| nemo | Afdal: huh. you don't? | 01:42 |
|---|---|---|
| nemo | I thought it was default for devuan desktops | 01:42 |
| Afdal | well I don't see it in my service list anyway... | 01:42 |
| nemo | $ ps auwx | grep dnsmasq | 01:43 |
| nemo | dnsmasq 1769 0.0 0.0 12700 580 ? S Feb15 0:00 /usr/sbin/dnsmasq -x /run/dnsmasq/dnsmasq.pid -u dnsmasq -7 /etc/dnsmasq.d,.dpkg-dist,.dpkg-old,.dpkg-new --local-service --t | 01:43 |
| nemo | etc.. | 01:43 |
| nemo | and I definitely have a dnsmasq service | 01:43 |
| nemo | although at least over here I also have a dnscache for the entire house | 01:44 |
| nemo | probably doesn't work as well as it used to, in this age of DoH | 01:44 |
| Afdal | hmm, there it is | 01:47 |
| Afdal | well for some reason it's not on autocomplete | 01:47 |
| Afdal | when I do sudo service blahblah | 01:47 |
| nemo | Afdal: I didn't know autocomplete did that.. distro specific thing? I know a lot of random stuff ends up in autocomplete.. sometimes stuff that's not a good idea | 02:14 |
| gnarface | Afdal: not sure if this is your issue but make sure ENV_SUPATH in /etc/login.defs has /usr/sbin | 07:44 |
| ski | hm, (using Mate) playing video in Firefox keeps stuttering all the time (even after restarting browser), but using e.g. `mplayer' doesn't stutter. restarting X doesn't seem to help, either. any idea what might be wrong ? maybe there's some service i could try restarting ? | 14:06 |
| ski | (this has happened, on occasion, in the past as well. rebooting seems to help, but i'd rather not resort to that, having open context running under GNU Screen) | 14:07 |
| gnarface | ski: go into plugins settings and disable the cisco h264 one | 14:20 |
| gnarface | "OpenH264 Video Codec provided by Cisco Systems, Inc." < disable this, our native support on linux is better | 14:20 |
| gnarface | there's some possibility you're missing a particular package or something like that too; mplayer may be running better but that's not proof it's running optimally | 14:22 |
| gnarface | which actual package or packages that might be, most likely would vary depending on whether you're using nvidia or mesa | 14:22 |
| ski | hm, where do i find these plugins settings ? | 14:29 |
| * ski checked Firefox settings, Control Center, Preferences->Hardware | 14:29 | |
| ski | checking packages, i appear to have a bunch of ones with "mesa" in the name, but not many with "nvidia" | 14:34 |
| fsmithred | ski, Tools, Addons and Themes, Plugins | 15:15 |
| ski | hm, it is already disabled (apparently) | 15:21 |
| ski | (apparently i could find this, by going to <about:addons>) | 15:22 |
| * ski also forgot there was a "Tools" menu, usually have the meny bar turned off | 15:24 | |
| gnarface | ski: about:plugins should list them, not sure if about:addons is the same thing or not. do you have a nvidia card? if so, another possibility is that you're using nouveau instead of the official nvidia (closed-source) drivers | 15:42 |
| gnarface | ... and if you've got one of those weird laptops with two video cards of different brands in it, then there's other possibilities of things to go wrong... | 15:43 |
| ski | <about:plugins> lists installed plugins, but doesn't seem to allow you to actually do something with them | 15:58 |
| ski | not sure whether i have an NVidia card, how would i know/check ? | 15:59 |
| ski | this is a laptop, yes (HP) | 15:59 |
| * ski 's not much of a hardware person | 15:59 | |
| debdog | ski: lspci | grep -i -e vga -e display | 16:01 |
| ski | searching for `nouveau' lists `libdrm-nouveau2',`xserver-xorg-video-nouveau' as installed | 16:03 |
| ski | 00:02.0 VGA compatible controller: Intel Corporation 3rd Gen Core processor Graphics Controller (rev 09) | 16:04 |
| gnarface | only that one shows up? | 16:04 |
| ski | yep | 16:04 |
| gnarface | seems like if it installed nouveau you probably have the dual video card thing going on too | 16:04 |
| debdog | wait, they show up as 3D | 16:05 |
| debdog | ski: lspci | grep -i -e vga -e display -e 3d | 16:05 |
| ski | same output | 16:06 |
| gnarface | do we know for sure it even shows up if it's not the active one? | 16:06 |
| debdog | lspci -k | grep -i -e nouveau | 16:06 |
| ski | no output | 16:07 |
| debdog | ok | 16:07 |
| debdog | compared to my HP laptop, there should™ not be a second one | 16:07 |
| gnarface | hmm | 16:08 |
| gnarface | why would it have installed nouveau by default though? that seems weird | 16:08 |
| gnarface | well, whatever | 16:08 |
| debdog | mayhap a "aptitude why xserver-xorg-video-nouveau" will shed some light on that *shrug* | 16:09 |
| ski | installed Devuan through live-cd ISO image, early last year, or so | 16:09 |
| gnarface | ah, maybe that's why | 16:09 |
| ski | (on a Ventoy USB stick, though i guess that doesn't matter) | 16:09 |
| ski | i task-desktop Depends xserver-xorg-video-all | 16:10 |
| ski | i A xserver-xorg-video-all Depends xserver-xorg-video-nouveau | 16:10 |
| gnarface | yea, nevermind that | 16:10 |
| gnarface | stand by, getting you a list of mesa packages to check | 16:11 |
| gnarface | ski: check to make sure you have all these (or the i386 ones if your hardware is 32-bit only, or both if you are using wine or proton) http://paste.debian.net/1307706/ | 16:12 |
| gnarface | ski: for Intel graphics you might need a couple others... like i965-va-driver intel-media-va-driver libvdpau-va-gl1? | 16:14 |
| gnarface | (maybe others i forgot) | 16:14 |
| gnarface | start with those anyway | 16:14 |
| gnarface | there is also the option of using the actual intel xorg driver (by default it uses the kernel modesetting driver) | 16:15 |
| nemo | for anyone else who encountered the nvidia-kernel-dkms issue, looks like debian upstream fixed it | 16:17 |
| ski | no `libd3dadapter9-mesa',`libgl1-mesa-glx',`libopenvg1-mesa' (unlisted),`libosmesa6-dev',`mesa-common-dev',`mesa-utils',`mesa-utils-bin' (unlisted). have `libegl-mesa0',`libgl1-mesa-dri' (and `:i386'),`libglapi-mesa',`libglu1-mesa',`libglu1-mesa-dev',`libglx-mesa0',`libosmesa6',`mesa-va-drivers' (and `:i386'),`mesa-vdpau-drivers' (and `:i386'),`mesa-vulkan-drivers' (and `:i386') | 16:22 |
| ski | gnarface ^ | 16:22 |
| ski | have `like i965-va-driver' (and `:i386'),`intel-media-va-driver' (and `:i386'),`libvdpau-va-gl1' (and `:i386') | 16:25 |
| ski | should i install the missing ones ? | 16:25 |
| gnarface | ski: yes, though before you should do that, check the output of "cat /proc/cpuinfo" to make sure you should be using i386 instead of amd64 (you might have been lied to by intel if that's a core2duo you bought with 32-bit windows pre-installed. there are no 32-bit core 2 duos. | 16:26 |
| gnarface | ) | 16:27 |
| gnarface | heh, for real | 16:27 |
| ski | vendor_id : GenuineIntel | 16:27 |
| ski | model name : Intel(R) Core(TM) i3-3120M CPU @ 2.50GHz | 16:27 |
| gnarface | "Core" | 16:27 |
| gnarface | yea, i'm sorry you installed the whole wrong architecture | 16:27 |
| gnarface | but also, without libgl1-mesa-glx at minimum, none of your hardware acceleration actually works | 16:28 |
| gnarface | mplayer probably is just not skipping because it was created in an era before hardware acceleration | 16:28 |
| ski | "wrong architecture" -- being ? | 16:28 |
| ski | (i have Wine installed, although i haven't used it that much, yet) | 16:29 |
| gnarface | basically you installed the 32-bit linux on a rig that can do both 32-bit and 64-bit at once, but only if the base install components are the 64-bit ones | 16:29 |
| * ski 's no idea of the significance of "Core" | 16:29 | |
| gnarface | if you install those missing mesa packages, hardware acceleration probably will all still work, but you'll lose out on some performance gains that are 64-bit only for other workloads | 16:30 |
| ski | (and it came with 64-bit windows, iirc) | 16:30 |
| gnarface | oh, well that's good. some people got short-changed for the brief period that microsoft didn't have a 64-bit windows offering available | 16:31 |
| gnarface | but that means you should have picked a "amd64" iso instead of a "i386" one | 16:31 |
| gnarface | (i know that might be confusing, but linux calls these "amd64" because despite being intel hardware, AMD was actually the first to market with this general type of cpu) | 16:32 |
| ski | i think `devuan_chimaera_4.0.0_amd64_desktop-live.iso' may be the name of the ISO i installed from | 16:33 |
| gnarface | the significance of "Core" in this context just being that off the top of my head i knew that moniker dates it well past the 32-bit era | 16:33 |
| gnarface | ski: heh, i think i misread you list and skimmed out the "and" parts... maybe you're looking at the right thing. check just check to make sure this says amd64, sorry: umame -a | 16:34 |
| gnarface | sorry again: uname -a | 16:34 |
| ski | Linux foobar 5.10.0-10-amd64 #1 SMP Debian 5.10.84-1 (2021-12-08) x86_64 GNU/Linux | 16:35 |
| gnarface | and then yes, install all those missing mesa packages then reboot | 16:35 |
| gnarface | 5.10 suggests you're not on current stable too | 16:35 |
| gnarface | might not matter but fyi daedalus is released now | 16:35 |
| ski | "chimaera" | 16:35 |
| gnarface | yea... i think you could in theory upgrade this to daedalus in place even though it was a livecd install, but there's a lot more to download then | 16:36 |
| ski | "the \"and\" parts" ? | 16:37 |
| gnarface | start with those mesa packages, though i pasted those from daedalus and i'm not sure the names are all unchanged | 16:37 |
| gnarface | just install the mesa packages and ignore the rest, i got confused there for a minute | 16:37 |
| gnarface | though if you instsalled from livecd i'm not sure why you'd be missing those | 16:38 |
| gnarface | alternately you can try "apt-get update && apt-cache search mesa" as root, and install all those instead | 16:39 |
| gnarface | you should have well over 20 mesa packages if you want everything working | 16:39 |
| ski | i guess i don't need `libd3dadapter9-mesa-dev' ? (and i seem to have `libgl1-mesa-glx-dev' installed, but not `libgl1-mesa-glx', for whatever reason) (also have `libosmesa6' installed) | 16:40 |
| gnarface | yea, maybe not all of them are important for what you're doing | 16:40 |
| gnarface | only thing was i think one of those turned out to be mandatory for DXVK to work right in Wine | 16:41 |
| gnarface | you would also probably need libvulkan too, if that hardware supports it (and i don't know whether it does or not0 | 16:41 |
| gnarface | ) | 16:41 |
| ski | (as well as having `libelg1-mesa-dev', but not `libegl1-mesa', installed) | 16:42 |
| gnarface | have you updated this system after installing it from the iso? maybe an update went awry and removed some things but you didn't notice? | 16:42 |
| gnarface | it's really a mystery that libgl1-mesa-glx would be missing, i could have sworn these isos had all that stuff | 16:43 |
| ski | `mesa-common-dev' wants to also bring in `libdrm-dev' | 16:43 |
| gnarface | that's direct rendering manager, not digital rights management | 16:43 |
| ski | (there's also a `meta-utils-extra', guess i don't need that. it mentions `es2gears',`es2_info',`es2tri') | 16:44 |
| ski | okay | 16:44 |
| gnarface | utils are just for testing that things are working right | 16:44 |
| gnarface | no, you probably don't need them either, but glxgears for example is useful as quick test to make sure opengl acceleration is functioning (it's not meant for benchmarking, but there will still be a noticeable framerate improvement when you have everything working) | 16:45 |
| gnarface | glxinfo is also often useful for troubleshooting | 16:46 |
| ski | `libvulkan-dev',`libvulkan1' (and `:i386') already installed (iirc, i think my brother suggested those, for Wine ?) | 16:50 |
| gnarface | yes, that is what dxvk uses, it's not a stock part of wine but you can get it from winetricks and it can give certain workloads a dramatic improvement (mostly direct3d stuff that translates more poorly to opengl) | 16:51 |
| ski | "have you updated this system after installing it from the iso?" -- no dist upgrade, just installed various packages | 16:51 |
| gnarface | hmm, whenever you install packages, sometimes it will have a REMOVES section | 16:51 |
| gnarface | pay very close attention to that | 16:51 |
| gnarface | you don't want to let it remove stuff you're using | 16:52 |
| ski | hm, okay | 16:52 |
| gnarface | and keep it updated; possibly it did so as a side effect of the other packages not being updated | 16:52 |
| gnarface | ...when you installed things | 16:52 |
| ski | well, it just says "To be installed" and "Unchanged", no "Removes" | 16:52 |
| gnarface | yea, removes won't always be there | 16:52 |
| gnarface | but you gotta remember to always look for it | 16:53 |
| ski | yea, i meant right now | 16:53 |
| ski | mhm | 16:53 |
| gnarface | especially when it's a big package list | 16:53 |
| ski | ok, will try installing those (except `libopenvg1-mesa',`mesa-utils-bin' which are not in the package list) | 16:54 |
| gnarface | you can try "apt-cache search ^libopenvg" to see if maybe it's there under a similar name | 16:55 |
| ski | nothing | 16:55 |
| gnarface | hmm, maybe it's new, not sure | 16:55 |
| gnarface | libvg maybe? | 16:56 |
| ski | nothing | 16:56 |
| gnarface | eh, whatever. i don't think it's important for this | 16:56 |
| gnarface | nope, vector graphics | 16:57 |
| ski | ok, installed those (and saved a log of the conversation, to remember context). i think i'll not reboot right now (since i have stuff running, context i'd prefer to not lose atm), but i'll keep this in mind and see if i notice any difference in the future (might be back later, i guess) | 17:00 |
| ski | thank you for all the diagnostics, suggestions and help ! | 17:01 |
| gnarface | you will definitely have to at least restart the window manager and your graphical login manager to make it work right, and with these libraries you risk weird crashes over time if you don't, just fyi | 17:02 |
| gnarface | it might be easier to just reboot | 17:02 |
| ski | well, i could reboot X (have already done that, today, to see if it might help) | 17:03 |
| gnarface | don't forget to run ldconfig | 17:03 |
| gnarface | (as root) | 17:03 |
| gnarface | whether you have to reload the kernel module too or not i'm not sure, though in this case i doubt it | 17:03 |
| ski | the installation process doesn't run it, automatically, at the end ? | 17:04 |
| gnarface | ...though you should also do a full update after you reboot for this update too anyway | 17:04 |
| gnarface | uh, i don't remember for sure if it runs ldconfig all the time or not, i feel like it might not have, it can't hurt to run it again either way | 17:04 |
| gnarface | if it doesn't complete instantly when you run it, it probably didn't run it | 17:05 |
| ski | well `/etc/ld.so.cache' seem to've been touched, when i installed | 17:07 |
| gnarface | the other package updates you haven't downloaded might contain some stability or performance fixes too, but most of them will probably be security fixes | 17:07 |
| ski | $ find /usr/lib* -name 'libd3dadapter9-mesa*' -o -name 'libgl1-mesa-glx*' -o -iname 'libosmesa6-dev*' -o -iname 'mesa-common-dev*' -o -iname 'mesa-utils*' | 17:12 |
| ski | lists nothing | 17:12 |
| gnarface | what are you trying to do? you list package contents with dpkg -L [package] | 17:13 |
| ski | checking if `ldconfig' had already created links ? | 17:14 |
| gnarface | i wouldn't necessarily expect any of these packages to contain files or directories matching their package names | 17:15 |
| ski | mhm, figures | 17:15 |
| gnarface | try glxinfo and vblank_mode=0 glxgears | 17:16 |
| gnarface | glxinfo |grep -i direct\ rendering | 17:16 |
| gnarface | glxinfo |grep -i version | 17:17 |
| ski | direct rendering: Yes | 17:19 |
| ski | GLX version: 1.4 | 17:19 |
| ski | OpenGL core profile version string: 4.2 (Core Profile) Mesa 20.3.5 | 17:19 |
| ski | (and a bunch more that looks redundant, or not relevant) | 17:20 |
| gnarface | those are probably all right | 17:20 |
| ski | `glxgears' appears to run alright | 17:20 |
| ski | (although i don't really know what i ought to be looking out for) | 17:20 |
| gnarface | i don't know either, for your particular hardware, but what you'd be looking out for usually is a multiple order-of-magnitude increase in the framerate | 17:21 |
| ski | well, i haven't run `glxgears' in many years (and certainly not on this machine) .. | 17:21 |
| gnarface | what do you get? | 17:22 |
| ski | 5978 frames in 5.0 seconds = 1195.530 FPS | 17:22 |
| ski | 6275 frames in 5.0 seconds = 1254.896 FPS | 17:22 |
| ski | then around ther | 17:22 |
| gnarface | hmm, yea, probably working right | 17:22 |
| gnarface | it probably will go up more if you put the cpu in performance mode | 17:22 |
| gnarface | (glxgears on its own isn't heavy enough to trigger the default cpu governor to scale) | 17:25 |
| ski | how would one "put the cpu in performance mode" ? | 17:26 |
| gnarface | echo 'performance' > /sys/devices/system/cpu/cpu0/cpufreq/scaling_governor | 17:30 |
| gnarface | you gotta do this for each core | 17:30 |
| gnarface | echo 'performance' > /sys/devices/system/cpu/cpu1/cpufreq/scaling_governor | 17:30 |
| gnarface | echo 'performance' > /sys/devices/system/cpu/cpu2/cpufreq/scaling_governor | 17:30 |
| gnarface | etc... | 17:30 |
| gnarface | or you can find a gui widget for your desktop maybe somewhere | 17:30 |
| gnarface | you wouldn't want to leave it like this by default though | 17:31 |
| ski | reading from those, it says `schedutil' | 17:31 |
| ski | right | 17:31 |
| gnarface | yea, you can set it to other stuff, schedutil has replaced the previous default "ondemand" but there's a couple others | 17:32 |
| gnarface | there's other info in that directory, like available clock speeds | 17:32 |
| gnarface | oh, that reminds me, make sure you have acpid installed (though i would have thought task-desktop contained that make sure) | 17:33 |
| gnarface | i think you probably also want acpi-support-base if you don't have it | 17:33 |
| ski | `acpid' and `acpi-support-base' already installed | 17:52 |
| gnarface | oh good | 17:54 |
| gnarface | any luck with firefox? | 17:55 |
| gnarface | i don't think mplayer will even use that libva thing by default but there was some way to make it, and it should lower cpu usage further | 17:55 |
| ski | tried rebooting firefox, still stutters. let me `<Ctrl>-<Alt>-<BackSpace>', then | 17:58 |
| gnarface | hmm, i wonder what it's missing | 18:01 |
| gnarface | with intel graphics i am not sure, i probably have forgotten something important, but you can change the xorg driver to the real intel one instead of the default more generic one, which might help | 18:02 |
| gnarface | it helps some stuff and hurts others, it's just a guess really | 18:02 |
| gnarface | varies depending on the particular graphics chip you have | 18:02 |
| ski | hm, black (well, dark grey, i guess) screen, when i try `startx mate-session -- :1.0' | 18:03 |
| gnarface | in general, the default one is better at auto-detection of basic setup stuff, but the intel one sometimes has support for hardware features the default one doesn't | 18:04 |
| ski | oh, timed out now | 18:04 |
| gnarface | your user is in the video group, right? | 18:04 |
| ski | and, retrying, seems to start up X now | 18:04 |
| ski | `id' displays `44(video)', yes | 18:06 |
| gnarface | doesn't really tell me anything, i'm not familiar with their lineup | 18:07 |
| gnarface | to try the intel xorg driver, you'd just have to create a few lines xorg.conf (you don't need the whole thing anymore) and put it in /etc/X11/ | 18:09 |
| gnarface | then restart the gui stuff again | 18:09 |
| gnarface | you'd see the change appear in the xorg log, which should be at ~/.local/share/xorg/ | 18:10 |
| gnarface | maybe /var/log/Xorg.0.log but probably ~/.local/share/Xorg.0.log | 18:10 |
| gnarface | er, ~/.local/share/xorg/Xorg.0.log | 18:10 |
| ski | `/var/log/' has `Xorg.0.log{,.old}', `~/.local/share/xorg/' has `Xorg.[123].log{,.old}' | 18:12 |
| gnarface | well, only one of those should have a timestamp of when you started xorg last | 18:13 |
| gnarface | though i guess it might be possible that you'd get a different one for the gui login itself | 18:13 |
| gnarface | afaik .1, .2, .3 etc are :1.0, :2.0, :3.0, etc | 18:14 |
| gnarface | and it copies the last one to .old every startup | 18:14 |
| ski | hm, seems Firefox restarted now, let me try .. | 18:15 |
| ski | .. hm. no sound at all | 18:15 |
| * ski checks sound preferences | 18:16 | |
| gnarface | just in firefox or everything? | 18:16 |
| ski | hm, oh, now i've got sound | 18:16 |
| gnarface | did pulseaudio get pitched overboard when you were doing installs? | 18:16 |
| ski | guess either muting and unmuting, or unplugging and replugging headphones helped that | 18:16 |
| ski | .. and there appears to be not stuttering at all ! | 18:17 |
| gnarface | oh, auto-mute mode might be screwing with you. you can disable that | 18:17 |
| ski | "afaik .1, .2, .3 etc are :1.0, :2.0, :3.0, etc" -- yep | 18:17 |
| gnarface | (but then you will have to mute/unmute your headphones manually) | 18:17 |
| gnarface | headphones/speakers* | 18:18 |
| ski | "did pulseaudio get pitched overboard when you were doing installs?" -- no idea (and no idea how to check) | 18:18 |
| ski | oh, i guess you were asking if it got uninstalled ? | 18:18 |
| ski | (i suppose it might use `~/.local/share/xorg/', when i manually use `startx', from my user) | 18:19 |
| ski | ("doesn't really tell me anything, i'm not familiar with their lineup" -- i did `id' from my user, which tells me which groups i'm in) | 18:22 |
| gnarface | yes, home directory when you use startx and /var/log when it's running in "legacy" root suid mode | 18:22 |
| ski | makes sense | 18:22 |
| gnarface | oh sorry, i thought you were pasting the video device id lol | 18:22 |
| ski | well .. i dunno how to check that :p | 18:23 |
| gnarface | actually it's not based on whether you use startx or not | 18:23 |
| ski | but which user is running it ? | 18:24 |
| gnarface | nvidia binary drivers also use legacy mode by default | 18:24 |
| gnarface | regardless | 18:24 |
| ski | mhm | 18:24 |
| gnarface | in your case startx will just run it as your user though | 18:24 |
| gnarface | what i don't know is if your graphical login is running as root or not | 18:24 |
| ski | USER GROUP PID PPID PGID SID %CPU %MEM VSZ RSS TTY TPGID STAT STIME TIME CMD | 18:26 |
| ski | root root 32375 32372 32375 32375 3.5 0.1 1004044 12912 tty7 32375 Ssl+ 13:04 00:11:26 \_ /usr/lib/xorg/Xorg -nolisten tcp -auth /var/run/slim.auth vt07 | 18:26 |
| ski | while `mate-session' runs as my user | 18:26 |
| gnarface | maybe that's where the /var/log/Xorg* logs are coming from | 18:28 |
| ski | well .. as the video replay issue appears to be resolved now (although i might have to restart my original X session, if i want it to work in there), thanks a whole bunch for the tele-diagnostication session and the suggestions ! | 18:39 |
| gnarface | no problem | 18:40 |
| ibanja | I am trying to install using ZFS via debootstrap and got stuck on the zfs-zed package. The error is here: https://paste.debian.net/1307728/ | 20:46 |
| rwp | ibanja, It looks to me that the package postinst script is failing while calling an init script action. It looks like you are running this in a chroot? | 20:53 |
| ibanja | yes, a chroot | 20:53 |
| ibanja | I an using one kernel older than the chroot. Is that a problem? | 20:53 |
| rwp | The package postinst script is available at /var/lib/dpkg/info/$PACKAGE.postinst and the best thing is to review it and run it and see what is failing. | 20:53 |
| ted-ious | ibanja: Is this a vm or a real server? | 20:54 |
| rwp | Generally one might run into problems with hosting using an older kernel than in the chroot but Linux is very long-time compatible and I don't think that is the problem shown here. | 20:54 |
| ibanja | it is a chroot installation on an SSD drive on my desktop | 20:54 |
| ted-ious | Oh so you can get physical access to everything no problem right? | 20:55 |
| ibanja | yes | 20:55 |
| rwp | Review the package postinst script by eye using less to browse it: less /var/lib/dpkg/info/zfs-zed.postinst | 20:55 |
| ted-ious | You might find it way easier to do a regular install onto ext4 and get all the zfs parts working properly first. | 20:55 |
| ted-ious | Then tar it all up and reformat the drives with zfs and copy the system back. | 20:56 |
| _ds_ | Chrooted environment. Only one kernel in use. | 20:56 |
| rwp | For GNU/Linux systems I would recommend against trying to boot from ZFS. I would boot from XFS (better than ext4) and then use ZFS as the data array. Since ZFS is bolted on using dkms recompiles with every new kernel install. | 20:57 |
| _ds_ | I'd just “mount --rbind /dev /path/to/chroot/dev” in the parent environment. | 20:57 |
| _ds_ | (unless the chrooted env. is intentionally limited in some way in which case, well, create your own device nodes etc.) | 20:58 |
| ibanja | rwp: I have ZFS on my voidlinux box and it recompiles every time it's upgraded. I'm OK with that. | 20:59 |
| ibanja | _ds_: I just exited chroot. I will chroot and try mounting the dev | 21:00 |
| rwp | +1 for _ds_'s suggestion to mount /dev into the chroot. Because almost certainly the zfs init script won't be functional with out. And almost certainly the postinst script is exiting with an error if it is not functional. | 21:01 |
| _ds_ | Also worth mounting similarly: /sys and /proc | 21:01 |
| rwp | Just by comment for general hacking, sometimes due to reason of bugs or for bending the machine to your will, it is sometimes needed to edit the postinst or the pre* scripts to avoid running them. For example by putting an "exit 0" immediately following the #!/bin/sh line. Then they do nothing, and exit successfully, allowing dpkg to mark the package as fully installed. Or fully removed/purged. | 21:03 |
| rwp | And then after that point doing further hacking and doing whatever it is that needs to be done manually. | 21:03 |
| ibanja | I have /sys /proc and now /dev mounted. Unfortunately it didn't fix the problem. I will have a look at the postinst script. | 21:03 |
| ibanja | rwp: good idea... | 21:04 |
| rwp | Almost certainly the postinst script is calling into another script (probably an init script) and depending upon the source of this package it might be calling systemd capability which does not exist on Devuan. This might require editing of either the postinst script or the init script to fix it. | 21:04 |
| rwp | One of the things I really like about dpkg (over rpm and some other formats) is that the pre and postinst scripts exist as regular files on disk available to browse and to hack upon if needed. Have used that capability many times. | 21:06 |
| ibanja | ok, thanks for the tip. I'll digging into the postinst now. | 21:06 |
| _ds_ | Worth checking what's currently mounted within that chroot env. in case /dev/pts (still) isn't (just --bind, not --rbind, for that one). | 21:07 |
| rwp | The postinst script is expected to be given a couple of arguments when being configured. "sh -x /var/lib/dpkg/info/zfs-zed.postinst configure 2.0.3-9+deb11u1" I think for your specific case should trace through the script operation. I think. Double check that. | 21:09 |
| ibanja | rwp: I put the exit 0 after bash and was then able to remove and reinstall and that worked. Thanks for the help. | 21:23 |
| ibanja | _ds_: yep mounted... thanks | 21:24 |
| rwp | That just allows the packages to be marked as installed and to get past that packaging step. Afterward you will need to manually understand and do whatever that postinst attempted to do. | 21:26 |
| rwp | Of course if what it was attempting to do only makes sense at boot time then it can be ignored entirely. | 21:26 |
| rwp | And then at boot time the installed init scripts will do what they are supposed to do at boot time okay. | 21:26 |
| rwp | I'll mention as a statement off in the sidebar that if one is maintaining a persistent chroot that the /usr/share/doc/init-system-helpers/README.policy-rc.d.gz layer can be used to prevent daemons from being started in a chroot at package upgrade times. The invoke-rc.d script (which was mentioned in your paste) uses that policy layer. | 21:28 |
| ibanja | ok, thanks. | 21:38 |
| rwp | Are other people having problems with paste.debian.org resolving? My system validates DNSSEC and is hitting the "validating apu.snow-crash.org/CNAME: verify failed due to bad signature (keyid=41523): RRSIG has expired" problem. | 21:50 |
| rwp | Here is an automated report: https://dnsviz.net/d/snow-crash.org/dnssec/ | 21:50 |
| fsmithred | rwp, yes ff says it can't find the site. I always use paste.debian.net and that is working. | 21:54 |
| rwp | I can't resolve paste.debian.net here due to the DNSSEC errors voiding it. I can override it with a /etc/hosts entry, which I have done in order to see pastes. But... | 21:56 |
| rwp | I guess I should not expect anything to be fixed on a weekend, and Monday all IT techs will always be busy, and should wait until Tuesday before getting excited about really getting this fixed. | 21:58 |
| rwp | It's just bugging me because I tend to use https://paste.debian.net/ a lot and like in the above I can't see other people's pastes if I can't resolve the IP address of it. | 21:59 |
| fsmithred | I'm using opennic in /etc/resolv.conf | 22:01 |
| rwp | Firefox using DoH for DNS-over-HTTPS is also complaining so I assume the upstream DoH provider (CloudFlare here) is also voiding it due to the DNSSEC. But I can /etc/hosts override to get to it as a temporary workaround. | 22:10 |
| devurandom | I recently updated my devuan system and I am having performance problems in graphical applications such as games. My graphics card is a Nvidia GeForce GTX 1050 Ti and I am using the proprietary nvidia driver. Anybody else having this problem? | 22:13 |
| devurandom | and yes I have restarted my machine. | 22:14 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!