libera/#devuan/ Saturday, 2025-02-01

gnarfaceNICAD: the motherboard bios setup panels should have an option to set the expansion card as your default display adapter instead of the motherboard one. if you don't see it right away, try checking for something like "advanced options"02:31
gnarfaceif you do that, xorg will most likely pick it up automatically then, however a custom xorg.conf can do the same thing (and more!) let me know if you need help fabricating one...02:32
gnarface...well, it won't actually do quite the same thing in this case, your system boot-up and any console output before Xorg comes up will probably still go out the motherboard port, and maybe both or maybe only the motherboard port, so the BIOS setting is the preferred first step anyway, if furnished02:34
gnarfaceNICAD: *but with a custom xorg.conf, you could even use both at once for different displays03:27
gnarfacesome window managers may have built-in controls to set that up too, but not all of them03:27
gnarface(and if you want to try changing the setup by command-line after Xorg is started but your window manager doesn't have that feature, check out the xrandr binary)03:28
freemangordongnarface: any idea how to avoid this https://pastebin.com/n08q2LKT08:30
gnarfacefreemangordon: hey man, sorry but i've got real mental issues, i'm way too paranoid for pastebin, but if you put it at paste.debian.net or just /msg it to me, i'll take a look09:10
freemangordonoh, sorry, will know for the future :)09:10
gnarface(anti-flooding bot in this room i think triggers at either 3 or 4 lines)09:10
gnarfacei will take a look though if you deliver it in one of those two ways09:11
freemangordonhttps://paste.debian.net/1347828/09:11
gnarfacethanks09:11
gnarfacefreemangordon: interesting, this error "cycle found while processing triggers" is not one i've seen before, but it seems to be referring to the pre/postinst scripts in these packages causing the package installation equivalent analog of a http redirect loop09:13
gnarfaceusually in the stable distro, this type of thing is caused by packages from out of distro, or from another release09:14
gnarface(like backports)09:14
freemangordonthis is caused by gconf, for sure09:14
gnarfacebut it might just be a simple glitch09:14
gnarfaceto figure it out in more specifics, i'd have to actually see the pre/postinst scripts ... they're capable of quite a lot of havoc, but they're usually fine unless you are pulling from unstable09:14
freemangordonI need a minute to confirm, but installing gconf2-common_3.2.6-8_all.deb before the upgrade seems to fix it09:15
freemangordonso maybe gconf2 shall pre-depend on gconf2-common09:15
freemangordongconf2 has triggers on files that are distributed by gconf2-common09:16
gnarfaceyea, it might be something like that. unfortunately packaging isn't really my area of expertise, i've only dabbled with it a bit, but it seems like it's saying that gconf2 and sgml-base have triggers that both retrigger each other in a easily detectable but neverending loop09:16
freemangordonyeah09:17
gnarfacelike they each say to preconfigure the other first, and neithers' scripts can ever exit09:17
gnarfaceit's not something i can recall seeing myself or hearing anyone else complain about in daedalus around here, but stick around and maybe someone else has09:18
freemangordonright, but this prevents dist-upgrade from chimaera to daedalus09:18
gnarfaceyea, that part is weird, i've done a lot of dist-upgrades from chimaera to daedalus09:18
gnarfaceso it might be something you guys added09:19
freemangordonI don;t think so09:19
gnarfacebut it might be something super innocuous seeming like a change of location of some file it's checking09:19
gnarfacealthough... hmm09:19
gnarfacesomewhere a long time ago in the debian release notes for some pre-devuan debian release, i do recall seeing the advice to "apt-get upgrade" then "apt-get dist-upgrade" as a second step, when doing upgrades - after changing the sources.list to the target release, to avoid... something09:20
gnarfacemaybe something like this09:20
freemangordonok, confirmed https://paste.debian.net/1347829/09:21
freemangordonlemme check if dist-upgrade will work09:21
gnarfacei'd only ever had to actually do that once before myself, and i don't think it was on daedalus, but it could be different on arm/arm64... most my dist-upgrades have been on x86 architectures09:21
freemangordonthis is amd64 VM09:21
gnarfaceoh, hmm09:21
freemangordonyeah, it will take ages on devices :)09:22
gnarfaceonly other thing i can think of as a quick fix without actually debugging the hooked scripts line-by-line would be to simply remove one or both of the problematic packages before the upgrade, then re-install that package afterwards09:22
freemangordonremoving gconf2 will remove all the leste, basically :)09:23
gnarfacehow about the sgml-base?09:23
gnarfacedoes gconf2 depend on that?09:23
freemangordonthere is no issue with it09:23
freemangordonthats a red herring09:23
gnarfacehmm09:23
gnarfacedo stick around, there's smarter people than me around here, but usually they're asleep around now09:24
freemangordonyeah, ok, thanks09:24
gnarfacesystemdlete: by the way, in case nobody notified you of this already: <plasma41> If systemdlete returns, please let them know that '[ -d /run/systemd/system ]' is, as far as I'm aware, a reliable check for a running systemd.10:25
gnarface<plasma41> As an example, I used this exact check this week in my patchset submission to the snapper package: https://bugs.debian.org/cgi-bin/bugreport.cgi?att=7;bug=976888;filename=v2-1007-Make-cronjobs-defer-to-systemd-timers.patch;msg=2210:25
freemangordongnarface: in case anyone you know might be interested https://github.com/maemo-leste-upstream-forks/gconf/commit/779bdb1e11038462dc4c33a1064f0ffc3de926a012:39
systemdletegnarface, thanks.  That sounds ideal.14:58
AlverstoneIs there a conventional method so saving the bootlog? I need to save the shutdown log. Inject a script into shutdown sequence and dump /dev/vcs*?19:05
nemoAlverstone: /var/log/boot.log ?19:18
nemohuh. that's weird... I thought devuan had a /var/log/boot or something whaaat19:20
nemoI guess there's /var/log/dmesg but..19:21
nemoguessing that's not what you want19:21
nemohm... is there anything in the shutdown that would not be in dmesg...19:23
Alverstoneyes. the messages it prints on the screen19:24
nemoAlverstone: those aren't echoed to dmesg too?19:24
Alverstonethough I decided to abandon all pride and record it on video19:24
Alverstonenemo, why should they?19:24
nemodunno. don't know how the kernel logs things, but looks awfully similar to what I see at boot and shutdown19:25
Alverstonesysvinit scripts don't do anything special, just echo19:25
nemoohhhh the script output19:25
nemoduh19:25
nemo:D19:25
nemosorry was focused on kernel shutdown19:25
Alverstone /lib/lsb/init-functions is pretty simple19:25
nemoAlverstone: ok. I get it now. that's definitely interesting and useful, and yeah, AFAIK unless the script is written to do logging...19:26
AlverstoneI guess I could make a script that would execute as the last in shutdown sequence, mount / back to readwrite, dump /dev/vcs and shutdown, but duh19:26
nemoAlverstone: hm. quite a few of the devuan scripts do use syslog19:27
nemowhich ones don't19:27
nemooh wow. most of them don't19:27
nemoah. I see. what you pointed out. they are using the generic functions.19:28
nemook. that improves the % in my grep a lot ☺19:28
nemoand yeah. since they are using the generic one, adding a hook in there besides echo is probably pretty easy. cool19:29
nemoI wonder why that isn't default19:30
Alverstoneactually I tried redirecting all output into a file by altering the init, but for some reason the system didn't boot at all, just hang19:30
Alverstoneweird19:31
Alverstoneyou'd think magic should be simpler19:31
nemoredirecting all output of...19:31
Alverstoneinit=/usr/local/bin/dumpinit.sh: exec /sbin/init > /var/tmp/lol 2>&119:32
Alverstonedumpinit.sh*19:32
Alverstonedumpb19:32
Alverstonedumb19:32
* Alverstone is angry19:32
nemoummmm19:32
nemoI don't have dumpinit.sh and I don't know what the original is, but I can see messing with something like that possibly screwing up backgrounding if you aren't careful with quoting19:33
nemowhich would hang things19:33
nemooh. I see.19:34
nemoyou were trying to replace init with your custom wrapper19:34
nemoone that captures stdin/stderr, with the logic that all subs would echo to the same stdin/stderr19:34
nemoI don't know if that part is true but...19:34
nemoyou would think init would have a way to do that..19:35
nemohm19:36
nemoI guess another issue with your wrapper19:36
nemoit doesn't handle arguments19:36
nemoreading the man page for init, init definitely takes args19:37
nemoso you should pass them in to the wrapper call as well19:37
nemorunlevel being the one most likely to cause hangs I guess19:37
nemosince it wouldn't do anything at all otherwise?19:37
nemootherwise the idea seems cool and I'm surprised init doesn't have logging functionality already19:38
nemoAlverstone: IMO adding the logging to the initfunctions wuold be less fragile19:47
nemo.. welp. curious if you come up with something that works well (or exists already) sounds useful19:50
Alverstoneand how will it work, exactly? syslog requires a daemon, which is not available during boot and is stopped during shutdown. and what will you do after all filesystems are unmounted and root remounted readonly19:52
fsmithredAlverstone, there's a package called bootlogd19:57
Alverstone>bootlogd logs all messages printed to the system console during system boot19:58
fsmithredThe first line in my /var/log/boot says "Activatinjg swap"19:58
Alverstoneany luck with shutdown?19:58
fsmithredsyslog notes when you shut down, but I don't know what it keeps.19:59
fsmithrednothing about shudown in boot19:59
fsmithredhey if you decide that the way to get it all is to switch to initramfs to shut down, please tell me how you did it.20:01
fsmithredI have a dream of someday eliminating my cryptsetup error messages.20:01
AlverstoneI am not sure you can jump back to initramfs after you booted, though20:02
Alverstoneplus pivot_root removes the previous root recursively20:02
Alverstonewhich is a funny gotcha20:02
fsmithredoops20:03
fsmithredI didn't know that20:03
AlverstoneI don't have experience using custom initramfs hooks though, but dumping /dev/vcs is the only thing I can come up with, if you mean you have cryptsetup errors during boot20:04
AlverstoneBy the way why exactly you can't see them? I though it'd pause to prompt password20:04
fsmithredwhen I boot I see cryptsetup messages about bad shutdown20:11
fsmithredI'll see if I can find one20:12
masonfsmithred: What cryptsetup error messages? Timeouts when you shut down?20:13
fsmithrednothing in var/log20:14
fsmithredI think I will reboot and take a picture20:14
masonfsmithred: Those have existed for a few years now at least. I reported the issue but the general consensus was that it didn't exist, "doesn't affect me", so my fix is my own.20:14
fsmithredno timeouts20:14
masonah20:14
fsmithrednot that it doesn't exist, but that it doesn't matter20:14
fsmithredeverything still works20:14
masonWell, an extra minute waiting for shutdown matters.20:14
fsmithredbut cryptsetup close whatever does not happen at shutdown20:15
fsmithredno extra waiting20:15
masonInteresting.20:15
fsmithredthis is known issue for years20:15
masonAlright, yeah, that would generally need a shutdown-ramfs because as noted you want to pivot to free up the root mount, so you can unmount it. Even systemd as I understand it just eats the error quietly.20:16
masonFor me, if I don't say "skip=y" in cryptdisks-functions there's a minute or so of fumbling as it tries to deactivate the storage backing root and fails.20:17
Alverstonesomething very ugly I once did is deliberately break initramfs so it doesn't boot. then chroot into real root and init the system manually using scripts. maybe, just maybe, think could be used to do what you want20:20
masonAlverstone: You can always just modify the init script in the initramfs so it gives you a shell all the time. Or not modify it, and tell it you want a shell.20:21
Alverstonelater I remembered you can do init=/bin/sh20:22
fsmithredmason, cryptsetup-modified-functions might eliminate the delay. (it's a package)20:22
Alverstonewas enough for my purposes20:22
masonAlverstone: Oh yeah, that too!20:23
fsmithredactually, I think youknow that pakcage well.20:23
masonfsmithred: Ah, hadn't seen that. I bet it does more or less what I'm poking in with Ansible.20:23
fsmithredpretty sure you worked on that one with me at one time. (dpkg-divert)20:23
masonfsmithred: If I do it's slipped away from me.20:23
masonOh, maybe.20:23
fsmithredfwiw, I thought it was no longer needed, but the author said it is needed in some cases (his) so I updated it.20:24
masonThe need for a patch shows up for me mostly because of ZFS root.20:25
fsmithredtake a look at it and see if it helps.20:26
fsmithredor feel free to improve it20:26
masonInteresting, this is very different from what I do: https://git.devuan.org/devuan/cryptsetup-modified-functions/src/branch/master/cryptdisks-functions20:30
masonSo, they've removed the whole geometric backoff thing.20:33
masonThe same thing can be achieved in the upstream script by saying "skip=y" at the top of _do_stop_callback()20:34
masoncryptsetup-modified-functions is also specific to LVM, which I don't actually use.20:35
masonBut still, thank you for showing it to me, or reminding me of it anyway.20:35
masonBoth approaches are just a way to say "how do we handle the situation where we'll quite literally never be able to release the underlying crypto device?" But it ends up not being a huge deal as long as your filesystem that's sitting on it is relatively coherent at shutdown.20:37
Alverstonewhen i used systemd it introduced some stupid bug that prevented /home from being unmounted. I didn't care to fix it for a month maybe and everything worked fine. I don't remember what was the issue though20:44
AlverstoneOne of the things that upsets me about runit is that sometimes it silently spams my pid table. I just started to wonder how the heck my PID got > 150000, appears polkitd fails to start because systemd devs enjoy breaking compatibility (some symbols missing or whatever). It does not even output any messages anywhere, I'd never notice otherwise22:17
AlverstoneYou'd think runsvdir would notify you, no?22:18
Alverstonewhat I end up doing is `sv status /etc/service/* | grep 'want up'`22:19
schillingklausone reason more to boycott polit and everthing depending on it22:20
Alverstonefucking stupid piece of shit, not even a way to install from daedalus (I use excalibur)22:25
AlverstoneI don't give a fuck22:25
AlverstoneI removed it22:25
Alverstoneand let it burn22:25
Alverstoneif something will break22:26
AlverstoneI don't give a damn22:26
AlverstoneIt's not my problem they're retarded22:26
* Alverstone is very angry22:26
* Alverstone is sorry too22:26
AlverstoneIs there a daemon that will shutdown my system when power gets too low? I use laptop22:27

Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!