e3d3 | What is the advised way to disable sysv startup applications; sysv-rc-conf, update-rc.d or ? | 01:24 |
---|---|---|
fsmithred | yes | 01:28 |
fsmithred | either way | 01:28 |
e3d3 | okay, thanks. After 2 hours reading about it, I found the same MX Linux advise 1 second after asking here. Good to have it confirmed. Thanks. | 01:31 |
rwp | e3d3, There are multiple opinions available on the matter. I know of at least four different methods and I use at least two of them myself. | 01:57 |
rwp | Better advice might be had if we knew what daemon you wanted to have installed but disabled. Since that by itself is an odd combination. | 01:57 |
e3d3 | rwp: I guess those 2 are not the ones I mentioned above | 01:57 |
e3d3 | rwp: I want mdadm, saned and bluetoothd disabled | 01:58 |
rwp | For mdadm it would probably make more sense to remove/purge it from the system. | 01:58 |
rwp | For saned and bluetoothd are those pulled in by some DE dependencies? | 01:59 |
e3d3 | I guess so but am not sure. I have XFCE, and both are not enabled as startup applications (GUI list) | 02:00 |
rwp | For avahi that is Depended upon but I want disabled I put an "exit 0" in /etc/default/avahi-daemon to prevent it from running. | 02:00 |
rwp | Putting an "exit 0" into any of the /etc/default scripts completely disables the entire init script from all actions. | 02:01 |
e3d3 | I have not looked into that dir, yet/never. Is the reason that you advise removing mdadm because its not in the /etc/default dir ? | 02:05 |
e3d3 | And is it true that via init.d disabled services are started again when updating the dependencies ? | 02:05 |
rwp | I always say simpler is better. Reducing the install base is always simpler. | 02:05 |
e3d3 | I agree completely | 02:05 |
rwp | My XFCE install has sane-utils and xsane installed but not saned. And has libbluetooth3 installed but not bluetoothd. | 02:06 |
e3d3 | I'll remember that, thanks | 02:10 |
rwp | Back in the old days users of Red Hat systems installed from cdrom often had the problem of DLL version Hell problems when trying to install something later. | 02:11 |
rwp | So the user culture became one that when they had the cdrom in their hand then install 100% of everything that was on it all at once. | 02:11 |
rwp | However now in Devuan with network installation if one has networking then it is trivial to install anything at any time. | 02:11 |
rwp | Because that was the only time when everything matched. But that required installing lots of cruft that was never used. | 02:11 |
e3d3 | I use mostly the shell & Emacs, most other things I consider distraction. Long ago I used minimal or netinstall but I don't have the energy anymore to keep reading about everchanging technical details, just to configure some 'secundairy' stuff, so I use XFCE desktop some years now | 02:16 |
e3d3 | Last 2-3 years I booted once in 1-3 months, then kill 2 services because I was too lazy to read about the init system. But because I installed Devuan on a laptop that I want to shut down at least once a day, I better disable unwanted services instead of killing. | 02:22 |
e3d3 | Do you know the answer on my other question; that upgrading a dependency will start a disabled service again ? | 02:23 |
rwp | Sorry, I didn't see that particular question. Upgrading _should not_ re-enable a disabled service. | 02:26 |
rwp | If the service is disabled by modifying /etc/default/foo then that is a conffile and changes there are required to be preserved, unless you say 'Y'es install maintainers version at package installation time. | 02:26 |
rwp | If a service is disabled by modfifying the /etc/rc?.d/S*foo symlinks then must leave at least one symlink behind to indicate that it is installed and configured. | 02:27 |
rwp | See https://paste.debian.net/1180399/ for the particular paragraph that describes this. | 02:27 |
rwp | I have often seen people remove all symlinks and then complain that they were re-installed so that is a common error. | 02:28 |
rwp | Leave at least one /etc/rc?.d/K*foo symlink somewhere so as to indicate that it has been installed but has been locally customized. | 02:28 |
rwp | Does that cover things for you? Questions? | 02:29 |
e3d3 | sorry but I need to read above more than once to understand it, so my reaction need some time | 02:29 |
rwp | No problem! I am also doing accounting simultaneously so might be slow to notice and respond myself. | 02:29 |
e3d3 | I think I understand what you said; upgrading foo will reinstall a removed symlink foo, so I better modifying the setting. | 02:36 |
rwp | e3d3, It will only reinstall a symlink if *ALL* of the symlinks are removed. | 02:37 |
e3d3 | you mean for all runlevels ? | 02:37 |
rwp | It will not modify symlinks if they are locally modified, removed, changes, as long as at least one symlink remains. | 02:38 |
rwp | Right. For all runlevels. find /etc/rc?.d/ -name '*foo' | 02:38 |
rwp | If all of those are removed then it appears as if it is a first time pristine installation and therefore the postinst will install symlinks. | 02:38 |
rwp | But if at least one remains to be found then it won't touch them if they are different from the defaults. | 02:39 |
e3d3 | so upgrading foo checks the symlinks in all runlevels | 02:40 |
rwp | Yes. | 02:40 |
e3d3 | I'm now in runlevel 2 but I guess I should disable bluetoothd and saned in all runlevels, not ? | 02:41 |
rwp | The difference between an "exit 0" in /etc/default/foo is that disables all init script actions. | 02:41 |
rwp | Removing /etc/rc[2345].d/S??foo avoids starting it but it will still be killed if manually started. | 02:41 |
rwp | Subtle difference but still a small difference. | 02:42 |
rwp | Have you ever used a different runlevel than 2? | 02:42 |
rwp | And I mean "ever" in all of the years you have been using the system? | 02:42 |
rwp | I consider myself a rather old-school user of things and even I can't remember the last time I used a different runlevel for any reason. (other than shutdown) | 02:43 |
e3d3 | I normally don't check my runlevel, but "know" that different distro/OS can use different runlevels, and I also don't know much about init systems at all | 02:43 |
rwp | (shutdown passes through runlevel 6 internally) | 02:43 |
e3d3 | 0,1 and 6 are told to be common in Linux | 02:44 |
rwp | So... Imagine yourself sitting in front of a university shared server for a hundred all trying to compile their programs to get them done before deadline. | 02:44 |
e3d3 | I do | 02:44 |
rwp | That's the environment in which runlevels were imagined. Runlevel 3 used to be networking and NFS enabled. Runlevel 0 (or was it 1) was single user mode. | 02:45 |
rwp | So one could switch to single user mode to keep anyone from being able to log into the system so that you could do some system administrative task. | 02:45 |
rwp | Like applying security patches. Or recovering from someone who had actually compromised the system. Or ran it out of disk space. Or whatever. | 02:45 |
rwp | Then could switch runlevels back to runlevel 3 with networking on and back online without rebooting. | 02:46 |
rwp | Because rebooting might take a while to flip those front panel switches to load in the boot loader IPL code to be able to load the next segment off mag tape. | 02:46 |
rwp | These days... Not needed so much. (shrug) | 02:46 |
e3d3 | I am the only physical user but see many users active on my laptop, and have no clue what the do when I'm not looking, maybe they cross runlevels !? | 02:48 |
e3d3 | nor if multi-user runlevel is required because of root & I | 02:49 |
rwp | It's your physical laptop but you see other users using what command? Try running the "w" command. Or "who". | 02:49 |
e3d3 | e3d3 :0.0 2021-01-09 18:38 (:0.0) | 02:50 |
e3d3 | 02:50:16 up 8:39, 1 user, load average: 0.13, 0.16, 0.11 | 02:50 |
e3d3 | USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT | 02:50 |
e3d3 | e3d3 :0.0 :0.0 18:38 ?xdm? 55:56 0.03s /bin/sh /etc/xdg/xfce4/xinitrc -- /etc/X11/xinit/xserverrc | 02:50 |
rwp | So you have one user and it is you. | 02:50 |
e3d3 | ??? | 02:50 |
e3d3 | I going to check it on my other primairy distro, 1 moment | 02:51 |
rwp | And it is the xdm (X Display Manager) login starting X graphics for you. Upon which you might have many terminals. | 02:51 |
e3d3 | I'm confused about that I the only user, or why I thought different | 02:52 |
fsmithred | 'ps aux' will reveal some system users | 02:53 |
rwp | A long ps listing (ps -efHww or ps aux) would show many processes running users but those are system users not people users. | 02:53 |
rwp | System users are used to be non-root and compartmentalized. For security reasons. | 02:53 |
fsmithred | ^^^ that | 02:53 |
rwp | So for example lightdm is run as the lightdm user. And bind/unbound is run as a bind/unbound user name. | 02:54 |
e3d3 | I thought indeed because of 'top/htop' listings, not that root, avahi, statd etc where real people, but still real users | 02:54 |
rwp | And www-data runs the web server. And postfix runs the postfix mailer. | 02:54 |
fsmithred | they are users with limited powers | 02:55 |
rwp | Well... They are different users. They are non-root users. That's the most important thing. And they are different compartments from each other. So they can't get out of their boxes. | 02:55 |
e3d3 | I believe you, mostly ;) about this | 02:56 |
e3d3 | so I, e3d3 and boxed companions will normally only stay in runlevel 2 ? And Lightdm only in 0 ? | 02:58 |
fsmithred | lightdm usually starts in 2 | 02:58 |
fsmithred | and 2-5 are the same in debian/devuan | 02:58 |
rwp | Type in "who -r" and it will say 2 | 02:58 |
rwp | Oh, there is also a "runlevel" command too. Same thing. | 02:59 |
e3d3 | that is my runlevel | 02:59 |
fsmithred | the default runlevel is set in /etc/inittab | 02:59 |
rwp | I recommend basically doing nothing interesting and just leaving everything at the default runlevel 2. Don't worry. Be happy. | 03:00 |
e3d3 | I've read that file to find out which runlevels Devuan uses | 03:00 |
e3d3 | :) | 03:00 |
e3d3 | I wish the same for you | 03:00 |
rwp | On Red Hat systems you will find it being runlevel 5 for everything. But same advice. Just don't touch it. Don't worry. Be happy. | 03:00 |
rwp | The only reason to even be aware of it is if you want to not start a daemon then you would need to know that to remove the /etc/rc2.d/S??foo link. | 03:01 |
e3d3 | I asked because with TUI app 'sysv-rc-conf' I can/have to set all runlevels, with rcconf only the current runlevel, if I believe the screenshots | 03:02 |
rwp | fsmithred, Do you recall the last time you changed to single user mode on a system? I can't. | 03:02 |
fsmithred | about 15 minutes ago | 03:03 |
fsmithred | and several times a day I do 'init 1' | 03:03 |
fsmithred | but that's because I'm testing stuff in a VM and don't want to do a full reboot | 03:03 |
rwp | Ha! "Few things are harder to put up with than the annoyance of a good example." --Mark Twain | 03:03 |
e3d3 | :) | 03:04 |
fsmithred | and I configure Refracta so that the display manager does not run in runlevel 3 | 03:04 |
fsmithred | that way it's possible to boot to console | 03:04 |
e3d3 | display manager in runlevel 3, not 2, like rwp said, if I understand it right | 03:05 |
fsmithred | default setting is for dm to run in 2-5 | 03:06 |
e3d3 | the multi-user spectrum | 03:06 |
fsmithred | yeah, but debian makes them all the same | 03:06 |
fsmithred | redhat and suse have graphical only on 4 and 5 | 03:07 |
rwp | One of these days I am going to get https://github.com/cosmos72/twin/ up and running for the main text console. ;-) | 03:07 |
fsmithred | 2 and 3 are multi-user console only | 03:07 |
e3d3 | rwp: I guess you don't use Emacs ;) | 03:08 |
rwp | I am in Emacs typing this right now using Emacs-erc. :-) | 03:08 |
e3d3 | rcirc | 03:08 |
tuxd3v | 3 is multiuser + network | 03:09 |
rwp | tuxd3v, That's the classic BSD convention, yes. But not on Devuan or Debian. | 03:15 |
e3d3 | because something went wrong, I repeat my last thanks for all the friendly help again and wish you all a good day|night. | 03:18 |
tuxd3v | rwp, yes :) | 03:22 |
systemdlete | I installed beowulf on bare hardware on two boxes. The first one works OK, but the other doesn't. The desktop comes up with low resolution and doesn't give me options to go higher resolutions. It's the same machine I am running ascii on, which works fine. So I am wondering what I need to do to get the desktop working properly. | 04:32 |
systemdlete | video is RS780L [Radeon 3000] [1002:9616] | 04:32 |
systemdlete | Does this require non-free software? (I did install that though) | 04:32 |
systemdlete | I'm guessing I need a certain x11 or xorg driver package or the like | 04:32 |
systemdlete | Oh... | 04:36 |
systemdlete | just found a wiki page | 04:36 |
systemdlete | (wiki.debian.org/atihowto | 04:44 |
systemdlete | When running (on ascii) grub-mkconfig, I get error: "/usr/sbin/grub-probe: warning: Couldn't find physical volume `(null)'. Some modules may be missing from core image.." This occurs multiple times. | 04:58 |
systemdlete | ddg'd this, but I can't seem to get a clear explanation of what this is. | 04:58 |
systemdlete | (doesn't hint which disk object it is complaining about) | 04:59 |
gnarface | systemdlete: you probably do need firmware-amd-graphics from non-free, yea | 05:10 |
systemdlete | right, gnarface. That wiki page spells it all out. | 05:10 |
systemdlete | Sorry to bug you about video drivers again. | 05:10 |
gnarface | dunno what that grub error means exactly, but it kinda seems like it's looking for something it sees in the config but not present on the machine | 05:11 |
systemdlete | "in the config" | 05:11 |
systemdlete | /etc/grub/* ? | 05:11 |
gnarface | the auto-generated config in /etc/grub.d | 05:11 |
systemdlete | ok | 05:11 |
gnarface | there's also /etc/defaults/grub | 05:12 |
gnarface | for hand edits | 05:12 |
gnarface | sorry, /etc/default/grub | 05:12 |
systemdlete | well, there is almost nothing in /etc/default/grub | 05:13 |
systemdlete | other than defaults, comments | 05:13 |
gnarface | there is also /boot/grub, i think | 05:13 |
gnarface | unless that's from the old version... | 05:13 |
gnarface | the thing is it's mostly supposed to be auto-generated | 05:14 |
systemdlete | well /boot/grub/grub.cfg is what I am trying to generate with grub-mkconfig | 05:14 |
gnarface | and you're only supposed to change some variables in /etc/default/grub | 05:14 |
gnarface | but obviously stuff goes wrong sometimes... | 05:14 |
gnarface | was this an upgraded system or a fresh install? | 05:14 |
gnarface | and is there like.... some weird removable drive bays in it or something? | 05:14 |
systemdlete | I was trying to coerce grub-mkconfig on ascii to "pick up" the new install on the beowulf partition | 05:14 |
systemdlete | no removable bays -- removed that unit months ago | 05:15 |
gnarface | i think i usually just run update-grub as root | 05:15 |
systemdlete | But that was never a problem in the past. It's a Kingwin 4000 that just uses passive connections. | 05:15 |
systemdlete | AH! | 05:15 |
systemdlete | thank you -- let me try update-grub | 05:15 |
gnarface | the thing you gotta figure out is why it's trying to load "(null)" ... that seems like it's literally missing a variable for a drive name but it's got a boot entry for it anyway or something? | 05:16 |
systemdlete | dang. Same error messages | 05:16 |
gnarface | there might be other errors somewhere in /var/log | 05:16 |
gnarface | or in dmesg | 05:16 |
systemdlete | I did have a usb drive plugged in, but even once it is removed, I still get errors from grub-mkconfig (or grub-update) | 05:16 |
systemdlete | I looked at system logs. Nothing of interest | 05:17 |
gnarface | weird | 05:17 |
gnarface | i got nothing | 05:17 |
systemdlete | Wonder if it is time to use Microsoft Tech: reboot | 05:17 |
gnarface | maybe missing package is what i'm thinking | 05:18 |
gnarface | but i couldn't guess what | 05:18 |
gnarface | you could always just use lilo instead | 05:19 |
systemdlete | Well, could be I guess. But why has this not been a problem until now? I rarely do this. I haven't installed a new OS on this box since ascii, and that was a couple of years ago. | 05:19 |
gnarface | yea, i'm really not sure. i can't recall seeing this before | 05:19 |
gnarface | i keep going back to it getting incomplete info it's trying to integrate into the config from somewhere | 05:19 |
gnarface | but where? | 05:19 |
systemdlete | it is not too important. The objective here is to move on to beowulf. So as long as I can get beowulf to boot in the other partition (along with those video drivers working) I'm good. | 05:19 |
gnarface | the most likely culprit would be leftover configs from after an upgrade, but still... never seen this | 05:20 |
systemdlete | It would be helpful if *nix utilities would tell the user what specific things they are complaining about. | 05:20 |
gnarface | did you have grub installed to multiple disks/partitions? maybe it's the grub config on some other block devices that it's picking up? | 05:20 |
systemdlete | Not sure. | 05:21 |
systemdlete | I have grub installed to /dev/sd[abc] (it's a RAID1 2 drives + 1 spare. | 05:23 |
systemdlete | both ascii and the proposed beowulf partitions are on partitions on those drives | 05:23 |
systemdlete | Fairly simple arrangement I think. | 05:23 |
systemdlete | And those partitions are NOT with lvm or anything fancy. | 05:24 |
systemdlete | ext4 and btrfs | 05:24 |
systemdlete | I hope that ascii's grub-mkconfig can read btrfs file systems, but idk | 05:24 |
gnarface | hmmm, also not using btrfs here | 05:31 |
gnarface | got any empty partitions? | 05:31 |
gnarface | unformatted? | 05:31 |
gnarface | unassigned? | 05:31 |
gnarface | i'm really grasping at straws | 05:32 |
gnarface | when grub doesn't work for me if i can't find a fix readily i often try lilo | 05:32 |
gnarface | i used to use it before, so i'm more comfortable with it anyway | 05:32 |
systemdlete | The only thing I note that might be relevant is that, while installing beowulf in another partition, I requested to install grub to /dev/sda, b, and c, yet... | 05:45 |
systemdlete | when I boot the box, I'm still getting the OLD grub menu. Now, I think I copied grub.cfg over from the ascii partition. But not sure if that has anything to do with the grub-mkconfig issue on ascii. | 05:46 |
systemdlete | heheheh. Yeah, when all else fails, fall back to simpler things. I agree. | 05:46 |
systemdlete | I was a huge fan of LILO until I ran into distros that refused to work with LILO so I was forced to use grub, then grub2 | 05:47 |
qin | is there a better solution for getting vaapi stuff working in beowulf? currently am using a symlink for iris_drv_video.so | 07:51 |
gnarface | what video hardware, qin? | 09:44 |
gnarface | i didn't have to make a symlink with nvidia or amd | 09:45 |
gnarface | intel hardware that doesn't support vaapi should be able to use the libvdpau-va-gl1:i386 wrapper lib | 09:46 |
gnarface | sorry libvdpau-va-gl1 (not specifically i386 necessarily) | 09:47 |
gnarface | not sure about other issues but in any case the situation is usually specific to the video card driver | 09:47 |
gnarface | and it's supposed to just be a matter of having the right packages installed, though mplayer will also need command-line options | 09:48 |
qin | gnarface: im using intel here, and thanks for the info | 09:56 |
systemdlete | Still struggling a bit to get beowulf configured satisfactorily. The main hangup at this point is that os-prober on beowulf is not detecting the other OS's installed. I was able to copy my ascii's grub.cfg to beowulf, and that works (except it doesn't have beowulf) but I'll still have to do some manual editing. | 10:16 |
systemdlete | I DDG'd a bit for anything related to os-prober in the past year (because wider time searches returned a lot of probably outdated info). | 10:17 |
systemdlete | At any rate, none of what turned up gave me any definitive answers. os-prober, run directly from the command prompt, returns just the kernels on the local (beowulf) partition. | 10:18 |
systemdlete | I tried cross-mounting the ascii partition but no better. | 10:18 |
systemdlete | I don't recall this much drama with grub.cfg generation in the past, but maybe I have overlooked a step. | 10:21 |
fsmithred | systemdlete, os-prober won't find encrypted systems. If that's the case, create entries in /etc/grub.d/40_custom | 13:15 |
bru | Hello, I need to set the memlock limit on my system with sysvinit, what's the best way to do so? | 16:25 |
bru | Nevermind, I chose to add a ulimit line on my program's init file | 16:40 |
tarzeau | §br | 17:31 |
tarzeau | bru: and that worked? i like nohang a lot | 17:31 |
bru | tarzeau: Still haven't tried it yet, lemme see with a quick reboot | 17:45 |
bru | tarzeau: Yep, adding "ulimit -l 65535" (as in 65535 KiB) before the running command on the start function (mpd_start in my case) worked perfectly | 17:53 |
tarzeau | how much memory do you have? | 17:56 |
bru | 16GB plus 8GB swap. The setting is just for a single program anyways, not a global rule | 17:57 |
tarzeau | yeah i just wonder how you exhaust that much memory | 17:57 |
tarzeau | i'm a fan of zram btw | 17:57 |
bru | It wasn't my idea to set that much memlock, it's a recommendation from the mpd manual: https://www.musicpd.org/doc/html/user.html#error-failed-to-initialize-io-uring | 17:59 |
bru | Hmm, I haven't heard of zram, is it specially for managing RAM across processes? | 17:59 |
tarzeau | ~ | 18:27 |
tarzeau | .~. | 18:27 |
tarzeau | bru: it's for compressed memory | 18:27 |
brandflake11 | Hey hey, does anyone have any good reading on using sysvinit? | 19:59 |
ErRandir | Start with /etc/init.d/README | 20:23 |
Xenguy | brandflake11: sysv-rc-conf seems to be a handy tool also | 21:48 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!