth switching. [I switched over *years* ago for precisely this reason.] But your experience may vary. I finally switched to Jessie (but still using SysV Init) a few months ago. This box and its predecessors have uses lilo (and SysV Init) since Bo was a pup. I have yet to see any real reason to switch from lilo to grub. I have never had a problem with lilo and I like having a config that I can directly control and understand. As Giovanni says, however, YMMV. Marc
Re: reasons to ditch LILO before upgrading to jessie?
On Tue, 05 Jul 2016, Giovanni Gigante wrote: > I am preparing my system for the upgrade from wheezy to jessie. > Since ancient ages, this system has been using LILO as the bootloader, > because, long ago, it was the only bootloader that was recommended for my > setup: this machine has two SATA disks in a software RAID 1 & LVM; that is, > in /etc/lilo.conf I have: > > boot=/dev/md0 > root=/dev/mapper/vg00-rootlv > raid-extra-boot = mbr > > My doubt is that I have read that LILO, besides being very old, is now > unmantained. However, I see that the jessie installation manual still > mentions it, so it does not seem deprecated yet. So the question is: > is there any serious reason to switch the system to GRUB before > upgrading, or can I just keep my current setup and proceed to jessie? There's always the possibility that you'll discover a new bug with lilo and newer kernels which no one else has seen, but that's probably fairly unlikely. In my experience, grub now works way more reliably than lilo ever did, and it's worth switching. [I switched over *years* ago for precisely this reason.] But your experience may vary. -- Don Armstrong https://www.donarmstrong.com 6: If we are one, then we can defeat 2. -- "The Prisoner (2009 Miniseries)" _Schizoid_
reasons to ditch LILO before upgrading to jessie?
Hello, I am preparing my system for the upgrade from wheezy to jessie. Since ancient ages, this system has been using LILO as the bootloader, because, long ago, it was the only bootloader that was recommended for my setup: this machine has two SATA disks in a software RAID 1 & LVM; that is, in /etc/lilo.conf I have: boot=/dev/md0 root=/dev/mapper/vg00-rootlv raid-extra-boot = mbr My doubt is that I have read that LILO, besides being very old, is now unmantained. However, I see that the jessie installation manual still mentions it, so it does not seem deprecated yet. So the question is: is there any serious reason to switch the system to GRUB before upgrading, or can I just keep my current setup and proceed to jessie? Thanks Giovanni