I've just completed an installation of AntiX on an old computer that I'm going to give to a friend. I found a couple of problems during the installation and I wonder if there is a quick way of reporting these to the AntiX team without having to subscribe to the AntiX forum. I don't want to belong to too many forums or I'll never get any work done!

This is antiX-16 (Berta Cáceres), base edition. It boots to a graphical desktop with an install icon. I clicked on it but didn't get very far; it crashed out while copying the files over. I don't know exactly where because I wasn't watching the screen at the time. When I next looked up, the install window was gone.

There is also a cli-installer so I switched to that and found it much more robust and quite easy to use. It ran to completion and reported no errors. But afterwards I couldn't boot from the hard drive. There was simply no sign of a bootloader although I had requested for grub to be installed in the mbr. This clearly hadn't happened. I found all the grub files in their usual places, so the grub package had been correctly installed but grub itself had not.

It took me quite a while to work out what had gone wrong. In the end I chrooted in and ran grub-install by hand and after that I could reboot normally. God knows what a newbie would have done!

Using a USB?

Has been an issue with the Debian installer on certain hardware from usb install media - the installer has trouble determining the correct mbr automatically.
It works if you choose to specify the device to install to (provided you get it right ).

I was installing from a cdrom on the built-in cd reader (IDE interface). The hard drive is SATA. I'm pretty sure I specified /dev/sda but I can't prove it!

I'm bumping this thread because I've recently done another installation on a different machine and with a different image. This time it was 32-bit AntiX-base (the other one was 64-bit). This is a very old machine; you can read the details in the General thread on old PCs. I knew it wouldn't be able to cope with a graphical install so I used the cli installer straight away. Again the installation device was a cdrom.

In view of what Descendant Command said above, I made sure this time that there were no usb devices present. In the other case, there was a wifi dongle plugged in, and I know that these register initially as storage devices so it is just possible that GRUB got installed there by mistake. This time around, the only writeable drive was the hard drive.

Everything went smoothly and, when it came to installing GRUB, I specified the MBR. But when I rebooted, I got LILO, which was what was there before. I had intended to boot another partition, which had Slackware on it, and then chroot over to AntiX and install GRUB. But from force of habit, I booted AntiX. To my surprise the hand-rolled kernel I had made for my previous AntiX came up. Obviously it hadn't been overwritten by the new install and LILO still had the address. So in the end, I didn't have to chroot and I was able to install GRUB successfully.

Oscar Wilde wrote that once could be a misfortune but twice looks like carelessness! I have now done this installation twice on two different machines with two different architectures and both times I hit the same bug. In this second case it could not be confusion between the hard drive and a usb device because there was none present. Either grub-install is not being run at all or perhaps it is wrongly installing GRUB in the AntiX partition.

Just out of curiosity, what is the disk organization, in terms of partitioning type and file system? Is this using the old BIOS-based, four primary partition layout or the more modern GPT? Similarly, is it using GRUB Legacy (1, 0.97) or is it a variation of GRUB 2 or GRUB 2 EFI?

I hadn't encountered any issues with antiX and GRUB until I started using UEFI/GPT. AntiX will work with it, but I've not been able to have AntiX MANAGE that environment and I have not figured out how to do it. I've installed GRUB 2 to the antiX root partition, and the symptoms I saw were similar to yours.

I am able to boot antiX either by chainloading to the GRUB 2 root installed by antiX from another system, and/or by having PCLinuxOS, Fedora (23, 24, 25), or openSUSE 42.* manage GRUB.

All my computers are second-hand and have BIOSes. I don't have anything new enough to sport a UEFI. The drives are likewise old and therefore use MSDOS partitioning. All partitions are primary with ext4 filesystems. The GRUB is GRUB2.
Other people haven't had problems but I think they were using the graphical installer. My guess is that it is only the cli-installer which has this bug. btw, if you are using UEFI, you need to have the second stage of GRUB installed in your EFI system partition where UEFI can find it. There are some options you can use with the grub-install command to tell it to treat the system as EFI/GPT.

A bit different, but I too seem to have ended up with NO [bootloader], as evidenced by:
nothing on screen, no disk activity (except maybe 1 read I missed); cpu [VBox] 100%.
boot-repair-disk fixed it.
I'm using antiX-16_386-core-libre.iso, so no GUI; cli-install only. (also no IP)
(I'm clueless about the F5 persist/frugal; probably just ignore it here)

NOW, I'm trying with a new VBox. cfdisk Seg-faulted, so I chose Custom instead of Beginning, changed default 0.03M to 1 (msdos label). Only 1 partition, so I answered sda1 for root.
Blew up with "no space" errors for everything, because /media/sda1 is a 10M! tmpfs (not /dev/sda1).
(Make FS seems to be Unimplemented in cfdisk! Did I need to change type to ext3?)
I answered ext3 (#2 of 7 fs choices) later. Confusing, but now it mounted properly.
(cp -a /live/aufs/* . took about 10minutes) I took default to install GRUB in MBR

Well, it 'won't boot', just like the first! isoCD: od /dev/sda|more & saw the 125125 [55aa]
so I don't know what's wrong. I can try anyone's suggestions...

Try booting from the installation disc, then mount your partition and create dynamic directories on it as in this extract from the LFS book. Then chroot to it and run grub-install by hand. That's what I did when setting up my friend's computer.

Someone ought to report this but I'm damned if I'm going to join the AntiX forum just to report a bug! Anticapitalista is the guy who should be dealing with this. Does anyone know if he visits this forum?

Let's fix cli-install script for them! (tomorrow/...) (OR is grub-install broken?)
It seems to do (after mount -o tmpfs --bind /{dev,proc,sys}/) chroot ... hd0
Is hd0 ok or is /dev/sda needed??

I wanna do what grub-install does, *manually*

You need to use the normal device name as an argument: grub-install /dev/sda. hd0 is what the bootloader calls the disk, but this is a Linux command, so you must call it what the kernel calls it.

What it does is to put stage 1 of Grub (boot.img) into the mbr and stage 2 (core.img) after the mbr.

Should be fixed with an upgrade via the repos. Obviously, do this before running the installer.

The script really needs re-writing. When I have time ...

Is that the version I just read on github? I checked out the grub installation sequence and it looked fine to me. Thanks for fixing it so quickly. Just out of curiosity, what was the script doing before?

It is deb version 3.9 in the repos and here at Github: https://github.com/antiX-Linux/cli-installer-antix

The script was trying to install directly to hda (as an example) rather than /dev/sda.

As I posted earlier, the script really needs a complete re-write to cater for UEFI for example. Still it does get the job done now thanks to your bug report.

1 members found this post helpful.