Freshman | yep everything grate after upgrade | 00:00 |
---|---|---|
Freshman | with only sed holded back | 00:01 |
Freshman | how report it? | 00:05 |
Freshman | which portal? | 00:05 |
fsmithred | best would be to test it in debian and then report to debian | 00:05 |
fsmithred | bugs.debian.org | 00:05 |
Freshman | thx. so devuan bugs flows to debian? are there own bugtracker? | 00:07 |
Freshman | or it is about devuan independent bug? | 00:08 |
fsmithred | bugs.devuan.org is for packages that we forked | 00:08 |
fsmithred | you can tell those because "devuan" is in the version | 00:08 |
Freshman | i see. thank you | 00:09 |
fsmithred | but it would not be good to report a bug to debian and then find out it's only a bug on devuan. | 00:09 |
rrq | somewhat likely "sed" is now installed at /usr/bin/sed and that that causes brekage | 00:09 |
fsmithred | They don't care and they'll just get angry. | 00:09 |
fsmithred | oh, that would make a difference. Should there be some script that fails and leaves a note in syslog? | 00:10 |
fsmithred | should/is | 00:10 |
fsmithred | dinner. bbl. | 00:10 |
fsmithred | I'm upgrading my ceres VM now and will get the new version of sed. I'll see where it goes, what happens when I boot, and what happens if I make the proper symlink. | 01:00 |
fsmithred | oh this will take more than an hour | 01:03 |
Freshman | interesting | 01:22 |
Freshman | which kind of VM do you use? | 01:23 |
rustyaxe | Spent an hour trying to figure out why my ttyS0 was getting hijacked.. Thanks brltty... | 01:44 |
brocashelm | i think it was documented somewhere, but how to prevent getty-run from littering tty1 with service scripts? i just want a non-DM interface to log in | 01:46 |
brocashelm | i use runit btw | 01:46 |
brocashelm | should i look somewhere in inittab or rc.d configs? | 01:47 |
onefang | Why not use tty2? | 02:04 |
al1r4d | Morning ☕ | 02:04 |
brocashelm | because tty1 is #1 ofc | 02:06 |
brocashelm | too lazy to ctrl+alt+f2 | 02:07 |
brocashelm | maybe i'll go back to a dm, but use something simpler like xdm or wdm so that it doesn't really get in the way of my settings (lxdm is a no) | 02:08 |
fsmithred | brocashelm, /etc/sysctl.conf to control verbosity of messages | 02:49 |
fsmithred | also, I think booting with 'quiet' turns off console messages | 02:50 |
fsmithred | Freshman, I'm using qemu at the moment. Sometimes vbox. Upgrade is not done. | 02:50 |
fsmithred | almost... | 02:52 |
fsmithred | Oh, cool! I have modules for three kernels in /lib/modules and one for the latest kernel in /usr/lib/modules. Is it safe to reboot into the new kernel? (Tell me or wait and I'll tell you.) | 03:00 |
brocashelm | fsmithred: just uncomment this line? kernel.printk = 3 4 1 3 | 03:00 |
fsmithred | wow, I don't recall what the numbers mean | 03:00 |
brocashelm | that's what the refracta ISOs came with | 03:01 |
brocashelm | i'll check | 03:01 |
fsmithred | yeah, that's what the comment says | 03:01 |
brocashelm | https://www.kernel.org/doc/html//v5.12/core-api/printk-basics.html | 03:02 |
brocashelm | i'm checking here | 03:02 |
brocashelm | so it would be KERN_ERR (3), KERN_WARNING (4), KERN_ALERT (1), and back to KERN_ERR (3)? | 03:03 |
brocashelm | just uncommented and rebooting now | 03:04 |
fsmithred | Nope. Ceres does not boot after upgrade. Cryptsetup cannot fine any devices. | 03:04 |
brocashelm | lame | 03:05 |
brocashelm | ok, the messages are gone. thanks | 03:06 |
fsmithred | I'm installing usrmerge now | 03:12 |
fsmithred | my current autremove list it pretty big. | 03:12 |
brocashelm | good luck | 03:13 |
fsmithred | it's just a test VM | 03:14 |
brocashelm | true, but it's a pain to work with, last i checked | 03:15 |
fsmithred | I'm expecting to learn new ways that the system is broken. | 03:15 |
brocashelm | thinking of experimenting with that on a ceres vm soon | 03:15 |
brocashelm | newer kernels are definitely altering stuff | 03:16 |
brocashelm | 6.4 and 6.5 aren't finding my iptables/firewall for ufw when i run a lynis audit (even though firewall is active), and swap is never being used (even after setting vm.swappiness = 95 and using the machine for days/weeks with browsing and videos in between) | 03:17 |
fsmithred | udev rules file for gphoto exists in two places and installing usrmerge fails | 03:19 |
fsmithred | it says I can run /usr/lib/usrmerge/convert-usrmerge until it completes without error. | 03:21 |
fsmithred | ...after fixing the problem | 03:23 |
fsmithred | and then it complains the same about libsane | 03:23 |
fsmithred | so I probably need to move /lib/udev/rules.d into oblivion. | 03:23 |
fsmithred | ok, I had to move all of /lib/udev and run /usr/lib/usrmerge/convert-usermerge again and that fixed it. | 03:27 |
fsmithred | The big autoremove list was misleading. Same packages were also being installed. | 03:27 |
rjp19ateseven | happy happy joy joy? | 03:29 |
fsmithred | reboot in new kernel fails again because cryptsetup can't find devices | 03:33 |
fsmithred | reboot to old kernel boots, but now I can't type to log in to slim. | 03:35 |
fsmithred | I booted with live-iso and can chroot the system but I can't run update-initramfs. | 03:49 |
fsmithred | in chroot I installed sysv-rc-conf and turned off slim in runlevel 3, then booted the installed system to runlevel 3. I can type commands. | 03:55 |
fsmithred | ran update-initramfs -u -k all | 03:56 |
fsmithred | rebooted and new kernel (6.6.0-9) still fails on cryptsetup trying to find devices. Boot old kernel (6.5.0-4) and I can't type to log into slim. | 03:59 |
fsmithred | boot old kernel to runlevel 3 (no slim) and startx fail with a bunch of xorg errors | 04:02 |
fsmithred | I'm done for the evening. | 04:03 |
gnarface | fsmithred: i probably can't help with whatever you were trying, but just a quick sanity check, did you bind mount /proc, /sys, /dev, and /dev/pts in the chroot first? | 04:34 |
rwp | fsmithred, Could you make effective testing use of a chroot? It's a reduced set but things go much faster that way. (Remember to install /usr/sbin/policy-rc.d with an exit 101 to keep it from starting daemons on the upgrade.) | 05:31 |
Freshman | fsmithred cool | 08:30 |
Freshman | btw 'sed' is the first time i holded smth for stability | 08:31 |
Freshman | and there's no bug reports on debian | 08:33 |
fsmithred | startx fails with segfault. Here's Xorg.0.log: https://termbin.com/k79k | 12:01 |
fsmithred | gnarface, yes I did the bind mounts. | 12:02 |
fsmithred | rwp, I'm not sure I understand what you want me to do. | 12:03 |
fsmithred | Freshman, sed is not the problem with your system. usrmerge is the problem. | 12:03 |
ykk0t | My devuan 5.0 netboot.iso boots into a blank screen. I downloaded from two different mirrors and checked the SHA256 on the last one and everything seems to be in order. I'm attempting to boot from an USB stick and I burned the ISO with Rufus. I used the very same program for Debian, Mint and Gentoo and they worked so I don't think it's the culprit. | 12:06 |
ykk0t | What gives? | 12:06 |
Freshman | fsmithred, wow. the job you made | 12:11 |
Freshman | what if i check this package on my sys? | 12:12 |
fsmithred | I installed lightdm, and ps shows that it's running after booting up, but I'm stuck on console. | 12:12 |
fsmithred | what package? | 12:12 |
Freshman | 'usrmerge' | 12:12 |
fsmithred | as far as I can tell, that's a good way to fuck up your system | 12:13 |
fsmithred | you can try it - you have backups | 12:13 |
freshman | oh yeah. didn't tell but i already fucked up my system yesterday. deleted some python lib which draged timeshift as well^^ | 12:14 |
djph | "oops" | 12:14 |
freshman | the python lib whuch was in problem candidate listl) | 12:15 |
fsmithred | [ 1296.442] (EE) xf86OpenSerial: Cannot open device /dev/input/event1 | 12:15 |
fsmithred | Permission denied. | 12:15 |
freshman | so no internet, no timeshift, no gui) | 12:16 |
freshman | thanks to rescue mode) | 12:16 |
gnu_srs1 | Seems like debian is fucking up more stuff due to usrmerge. E.g. libtool has hardcoded paths to /bin/grep and /bin/sed, and sed is now at /usr/bin. | 12:17 |
rrq | fsmithred: which xserver-xorg-core version ? | 12:17 |
freshman | fsmithred, usrmerge is not installed in the sys and wasn't | 12:17 |
fsmithred | /dev/input/ only has mice, mouse0 and mouse1 no event* | 12:17 |
rrq | (you may need seatd installed) | 12:18 |
fsmithred | I think I have seatd | 12:18 |
fsmithred | oh, installing seatd now | 12:18 |
gnu_srs1 | Works only on a merged /usr system. What to do project-wise (if one don't want to install usrmerge)? | 12:19 |
fsmithred | xserver-xorg-core 2.21.1.10 | 12:19 |
rrq | that's debian's version | 12:20 |
rrq | will require systemd | 12:20 |
fsmithred | oh, our version isn't ready yet? | 12:20 |
fsmithred | no, it's ours. 1devuan1 | 12:20 |
u-amarsh04 | gnu_srs1 I have been trying to catch when packages move the installed binary from /bin to /usr/bin (like sed just recently) then immediately after aptitude runs, add a symbolic link e.g. /bin/sed to /usr/bin/sed | 12:21 |
fsmithred | sorry, still running on first cup of coffee | 12:21 |
fsmithred | is there some reason the devs don't use 'command -v' instead of hard-coding stuff? | 12:21 |
rrq | ok. so then input access is facilitated either bmo elogind (and stuff) or by seatd | 12:21 |
u-amarsh04 | I posted in dev1galaxy.org about sed in the hope of reducing grief for others | 12:22 |
fsmithred | I do have elogind | 12:22 |
rrq | and dbus-x11 | 12:22 |
u-amarsh04 | if/when coreutils does the same thing I'll have to do a lot of work manually after booting from a rescuedvd | 12:22 |
fsmithred | I have dbus-x11 | 12:25 |
rrq | mmm the elogind facilitation is of course rather complex, but essentially xorg uses dbus to talk to the elogind daemon for opening the input devices, and pass on opened file descriptors... there would be some authorization thingy involved; policykit or consolekit I think | 12:29 |
rrq | also, the user need to own the /dev/ttyN involved | 12:29 |
rrq | with seatd started, xorg will rather try its facilitation; but if not started, it'll try logind facilitation... and if neither is available it will want to run a "vt-bound" seatd for itself | 12:33 |
rrq | in the latter case, the same user will run seatd and will then need rw access to the device nodes | 12:34 |
fsmithred | oh yeah, I should try statx after installing seatd | 12:34 |
fsmithred | nope. segfault again. | 12:35 |
fsmithred | any other logs I should save? | 12:35 |
rrq | maybe an strace of xorg ... edit /usr/bin/Xorg and prefix the start with "strace -f -o /tmp/my.x.log" | 12:37 |
rrq | (after "exec") | 12:38 |
fsmithred | ok | 12:38 |
rrq | .. then look for "/bin" paths without "/usr" prefix :) | 12:40 |
fsmithred | there are two 'exec' in an 'if/else' - should I do both? | 12:42 |
rrq | probably best; I guess the last one is the ine used (you don;t have a /usr/lib/xorg/Xorg.wrap ?) | 12:43 |
fsmithred | I have Xorg and Xorg.wrap | 12:44 |
fsmithred | and now I have strace, so it'll work better | 12:44 |
fsmithred | :) | 12:44 |
rrq | :) | 12:45 |
fsmithred | it didn't create the log | 12:45 |
rrq | hmm it bypasses .. the actual binary is /usr/lib/xorg/Xorg ... needs more involved wrapping | 12:46 |
rrq | like: mv /usr/lib/xorg/Xorg{,.real} | 12:47 |
rrq | echo "exec strace -f -o /tmp/myx.log /usr/lib/xorg/Xorg.real" > /usr/lib/xorg/Xorg | 12:48 |
rrq | chmod a+x /usr/lib/xorg/Xorg | 12:48 |
rrq | # end | 12:48 |
rrq | possilby that script should have an initial #! line I suppose | 12:48 |
fsmithred | it has shebang | 12:49 |
rrq | ? in /usr/lib/xorg/Xorg (not /usr/bin/Xorg) | 12:49 |
rrq | I expected /usr/lib/xorg/Xorg to be an ELF binary | 12:50 |
fsmithred | oops. I did /usr/bin/Xorg | 12:51 |
fsmithred | fixing it now | 12:51 |
fsmithred | ok /usr/lib/xorg/Xorg is binary file | 12:53 |
fsmithred | scrolling up to re-read... | 12:53 |
fsmithred | I did edit the right file. | 12:53 |
rrq | yes; messing with /usr/lib/xorg/Xorg is plan B, since plan A didn;t work | 12:54 |
rrq | thus move it to Xorg.real, then make that (previous) Xorg be a script | 12:55 |
rrq | .. and add "$@" (with the quotes) to the end of that echo-ed line btw | 12:55 |
fsmithred | I can't type curly braces - using vnc | 12:57 |
fsmithred | mv Xorg Xorg.real? | 12:58 |
rrq | yes (in /usr/lib/xorg) | 12:58 |
fsmithred | yeah | 12:59 |
rrq | maybe we should continue on pm .. leaving space for others... | 12:59 |
fsmithred | ok | 13:03 |
[-_-] | anyone using wayland? | 13:44 |
freshman | there are wayland on my system, but they somehow together with X | 14:48 |
n4dir | when i installed KDE i had that too | 14:49 |
n4dir | if i dpkg -l | grep -i wayland, i sure get a massive list. This installation has never seen KDE, not sure why all that is there. Nor what exactly it is, probably libs | 14:50 |
mason | freshman: Taking one random example from my local system, where I don't use Wayland: 'aptitude why libwayland-dev' says it's because of libgtk-3-dev | 16:39 |
joerg | *-dev is harmless | 16:50 |
joerg | it's just a "you COULD..." | 16:51 |
joerg | AIUI | 16:52 |
freshman | mason, i see | 16:54 |
freshman | i decided to migrate of X to Wayland. installing it and some manipulations ended by nothing. | 16:56 |
freshman | as i used to think it do function but still X is required | 16:57 |
freshman | there are talks about Wayland as alternative, that's why i decided to try | 16:57 |
mason | freshman: I have no interest in Wayland. If it's ever feature-complete I'll consider it. | 17:05 |
mason | ...if and only if X completely rots away first. | 17:05 |
jonadab | /w 35 | 18:41 |
fsmithred | freshman, fwiw I installed usrmerge on daedalus VM and then upgraded to ceres, and it's working ok with the new kernel. | 19:01 |
fsmithred | I don't have a full desktop system installed - just window manager and a few other things. Also, this happens to be a no-dbus system. That might make a difference. | 19:01 |
fsmithred | 6.6.9 kernel | 19:01 |
Besnik_b | gnarface, during my attempts to add swap to my Devuan, the machine died. I tested the PS and was dead. Probably that made the cooler stop and the CPU temperatures raise. So I had to built a new machine and here I am under UEFI now… :) | 20:10 |
mathew | So... If i install the nvidia-driver with "apt install nvidia-driver --no-install-recommends"! The installation runs through. However, the driver is not loaded during boot and loads the xorg nouveau driver... It seems, that critical packages are not installed correctly using the said method. | 20:40 |
n4dir | i belong to the ones who avoid installing without Recommends. Might be worth a try in that case, but i doubt it is the source of the problem | 20:41 |
mathew | Is it a good idea to install every recommended package but "nvidia-persistenced"? That would be my next try i guess. | 20:47 |
mathew | is there some "-exclude" like option for apt? | 20:48 |
n4dir | if it was me i'd install it and then remove it, assuming i understand your case correct | 20:49 |
n4dir | remove it right away. | 20:49 |
mathew | The problem is that apt does not run through and exits with error code 1 forcing me to run "apt install --fix-broken" which again ends in error code 1 | 20:50 |
mathew | i can only remove nvidia by running "dpkg -r nvidia*" then | 20:51 |
n4dir | you would need to pastebin the command and the full output | 20:51 |
n4dir | apt- shouldn't exit with error code. | 20:51 |
mathew | ok i like to show you! :-) | 20:51 |
n4dir | once at it show you sources list(s) | 20:52 |
fsmithred | n4dir, do you have an xorg.conf for nvidia? | 20:52 |
fsmithred | you don't need nvidia-persistenced and it generally block installation | 20:52 |
n4dir | no, mainly because i don't have nvidia. You probably mean him | 20:52 |
fsmithred | oh, sorry | 20:53 |
fsmithred | mathew, ^^^ | 20:53 |
n4dir | i am busy for a few minutes, mathew , bu the info will be good to troubleshoot anyway | 20:53 |
mathew | https://paste.debian.net/1302965/ | 20:55 |
n4dir | fsmithred: ^^ looks like all gets installed ok, but then the installation process fails with update-initramfs. | 20:56 |
n4dir | mathew: i think i understand, it gives what you said it would give, but i wouldn't know the problem behind. | 20:57 |
n4dir | your idea, install Recommends but exclude that package, sounds like a good idea though | 20:57 |
fsmithred | you have to exclude Recommends and add the ones you want | 20:59 |
fsmithred | apt --no-install-recommends install nvidia-driver nvidia-whatever... | 20:59 |
fsmithred | but not nvidia-persistenced | 20:59 |
mathew | This is what i did know: https://paste.debian.net/1302966/ | 21:00 |
mathew | but maybe i have to run through every recommended package by hand just to exclude this troublemaker "nvidia-persistenced". but i will reboot and see if it maybe works now | 21:01 |
n4dir | there is also that trick to quickly set the output to English, so it is more easy for others to understand the idea. mathew | 21:02 |
n4dir | i don't know how the trick is, but it is straight forward. | 21:02 |
n4dir | time for the Sauerkraut ... | 21:03 |
mathew | n4dir: lol. sorry for the sauerkraut. i do not know how to change the language but will look it up. sadly the driver did not load. i will know try to install every package by hand while using --no-install-recommends | 21:05 |
n4dir | mathew: well, most of the times folks will understand the output anyway. The basic info is the same. Just makes it a bit more easy | 21:06 |
mathew | yes, absolutly! | 21:06 |
fsmithred | I think you only need to install nvidia-driver and nvidia-settings with --no-install-recommends and you'll get what you need. | 21:12 |
Nietz | fsmithred: question about nvidia driver | 21:14 |
Nietz | comes with nvidia-persistenced, but this can be uninstalled without breaking. | 21:15 |
Nietz | do you know how/if persistence mode matters? ive been trying to get to the bottom of it | 21:15 |
fsmithred | I don't know the details. Description says it saves the state. | 21:16 |
fsmithred | I stopped using nvidia when the last video I looked at showed up like a tile mosaic on my desktop after a reboot. | 21:17 |
Nietz | something about useful for no video setups | 21:17 |
Nietz | so availability is not affected as compute-only device | 21:17 |
Nietz | nvidia should die but I still dont have the money to replace my old GTX 1650 super | 21:18 |
Nietz | mby not should die . . . i have used their drivers a lot apologies | 21:18 |
Nietz | do you use nouveau? | 21:19 |
fsmithred | nvidia should share their sources so someone can fix it | 21:19 |
fsmithred | nouveau on one machine (beowulf) no problems | 21:19 |
Nietz | ^ that's what i meant to say | 21:19 |
Nietz | ok, I used to play more steam games, but my card doesn't have ray tracing so it's silly at this pt | 21:19 |
mathew | This is what i did now: https://paste.debian.net/1302972/ (Please excuse the sauerkraut, setting LANG=en did not change the language of my konsole) But sadly the driver does not load. Running nvidia-smi tells that it failed to communicate with the driver. Also in /var/log/boot there are error regarding nvidia :-( | 21:20 |
mathew | I will try again if i have new infomation. for now i just life with the nouveau driver. devuan already runs great with it. | 21:20 |
n4dir | "please excuse the sauerkraut". ha ha ha | 21:21 |
mathew | fsmithred, Nietz. You both are absolutly right about nvidia. i got my gtx1080 so cheap that i could not say no... shame on me | 21:21 |
fsmithred | LC_ALL=C <command> | 21:22 |
fsmithred | ^^^ mathew | 21:23 |
mathew | fsmithred: LANG=C works! thx | 21:23 |
tavasti | Hello! I have trouble with enabling GPU rendering in my devuan. I can get one error with devuan packaged bleder, but actually I would want to use more recent blender version. Now I am not sure where to report bugs... | 22:13 |
tavasti | I have Vega 64 card, and I get: | 22:13 |
tavasti | Compiling HIP kernel ... | 22:13 |
tavasti | hipcc -Wno-parentheses-equality -Wno-unused-value --hipcc-func-supp -O3 -ffast-math --amdgpu-target=gfx900 -I /usr/share/blender/scripts/addons/cycles/source --genco /usr/share/blender/scripts/addons/cycles/source/kernel/device/hip/kernel.cpp -o "/home/tavasti/.cache/cycles/kernels/cycles_kernel_gfx900_77944C4F96BD4E71230409B7F1F7F920" | 22:14 |
tavasti | error: unhandled SGPR spill to memory | 22:14 |
tavasti | more recent blender versions give Failed to load HIP kernel from '/data/tavasti/NoBackup/blender-4.0.2-linux-x64/4.0/scripts/addons/cycles/lib/kernel_gfx900.fatbin' (Shared object initialization failed) | 22:14 |
freshman | fsmithred, you mean you could fix the bug (with sed and usrmerge)? | 22:20 |
fsmithred | no, freshman, I mean that the whole usrmerge mess is the cause of the problem | 22:32 |
freshman | ok. but you said that system was working well | 22:35 |
fsmithred | yeah | 22:35 |
fsmithred | read again. Different system. | 22:36 |
fsmithred | try 'apt -s remove dbus' to get an idea of what you would lose. | 22:36 |
fsmithred | also, it's possible that something is broken and I haven't found it yet. | 22:37 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!