| darwin | seems rc.local didn't run on start, despite it's 755 | 02:53 |
|---|---|---|
| darwin | or /run/user/1000 wasn't created, or maybe it was but deleted twice (used to be created, but now I had to create it using rc.local, which doesn't happen either) | 02:54 |
| gnarface | did you install the "initscripts" package? | 02:54 |
| darwin | this prevents start-pulseaudio-x11 from running | 02:54 |
| darwin | initscripts is installed | 02:55 |
| gnarface | you using something other than sysvinit then i assume? | 02:55 |
| darwin | of course not | 02:55 |
| gnarface | hmm, odd. which release? | 02:55 |
| darwin | 5 | 02:56 |
| gnarface | is that... daedalus? | 02:56 |
| darwin | yes | 02:56 |
| gnarface | cannot reproduce | 02:56 |
| gnarface | /run/user/1000 though is an elogind thing i think | 02:56 |
| darwin | maybe rc.local did run but for some reason 'mkdir /run/user/1000' and its chown didn't happen | 02:57 |
| darwin | i have elogind also | 02:57 |
| gnarface | completely optional, should not be related to rc.local | 02:57 |
| gnarface | try a simpler test of /etc/rc.local like tell it to write a file to /tmp/ | 02:58 |
| darwin | it did the commands I put in there when I ran it | 02:58 |
| darwin | but it also has other stuff now I didn't put in there, which is odd and should be in its own system scripts | 02:58 |
| darwin | it has 'if test -d /etc/boot.d ; then' 'run-parts /etc/boot.d' 'fi' | 02:59 |
| darwin | (three lines) | 02:59 |
| gnarface | those are stock additions | 02:59 |
| onefang | You replaced rc.local, or put stuff in /etc/boot.d/ which rc.local runs through. | 02:59 |
| darwin | i did not replace rc.local nor put anything in /etc/boot.d | 02:59 |
| onefang | These days rc.local will run any scri.t it finds in /etc/boot.d, which is the proper way to add youc own stuff. | 03:00 |
| darwin | i don't even have /etc/boot.d, in fact. Maybe that's the problem | 03:00 |
| gnarface | the "run-parts /etc/boot.d" thing was added to rc.local stock, i'm not sure when | 03:00 |
| gnarface | it's possible you upgraded and initially kept your old one but a later update overwrote it if it didn't look like it had been customized or something... | 03:01 |
| gnarface | no idea why /run/user/1000 isn't getting created, but keep in mind if your UID isn't 1000 then it'll be that number instead | 03:02 |
| darwin | it was a fresh installation of Devuan 5, not upgraded from 4 | 03:02 |
| onefang | I promise I'll type better after lunch. lol | 03:02 |
| darwin | it is 1000 | 03:02 |
| gnarface | do you see elogind running? | 03:02 |
| darwin | elogind-daemon is running | 03:02 |
| gnarface | and dbus? | 03:02 |
| darwin | /usr/bin/dbus-daemon --system and longer versions are running | 03:03 |
| gnarface | "elogind-daemon" that's weird | 03:03 |
| darwin | i should've put that in quotation marks | 03:03 |
| gnarface | it's just called elogind here | 03:03 |
| gnarface | er, wait... | 03:03 |
| gnarface | no, nevermind, elogind-daemon is right | 03:04 |
| gnarface | which login manager? | 03:05 |
| darwin | none | 03:05 |
| gnarface | hmm, might be part of the issue | 03:05 |
| gnarface | some user processes might not start without it | 03:06 |
| gnarface | i recommend lightdm just because it's given me the fewest problems | 03:06 |
| gnarface | that probably explains why pulseaudio isn't starting too | 03:07 |
| darwin | it did start before I set 'allowed_users=anybody' | 03:07 |
| darwin | /etc/X11/Xwrapper.config | 03:08 |
| onefang | Not just anybody can take your pulse, has to be qualified medica.... er carry on. | 03:08 |
| darwin | i did that to solve some other problem I forgot | 03:08 |
| darwin | that created at least two more problems | 03:09 |
| gnarface | well, on the one system i have here with elogind and lightdm and pulseaudio, i didn't do anything weird to the configuration and this is all working normally | 03:13 |
| gnarface | on the other systems where i don't have any login manager, i also don't have pulseaudio | 03:14 |
| gnarface | /etc/rc.local works on all of them though, independently of all the login/session/permissions management framework stuff | 03:14 |
| gnarface | so some of this might be a misdiagnosis, and some of it appears self-inflicted | 03:15 |
| gnarface | it is not unexpected for a bunch of random crap to show up in /tmp/ though and most of it i'm guessing isn't related to the issue at hand | 03:16 |
| gnarface | i don't think pulseaudio has to be run as your user, i think that's just the new hot way to do it. i think you might be able to change the pulseaudio config so it can be started directly at startup and just used as a system daemon by any user instead. i think that's actually the way it used to be. | 03:18 |
| gnarface | and if you're not using any login manager i wouldn't really expect /run/user/[UID] to be of any use at all | 03:19 |
| gnarface | hmm, i just had another idea about pulseaudio... maybe you could just add it to your window manager's startup programs directly instead? | 03:28 |
| darwin | it already is | 03:29 |
| gnarface | oh, hmm | 03:29 |
| gnarface | well, i'm not really completely aware of what all /run/user/[UID] is even for, or if there's a way to initialize it all without the rest of the normal software stack components, maybe someone else around here has ideas about that... | 03:30 |
| freem | <gnarface> no idea why /run/user/1000 isn't getting created, but keep in mind if your UID isn't 1000 then it'll be that number instead | 03:35 |
| freem | Is there is a /run/user folder, to start with? I hope the question was not asked yet... | 03:35 |
| gnarface | no, it was not and it's a good question | 03:37 |
| freem | as for this kind of stuff... I myself create a /dev/shm/${uid} that I export as XDG_RUNTIME_DIR, even if not using PA nor a login manager. I still have not migrated everything relevant to use that instead of /tmp though | 03:38 |
| freem | I never heard about /run/user before, it's why I asked. And it's definitely absent from my system (not surprising though, my system is weird) | 03:39 |
| freem | am I right to suspect that this folder is used for the same thing I created /dev/shm/${uid}, that is, to be used as XDG_thingy? | 03:40 |
| * gnarface doesn't know | 03:41 | |
| freem | this actually reminds me I *should* make ssg-agent create it's stuff in $XDG_RUNTIME_DIR instead of /tmp... I happen to consider /tmp as a scratchpad, and happily `rm -rf /tmp/*` from time to time to get some cleanup (it's a ramdisk here so I don't really enjoy it full of crap) which ofc makes my ssh-agent unusable... | 03:42 |
| rrq | "What is this folder /run/user/1000? .. Asked: 10 years, 1 moonth ago" | 03:42 |
| rrq | "171 | 03:43 |
| onefang | I guess a moonth is longer than a month. | 03:43 |
| freem | rrq: and the answer makes no sense on a devuan system... I quote (if that's the same link): "/run/user/$uid is created by pam_systemd" | 03:43 |
| freem | https://unix.stackexchange.com/questions/162900/what-is-this-folder-run-user-1000 is the one i found. And it turns out my guts were right, it seems to be mostly to store XDG_RUNTIME_DIR and other temp stuff | 03:44 |
| freem | (still thanks for the reminder to search, though, and sorry for asking I guess) | 03:45 |
| rrq | right; supposedly a per-user /tmp space | 03:45 |
| rrq | gnupg puts its socket(s) there | 03:46 |
| freem | apparently for ssh-agent I'll have to override $TMPDIR when starting it | 03:47 |
| * freem tries | 03:47 | |
| freem | hm... can't make stuff work that way. OTOH my user-land daemon management is a mess | 03:57 |
| freem | lol | 04:00 |
| freem | In Debian, ssh-agent is installed with the set-group-id bit set [...] This has the side-effect of causing the run-time linker to remove certain environment variables [...] including LD_PRELOAD, | 04:01 |
| freem | LD_LIBRARY_PATH, and TMPDIR. | 04:01 |
| freem | FILES | 04:01 |
| freem | $TMPDIR/ssh-XXXXXXXXXX/agent.<ppid> | 04:01 |
| freem | buggy doc. | 04:01 |
| freem | annoying, wonder if I can overcome this somehow | 04:01 |
| freem | sometimes I wonder if there is not a bit too much of security measures | 04:02 |
| freem | preventing to run ptrace(2) against this kind of process makes sense... on a multi-user system, but on a single-user one like a laptop or desktop? I wonder. | 04:03 |
| gnarface | it's probably not a difficult change to make in the code, but since it's ssh there's probably a good reason | 04:03 |
| freem | well... I have to say, I do not really trust debian on their ssh patches | 04:03 |
| freem | I know of at least 2 big fails there, one quite old which made bunch of keys fragile and the recent systemd one | 04:04 |
| gnarface | fair enough | 04:04 |
| freem | the good reason is described in the man-page, though | 04:04 |
| freem | In Debian, ssh-agent is installed with the set-group-id bit set, to prevent ptrace(2) attacks retrieving private key material. This has the side-effect of | 04:04 |
| freem | causing the run-time linker to remove certain environment variables which might have security implications for set-id programs, including LD_PRELOAD, | 04:04 |
| freem | LD_LIBRARY_PATH, and TMPDIR. If you need to set any of these environment variables, you will need to do so in the program executed by ssh-agent. | 04:04 |
| freem | but then they should probably have patched it to first take XDG_RUNTIME_DIR instead of... oh well, guess I can just check if I can patch it myself | 04:05 |
| freem | fun patch: % upl user-group-modes.patch -k | 04:07 |
| freem | https://p.mort.coffee/U1u | 04:07 |
| freem | Allow secure files (~/.ssh/config, ~/.ssh/authorized_keys, etc.) to be | 04:08 |
| freem | group-writable, provided that the group in question contains only the file's | 04:08 |
| freem | owner. Rejected upstream for IMO incorrect reasons | 04:08 |
| freem | the heck about accepting group writes in there "only if that group only have 1 user"?! | 04:08 |
| freem | anyway, that feature was even rejected by openssh's upstream. I should follow the links to make my opinion... | 04:10 |
| gnarface | you could always just use the upstream openssh version instead, i doubt it would be hard to build and package... | 04:13 |
| gnarface | ... not all debian patches are bad though | 04:13 |
| freem | ofc not all of them are bafd | 04:13 |
| freem | I trust debian a lot, despite what I said above. But debian maintainers are also humans who do mistakes | 04:13 |
| freem | i.e. adopting systemd and all it's bullshit changes | 04:14 |
| freem | and here, I clearly disagree with the maintainer | 04:14 |
| freem | https://bugzilla.mindrot.org/show_bug.cgi?id=1060#c3 | 04:14 |
| freem | Suggestion 5 is about trawling through users' home directories on package | 04:14 |
| freem | installation, which is entirely unacceptable and a cure worse than the disease. | 04:14 |
| freem | Sorry. | 04:14 |
| freem | well, sorry, but *I* would have prefered to be able to put the damn files outside of /tmp, as they are part of the functionning of my system, and I use /tmp as a junkyard for temporary files, not files necessary for software to function | 04:15 |
| freem | so I would have prefered a warning rather than a "reduce constraint" which actually removes features | 04:16 |
| freem | I still have to read the whole stuff, I might change my mind. I'll probably also go on openssh's IRC to know if using vanilla I'd be able to do what I have in mind | 04:16 |
| freem | I'm surprised there's that many patches in openssh-client though. 36 patches is not nothing | 04:22 |
| AlexLikeRock | o/ | 05:54 |
| AlexLikeRock | gnarface, | 05:54 |
| darwin | i think what I had to do was create /run/user/1000 and then run start-pulseaudio-x11 as root | 05:56 |
| AlexLikeRock | i remember pulse audio : no need run as root | 05:58 |
| darwin | i don't remember why I changed 'allowed_users=console' to 'allowed_users=anybody' but now I can't actually login as root then wheel then startx--I have to login as wheel first, so that line doesn't seem to do what it says. Maybe before that I could no longer startx at all | 05:59 |
| AlexLikeRock | cool | 06:01 |
| darwin | no | 06:01 |
| freem | darwin: wdym you mean you can't login as root then startx? What's the error message? | 06:04 |
| darwin | it needs a console user | 06:05 |
| darwin | that's not what I said | 06:05 |
| freem | sorry then, failing to understand what you said I guess | 06:06 |
| darwin | i said I login as root then wheel | 06:06 |
| darwin | so the wheel user isn't a console user--root is there | 06:06 |
| freem | ah... so, login as root, changing user with `su -l` and then running startx? | 06:07 |
| darwin | 'su -' | 06:07 |
| darwin | 'su - user' | 06:07 |
| rrq | Xorg requires teh running user to own the /dev/ttyN devnode for the virtual terminal | 06:07 |
| freem | '-', '-l' and '--login' are the same. I only keep the -l because it's easier for me to remember, and less confusing | 06:08 |
| freem | but right, - is shorter | 06:08 |
| darwin | the line I changed should've removed that requirement | 06:08 |
| rrq | I don't know what Xwrap does | 06:09 |
| freem | just checking, but I suppose needs_root_rights is set to yes? | 06:11 |
| darwin | no; startx didn't work with that | 06:12 |
| freem | really? weird, I *think* I have the exact inverted behavior... but that install is relatively old, maybe something changed and it's no longer required... | 06:12 |
| rrq | Xorg doesn't need to be run by root, but there are three ground requirements in the running user's setup | 06:15 |
| rrq | firstlym the user needs to have rw access to /dev/ttyN for the VT concerned | 06:15 |
| rrq | secondly the user needs rw access to /dev/dri/card0 | 06:16 |
| rrq | thirdly the user needs input device mediation | 06:16 |
| freem | logind & friends, the input device mediation? | 06:17 |
| rrq | via seatd or logind+dbus | 06:17 |
| freem | thx | 06:17 |
| rrq | many weeks, evne moths and years ago, non-root user needed a different setup via xserver-xorg-legacy ... primaryily that Xorg had SUID=0 | 06:19 |
| freem | well, I still don't run an input device mediator tool, so I still run Xorg as root | 06:20 |
| freem | what are they for, exactly? | 06:20 |
| freem | and don't *they* run as root anyway? | 06:20 |
| rrq | well, it actually runs a "personal seatd" for that anyhow | 06:20 |
| rrq | yes seatd runs with permission to /dev/input/* | 06:20 |
| onefang | Hmm, got seatd AND elogind-daemon running. Wonder if I can get that down to just seatd? | 06:21 |
| rrq | mayby.. elogind isn't a "do one thing well" kind of thing | 06:21 |
| rrq | me spelling is shocking :) | 06:22 |
| freem | I don't think there are still many "do one thing well" tools in what they call modern linux though | 06:23 |
| * onefang wanders into the kitchen to try that on the test box. | 06:24 | |
| freem | I will notably refuse to keep dbus even installed, because that moronic tool will start itself automatically without any way for me to prevent it or not | 06:24 |
| n4dir | i just gave up. sad, but true | 06:25 |
| freem | I guess wayland have the "do one thing" part... sadly it does not have the "well" one :D | 06:25 |
| n4dir | lol. for the lulz i booted Fedora Jam, audio Fedora, and for the life of me couldn't change the keyboard to german in Wayland | 06:26 |
| freem | on this https://linuxfr.org/news/gimp-3-0-rc1-est-sorti (sorry, french link) there's a gimp dev who baically says that all professionals they met still use X11 because apparently wayland's get colours wrong | 06:27 |
| freem | which is kind of a critical issue, for a tool meant to show stuff on screens... | 06:28 |
| freem | linking to this: https://gitlab.freedesktop.org/wayland/wayland-protocols/-/merge_requests/14 | 06:28 |
| freem | I am not an artist, so I can't judge. I also know that bunch of people who like writting their own tools tried with wayland, and went back to writting their X11 WM instead | 06:29 |
| freem | less troubles | 06:29 |
| rrq | other Q: anyone knows whether/how the kernel uses the "chosen/bootargs" attribute on FDT boot? | 06:30 |
| freem | FDT? (which implies: not me, sorry) | 06:30 |
| rrq | "Flattened Device Tree" ... the alternative to ATAGs, a I believe | 06:31 |
| freem | ah, like, U-boot? | 06:31 |
| rrq | u-boot deals with them too, yes, but afaiui the kernel's 3rd argument is either an ATAG table or an FDT blob | 06:32 |
| rrq | (pointer) | 06:33 |
| onefang | OK removing elogind and friends looks like it requires some rabbit hole exploring. No time for that. | 06:33 |
| freem | damn I worked with that stuff a few years ago... but do not remember well that part (despite I had to go through a lot of work to get the mess working, because almost no doc back then) | 06:33 |
| freem | IIRC I used to place a file at the 1st partition with some kind of lilo-style config in it | 06:34 |
| freem | maybe I can boot one of the BBBs I salvaged from destruction and see what I did back then | 06:35 |
| freem | dunno if I have an appropriate power source though | 06:35 |
| * freem probably can guess back the static IP so lack of screen/keyboard should not be a prob | 06:36 | |
| rrq | my headline issue is to boot a stock armhf kernel on arm64 hardware | 06:36 |
| rrq | the u-boot/kernel handover happens, but the kernel just hangs there quietly | 06:37 |
| onefang | #devuan-arm might help. | 06:38 |
| rrq | more people here :) | 06:39 |
| freem | ah, found some old doc I wrote back then | 06:42 |
| freem | stuff I wrote so that others would stop bothering me and actually do their job | 06:43 |
| freem | I'm a bit too tired to read and understand what I wrote years ago correctly, but there might be useful hints in this: https://p.mort.coffee/kJx | 06:46 |
| freem | 5 years already... | 06:47 |
| freem | also there are probably many shortcomings and errors, that was the 1st time I'd set a PXE setup to get automatic deployment | 06:48 |
| freem | only had heard about the possiblity before | 06:48 |
| freem | ouch | 06:49 |
| freem | pxe% upl ./host_srv/nfs/installer/bbb/uEnv.txt -k | 06:50 |
| freem | https://p.mort.coffee/x2o | 06:50 |
| freem | that is the uEnv.txt file I mentioned in this, rrq, it might be of use. I remember I put everything in a damn one-line because U-boot would refuse to work correctly otherwise | 06:50 |
| freem | /nfs/installer/bbb% upl boot.correctbug -k | 06:51 |
| freem | https://p.mort.coffee/dlF | 06:51 |
| freem | that one is readable, but no guarantee that the content strictly matches. It should though. | 06:51 |
| rrq | thanks; I have the u-boot stages in hand, but the kernel doesn't want to play nice | 06:52 |
| freem | you sure it receive the correct parameters? | 06:52 |
| freem | I'm digging in that old repo, but that's really old to me, and it's about armhf systems, too | 06:52 |
| freem | there might be differences | 06:52 |
| freem | (and I should save this stuff somewhere else, too... those scripts were not bad, can always be reused) | 06:53 |
| rrq | parameters should be right, but the armhf kernel is different.. | 06:53 |
| * rrq afk | 06:55 | |
| freem | it's fun to read one's old fully automated install scripts. All the param stuff is even declarative style | 06:55 |
| freem | yay, I even wrote maintenance docs so that a future me, if I would have stated around, would not have too much troubles | 06:59 |
| freem | someone was not too stupid back then | 06:59 |
| freem | there's nothing more of interest for you in there though, I'm afraid, rrq | 07:01 |
| friedhelm | \join #devuan-ci | 19:44 |
| paculino | I just restarted mousepad for the first time since using emacs gtk-key-theme-name, but it doesn't work like in icecat/firefox. Also, an extra titlebar appeared (icecat now lacks the title in the titlebar) and it and the menubar are now a different colour. | 21:11 |
| nemo | Has anyone here gotten tinydns running on devuan? | 21:16 |
| nemo | the init script doesn't seem to actually start anything.. | 21:16 |
| paculino | What init are you using? | 21:25 |
| paculino | Based on an old forum post, it supports runit | 21:25 |
| nemo | hm. apparently "sysvinit" | 21:29 |
| nemo | I assume it's the devuan default | 21:29 |
| nemo | (I installed this VM a long time ago) | 21:29 |
| nemo | paculino: it installed an /etc/init.d/tinydns that looked very different from the gentoo setup | 21:30 |
| nemo | no svscan | 21:30 |
| nemo | #!/usr/bin/env /lib/init/init-d-script | 21:30 |
| paculino | hendrikboom3 on the forums seems to have had it working in January at least | 21:31 |
| nemo | the start/stop seems odd | 21:31 |
| nemo | daemon_ () { /usr/bin/daemon -D "/etc/sv/${NAME}" --noconfig --name "${NAME}" "$@"; } | 21:31 |
| nemo | do_start_cmd_override () { daemon_ -- /usr/bin/chpst -e conf /bin/sh ./run; } | 21:31 |
| nemo | paculino: have a link? would rather not have to rework the init for everything on here | 21:31 |
| nemo | just trying to quickly switch my DNS over to something I have running recently while I unbreak this router by getting openwrt on it | 21:31 |
| nemo | at present !@#$ thing seems to be hard blocking 22 and 53 | 21:32 |
| paculino | There is not a post on fixing it as far as I can tell, but hendrikboom3 mentioned using tinydns to try to troubleshoot another problem, so you'd have to message | 21:32 |
| paculino | https://dev1galaxy.org/profile.php?id=6944 | 21:32 |
| paculino | I'm not familiar with tinydns unfortunately; I barely got wifi working | 21:33 |
| nemo | mm... | 21:37 |
| nemo | might be easier to just roll my own, based on what is working (gentoo) | 21:38 |
| nemo | paculino: tinydns is pretty nice. less to keep track of than bind | 21:38 |
| nemo | might also just be faster to unbreak the router... a project for tonight | 21:38 |
| paculino | A couple friends fixed their server last night just by rebooting everything. The router and modem were at fault, not the new configs as feared. | 21:41 |
| nemo | well this router has been bounced several times during the setup process | 21:41 |
| nemo | and ports 22 and 53 are definitely whitelisted. I think they are just being idiots | 21:42 |
| nemo | I'm going to try ASUS Merlin since it's a more minimalist mod and I'd had it working w/ a config like this in past. And if that fails, openwrt which also supposedly supports this model | 21:42 |
| nemo | paculino: just to get something working I installed svscan. it seems to be running stuff now, I'll figure out the "right" devuan/debian approach later | 21:56 |
| AlexLikeRock | lest HACK ! :-> | 22:06 |
| rwp | paculino, I don't think the gtk-key-theme-name change should change anything you can see as that is controlled by other themes. To test this though you could remove that file and restart the apps to verify. I could be wrong. | 22:11 |
| paculino | My gtk settings are hacky in order to get two rows of windows on the bar, so I fear it is an interaction there. | 22:12 |
| paculino | I've always had failure (for gtk apps, not fltk) to load both colorreload-gtk-module (maybe the cause of the different colour) and window-decorations-gtk-module (probably related to the titlebar issues) | 22:13 |
| rwp | I mainly do everything in terminals. For me the only GUI app is Firefox. I use that config to have Firefox use Emacs keybindings. And as far as I can see there are no changes to be seen visually. But I do no other customization. Sorry I don't know more. | 22:28 |
| paculino | Okay, thank you | 22:30 |
| gnarface | same here, i use that one line to get emacs style bindings in various gtk programs (where the defaults are otherwise vi or windows style ones) and it works everywhere i've tested except that various programs skip some of the bindings, and not the same ones, so there's something program-specific overriding just key things in different cases... | 22:33 |
| gnarface | ...but i've never once seen it cause an actual visual change in anything | 22:33 |
| gnarface | (and the ones that have windows style bindings too usually keep them) | 22:33 |
| paculino | Tonight I'll work on it more; thank you for showing the line to add. My problem is not a big deal at all. | 22:35 |
| paculino | If curious, here is what it looks like: https://xabber.de/upload/40a61b2ffd4899329d64361467c8a6ff8ebc174a/BG0doqF0FKW5FsHcbNPJq1kfFOAtzrwYrfna8eH5/076aad5b-a946-40d4-848c-fc0f197de1ee.png | 22:36 |
| AlexLikeRock | looks so FLAT | 22:36 |
| AlexLikeRock | like Window$ 8 | 22:37 |
| AlexLikeRock | XDDDD | 22:37 |
| AlexLikeRock | UGLY !!! | 22:37 |
| paculino | Windows 8 was my favorite Windows in terms of appearance (although, to be fair, I had something to allow Windows 7 style widgets) | 22:38 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
