| grizzlybear | im having tons of trouble installing devuan 5.0 to a raid1 mdadm partition as a boot device spanned across two 1tb wd blue ssd drives, sata. | 04:11 |
|---|---|---|
| grizzlybear | grub didn't recognize the btrfs partition and failed to boot on one attempt. | 04:11 |
| grizzlybear | second attempt with the / partition formatted as ext4 gives me a grub error of "error: file '/boot/grub/i386-pc/normal.mod' not found." | 04:12 |
| grizzlybear | any hints? thinking of going back to hdd raid1 and/or using another os | 04:12 |
| gnarface | grizzlybear: do you have the package "grub-pc-bin" ...? | 04:18 |
| gnarface | or i guess just check for grub-pc | 04:18 |
| gnarface | not sure this is an issue specific to raid | 04:19 |
| rwp | I always use a separate /boot formatted as ext2, ext3, or ext4 which grub knows how to boot. AFAIK grub does not know how to boot from btrfs file system. I might be wrong there though. | 04:22 |
| rwp | Using btrfs as a single device works best. Which is why I presume you are using mdadm to create a RAID1 to host a single btrfs partition? I guess so. | 04:23 |
| rwp | I guess I do use grub to boot /boot on native XFS. (me looks sheepish) So that does work too. | 04:24 |
| gnarface | you might just have to add some line to the grub config to load a btrfs module, i don't know | 04:25 |
| gnarface | i do know it was generally recommended not to make /boot part of the raid, but i thought there was a way to do it both ways | 04:25 |
| rwp | I have always put /boot on an /dev/md0 mdadm RAID1. Works great. FTW! | 04:26 |
| rwp | I agree though that you might just be needing to explicitly load the btrfs module. | 04:27 |
| gnarface | grizzlybear: if you actually check, do you see that file present on the disk? | 04:28 |
| gnarface | grizzlybear: also, i assumed you have btrfs-progs installed but make sure | 04:29 |
| rwp | Also not having /boot/grub/i386-pc/normal.mod installed makes me think something failed during the setup. That should be there. | 04:30 |
| DelTomix | ^ could also be the location is changed when things are assembled - or - the partition/drive that grub was installed to doesn't see it at boot time | 04:31 |
| rrq | I guess you're talking about a device module for the boot loader? that boot loader must understand the filesystem it loads the kernel from | 04:32 |
| gnarface | yea i thought grub had... modules, or something | 04:32 |
| gnarface | but also, grub packages going missing during install for certain systems has not been unprecedented | 04:33 |
| rwp | In my poking around here I think grub-pc-bin.postinst should have installed the /boot/grub/i386-pc/* files and if they did not get installed then something there failed. Pretty sure. | 04:33 |
| gnarface | there's a copy in grub-pc-bin at /usr/lib/grub/i386-pc/normal.mod, so i'm assuming what failed was grub-pc | 04:34 |
| rwp | That should all get installed automatically when grub-pc-bin is installed though. | 04:36 |
| rwp | Since setting up mdadm raid1 in the installer is rather tedious I might assume that a mistake was made in the process somewhere. | 04:37 |
| rwp | I often set up partitions and mdadm raid outside the installer and then boot the installer and attach to the already set up raid partitions. It's easier. | 04:37 |
| rrq | there need to be a grub partition accessible by the boot loader prior to sofware raiding | 04:38 |
| rwp | mdadm raid is brought online in the initramfs | 04:39 |
| rrq | right; that's after loading the kernel and initramfs | 04:39 |
| rrq | so there needs to be a prior "fielsystem" for the boot loader | 04:40 |
| rrq | I'm afraid grub has changed a bit recently and all my knowledge is old, possibly outdated | 04:41 |
| rwp | The grub package abuses the debconf database as a registry too. It records the device used to store the boot loader. | 04:41 |
| rwp | Which it uses at upgrade time to refresh the boot loader. Which might have changed over time in a system if disks were replaced or whatever. | 04:42 |
| rwp | I like how the OP threw that question in here and has been busy offline apparently and we are just happily going down various rabbit holes of thought. | 04:42 |
| grizzlybear | awesome my technique of going to go take a shower while somebody replies to my comment works like a charm | 04:55 |
| grizzlybear | thanks! | 04:55 |
| grizzlybear | interesting so i might have some failures in setup | 04:58 |
| grizzlybear | failures to copy files | 04:58 |
| grizzlybear | i guess burning the iso to usb isn't such a great idea after all | 04:58 |
| grizzlybear | will try an optical storage medium.... | 04:59 |
| rwp | The installer can be copied to a USB storage okay. That's what everyone does these days. No need to make it an actual dvd/cdrom. | 05:06 |
| rwp | The installer booted and the installer ran so you know that part was working. | 05:09 |
| grizzlybear | lets rule it out anyway | 06:19 |
| grizzlybear | huh growisofs is missing from the latest ubuntu desktop AND k3b cannot detect my optical drive... grr | 06:44 |
| grizzlybear | will try it your way rwp, using a usb flash drive again but a different one. | 06:55 |
| grizzlybear | i give up | 07:14 |
| grizzlybear | "unable to install grub in dummy" | 07:14 |
| gnarface | why dummy? | 07:14 |
| grizzlybear | dunno why it chose dummy | 07:14 |
| gnarface | choose something else | 07:14 |
| grizzlybear | i'm a dummy though | 07:14 |
| grizzlybear | maybe its calling me a dummy | 07:14 |
| grizzlybear | i cannot choose something else? | 07:15 |
| grizzlybear | its guided partitioning | 07:15 |
| gnarface | i never touched that, i always use manual | 07:15 |
| grizzlybear | yeah even manual fails | 07:15 |
| grizzlybear | i've tried that | 07:15 |
| grizzlybear | will try it again though | 07:16 |
| gnarface | i think you need to do it a specific way | 07:16 |
| grizzlybear | well yeah thats why im here to learn the specific way | 07:16 |
| gnarface | i just remember something about needing LVM or needing to do it by hand | 07:17 |
| grizzlybear | ok | 07:17 |
| gnarface | someone around here does know | 07:17 |
| gnarface | if you try it again make sure you're in expert mode, just because i'm not sure if that matters too or not | 07:18 |
| gnarface | which image did you use? | 07:18 |
| grizzlybear | server amd 64 | 07:18 |
| gnarface | specifically, which link did you download it from i mean | 07:18 |
| grizzlybear | geez it was a while ago now | 07:19 |
| grizzlybear | devuan_daedalus_5.0.0_amd64_server.iso | 07:19 |
| gnarface | alright | 07:19 |
| gnarface | well 5.0.1 is up there now | 07:20 |
| gnarface | not sure if that will help though | 07:20 |
| gnarface | i do know people trip up with the RAID setup a lot, it's not obvious how to make it work right | 07:20 |
| grizzlybear | roger thanks | 07:20 |
| gnarface | it is possible it might be easier to just do the install to one disk normally then setup mdadm afterwards, but i haven't messed with software raid in years, i couldn't tell you specifics | 07:21 |
| grizzlybear | i'm sure it would be easier actually | 07:51 |
| grizzlybear | faster too. installing the base system now and its going really slow | 07:51 |
| grizzlybear | for some reason | 07:51 |
| grizzlybear | possibly because the raid1 is syncing dunno | 07:52 |
| grizzlybear | or its downloading packages | 07:52 |
| grizzlybear | nope that didnt work either. ended up going with a single drive typical installation and i'll do a btrfs conversion and setup mdadm | 08:46 |
| grizzlybear | post install | 08:46 |
| grizzlybear | thanks for your help anyway guys | 08:46 |
| gnarface | hmm, that raid thing keeps coming up, that's not a new issue with daedalus, but there's a way to do it... | 10:40 |
| gnarface | i think it just needs LVM maybe | 10:41 |
| rrq | grizzlybear: do you have UEFI or legacy bios? | 10:51 |
| ragnarov_unident | Hi | 15:58 |
| debdog | ahoy! | 16:00 |
| ragnarov_unident | sorry, I am a bit of a compromised position rn | 16:02 |
| ragnarov_unident | new setup, haven't got everything running, shouldn't have pinged so fast | 16:02 |
| ragnarov_unident | When I switch tty the tty containing Xorg freezes | 16:03 |
| ragnarov_unident | I can't get back to it | 16:03 |
| ragnarov_unident | I am starting Xorg from startx | 16:03 |
| ragnarov_unident | the all other ttys work | 16:04 |
| debdog | ragnarov_unident: yes, known issue, solutin is on the way | 16:04 |
| debdog | you can try with some packages from excalubur, gimme a minute | 16:04 |
| debdog | <rrq> CALL FOR TESTERS: if you run startx as non-root, using daedalus and ceres, you are invited to test the devuan/experimental build and send feedback to LeePen or rrq .. it's a version that is happy about VT switching (supported by either seatd or elogind), but we want more confirming tests. | 16:05 |
| debdog | xserver-xorg-core and (dependecncy) xserver-xorg-common | 16:05 |
| debdog | with seource.list line added, it'd be: apt-get install -t experimental xserver-xorg-core | 16:05 |
| debdog | the sources.list line is: deb http://deb.devuan.org/devuan experimental main | 16:05 |
| ragnarov_unident | aight | 16:07 |
| ragnarov_unident | debdog apt says xserver-xorg-common is already at the newest version | 16:20 |
| ragnarov_unident | brb | 16:36 |
| debdog | ragnarov_unident: sorry, I was just relying that message and that is all I know | 16:43 |
| ragnarov_unident | wai | 16:43 |
| ragnarov_unident | realy it agian | 16:43 |
| ragnarov_unident | I lost it, I am not registered, so I lot it | 16:44 |
| ragnarov_unident | debdog ^ | 16:44 |
| debdog | ragnarov_unident: the channel is looged, see topic | 16:46 |
| debdog | *logged | 16:46 |
| ragnarov_unident | thanks | 16:48 |
| ragnarov | guys the fix did not work for me | 17:32 |
| ragnarov | It freezes mouse & keyboard | 17:32 |
| debdog | rrq ^ | 17:36 |
| debdog | ragnarov: out of curiousity, what GPU and driver? | 17:37 |
| ragnarov | intel | 17:37 |
| debdog | ok | 17:37 |
| ragnarov | I have not configured GPU, it has and and radeon one | 17:37 |
| debdog | ok, just plain intel | 17:38 |
| ragnarov | Aye | 17:38 |
| debdog | oh, another question arises, ragnarov, is it fresh daedalus install or upgraded from previous version/s? | 17:39 |
| ragnarov | upgraded | 17:40 |
| debdog | interesting | 17:40 |
| ragnarov | Prev one was 3 months old, I am new to Debian/devuan | 17:40 |
| debdog | need to test that here, thanks | 17:40 |
| debdog | I have some really odd issues with upgraded versions here as well. everything seems fine so far with a fresh install | 17:42 |
| systemdlete | So I wrote a drop-in replacement script for os-prober. I find that if I place os-prober in /usr/local/sbin/, update-grub will find it and use it instead of its own. My script COMPLETELY IGNORES REMOVABLE DRIVES! So no thumbdrive trickery, which I think was one of the issues with os-prober security. (of course there could be other trickery, like removable SATA drives, but maybe that's not so | 20:34 |
| systemdlete | terrible, idk). | 20:34 |
| systemdlete | Anyway, my script is working, and I've testing it on all 3 of my hosts, which contain a variety of multiboot configurations. | 20:34 |
| systemdlete | I will post it here and if you see problems with it, let me know and I will see if I can easily modify it. | 20:35 |
| systemdlete | I've tried to be generous with comments in the script, and it also has a trace option. (no man page yet, sorry...) | 20:35 |
| * phogg awaits eagerly | 20:37 | |
| systemdlete | https://paste.debian.net/1292056/ | 20:38 |
| * systemdlete likes phogg's comment... | 20:39 | |
| systemdlete | There may be simpler ways to do this, but this one is working, so I went with it. | 20:40 |
| systemdlete | thanks to all for /constructive/ feedback | 20:41 |
| phogg | systemdlete: I can provide some feedback on sh things if you like | 20:46 |
| systemdlete | as you wish | 20:46 |
| systemdlete | (hit me!) | 20:47 |
| phogg | I'll do it in the form of a paste in a minute | 20:47 |
| * debdog hits systemdlete with a wet trout wrapped in a frozen towel SCNR | 20:48 | |
| * systemdlete thanks debdog for that; they needed it | 20:48 | |
| systemdlete | one thing I will say upfront: I don't care for K&R style coding. So no run-on "if [ x = y ] ; then" in my scripts. I find that harder to read later on. | 20:51 |
| phogg | I won't comment on style | 20:51 |
| systemdlete | just a preference I developed early on | 20:51 |
| phogg | we all have our preferences, mine are just as bad as anyone else's | 20:51 |
| systemdlete | phogg, are you re-writing my whole script? LOL | 20:51 |
| phogg | no | 20:52 |
| systemdlete | (some people will do that sort of thing) | 20:52 |
| phogg | I do that sort of thing, but I am just doing a couple things in this case. | 20:52 |
| systemdlete | I have added some comments since I posted that | 20:52 |
| systemdlete | are your changes portability/compatibility issues, or did I miss some cases? | 20:55 |
| * systemdlete is curious | 20:55 | |
| phogg | just some little efficiency things and a bit of correctness | 20:56 |
| phogg | most of the time is making sure I understand what you're doing | 20:56 |
| systemdlete | Yeah, I didn't go into much philosophy or strategy discussion there. | 20:56 |
| systemdlete | I just want to re-iterate: I did not change anything else besides the ONE os-prober script. So I didn't modify any of the scripts that the testing portion calls, e.g. | 20:58 |
| systemdlete | There may be issues with those, such as for Hyperbola (which takes a long time sometimes) test | 20:58 |
| phogg | systemdlete: https://termbin.com/7csp lots more I would do differently, but here are my suggestions. | 20:58 |
| systemdlete | OK, thanks. Generally, I find the IFS= business makes the code more difficult, if maybe more efficient, as you point out. | 21:00 |
| phogg | I figured two subprocesses was a bit too much | 21:01 |
| systemdlete | I'm ok with a little inefficiency if it makes code easier to maintain | 21:01 |
| phogg | you could also do something similar to avoid using ls in two more places, but one of them is "is dir empty?" which is very hard | 21:01 |
| systemdlete | and I know the error handling on umount is lacking... I was getting around to that. | 21:01 |
| systemdlete | phogg, have you tried this script? | 21:02 |
| phogg | did not test at all | 21:02 |
| phogg | didn't even shellcheck it... probably should do that anyway | 21:02 |
| systemdlete | I'm hoping someone will try it on a multi-boot system | 21:03 |
| systemdlete | Get some more data points for the outcome of the operation | 21:03 |
| phogg | shellcheck found a couple things and I noticed a couple more: https://termbin.com/xh87 | 21:03 |
| systemdlete | it can be run standalone, and all it will do is generate the entries for update-grub | 21:04 |
| phogg | I haven't got a multi-boot system right now, but I was thinking about grabbing an old box and throwing gentoo on it in the next week. I could maybe try it there after that. | 21:04 |
| systemdlete | thanks, then you will let me know | 21:05 |
| systemdlete | mostly, I am concerned with the security issues that have put os-prober in the klink | 21:05 |
| phogg | when I've tried it I'll report back one way or the other | 21:05 |
| systemdlete | that's why I avoided removable drives. | 21:06 |
| systemdlete | but I am not sure about all of the issues that made it a baddy | 21:06 |
| systemdlete | And I'm not claiming I will solve all of those | 21:06 |
| phogg | Are there open bugs for them? | 21:07 |
| systemdlete | you know, I actually visited the gnu site where they have the bug list, but from what I can tell, it seems that little to nothing has moved much for grub, generally, over the past few years. | 21:08 |
| phogg | two reasons for that: it's "done", and we are in the EFI era where people question the need for a bootloader at all | 21:08 |
| systemdlete | EFI is sort of a bootloader itself, I thought | 21:09 |
| phogg | not entirely, but it solves some of the same problems grub solves | 21:09 |
| phogg | I say throw out EFI and convince the industry to standardize on coreboot and grub instead. The world would be a better place. | 21:09 |
| systemdlete | aside from hobbyists, I wonder how much demand there is for multi-boot systems | 21:10 |
| systemdlete | I like the concept of EFI, but it seems to have gotten snagged in implementation more than design. Istr a lot of bugs and security issues early on. | 21:10 |
| phogg | probably zero, but there is a demand for purpose-built systems where you have control all the way down to the metal | 21:11 |
| phogg | EFi smacks of committee design, but anything further on this should probably move to #devuan-offtopic | 21:11 |
| systemdlete | agreed | 21:12 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!