rennj | Bcache patches for the Linux kernel allow one to use SSDs to cache other block devices. It's analogous to L2Arc for ZFS, but Bcache also does writeback caching (besides just write through caching), and it's filesystem agnostic. | 00:01 |
---|---|---|
rennj | nice | 00:01 |
yanmaani | buZz: Is there any package that does it for me? | 00:01 |
buZz | dont think so | 00:02 |
buZz | rennj: they arent patches anymore | 00:02 |
buZz | been mainline since ~forever | 00:02 |
rennj | yeah well ive i was doing lvm/ some fs bcache would be nice...or you know just go openzfs | 00:03 |
rennj | https://bcachefs.org/ the future | 00:05 |
rennj | Bcachefs is not yet upstream - you'll have to build a kernel to use it. | 00:06 |
rennj | meh..let buZz test it | 00:06 |
rennj | hehe | 00:06 |
buZz | yeah i didnt try bcachefs yet | 00:06 |
rennj | copy that sun tech | 00:06 |
rennj | reinvent the wheel | 00:06 |
buZz | hehehe | 00:07 |
buZz | quite sure bcache is older than zfs | 00:07 |
DHE | i'd like to point out that people give zfs' l2arc and slog features far too much credit | 00:07 |
buZz | but who knows | 00:07 |
buZz | i remember when LVM was new, and ppl were falling over each other to adopt it | 00:08 |
buZz | leading to a lot of dataloss | 00:08 |
rennj | https://blogs.oracle.com/bonwick/zfs-end-to-end-data-integrity | 00:08 |
rennj | your merkel tree | 00:08 |
rennj | linux lvm is from hp-ux lvm which is like ibm lvm | 00:09 |
rennj | pull 100drives from array cabinets once cause of bad bios | 00:12 |
buZz | causation? | 00:12 |
buZz | or correlation | 00:12 |
buZz | they arent often both | 00:13 |
buZz | LVM on linux in any way has no interest in whatever firmware you run | 00:13 |
rennj | lvm is just the volume manager..you have choice on fs | 00:13 |
rennj | lvm and veritas vxvm where almost identical | 00:13 |
buZz | i dno, i dont enterprise | 00:14 |
rennj | its just 1 part of the raid pieces..but with zfs you get in all in 1 | 00:14 |
rennj | the volume manager and fs..and crc integrity | 00:14 |
buZz | for me , all-in-one solutions just break tiny parts which then pulls everything off | 00:14 |
rennj | anyway im sure linux copy of that tech will eventually be production ready | 00:14 |
buZz | i dont even like phones with cameras | 00:15 |
buZz | i like cameras | 00:15 |
buZz | modern phone takes like 1 minute to boot, my camera boots in 1-2 seconds | 00:15 |
rennj | read those car camera with embedded linux boot in ms | 00:18 |
rennj | like backup rear view display | 00:18 |
rennj | https://www.yoctoproject.org/ spin of devuan version | 00:20 |
rennj | probably totally against the spirit of devuan | 00:22 |
rennj | include the binary no source in the final build | 00:22 |
rennj | i only use my phone to make phone calls, with bmc being blackbox..it worthless for computing | 00:25 |
golinux | Maybe OT would be a better place for car/phone chat? | 00:27 |
buZz | rennj: yeah one of my point&shoots is a gopro clone | 00:30 |
buZz | thats a full linux computer | 00:30 |
buZz | either way, i use bcache cause i can sprinkle it onto any fs | 00:31 |
buZz | similar to how zram can work for 'any system' | 00:31 |
buZz | i have no doubt that camera uses zram aswell | 00:31 |
buZz | should check some day i guess :P | 00:31 |
* wikan is watching you | 03:20 | |
mason | Ugh, unbound is exceedingly ill-behaved. | 03:59 |
fsmithred | what is it doing? | 04:00 |
mason | It doesn't write a pidfile on launch. I wonder if the sysvinit scripting has rotted. | 04:00 |
mason | Doesn't write a pidfile on launch, launches competing copies on restart, won't die because of the pidfile issue. | 04:03 |
fsmithred | I seem to recall gnu_srs finding something like that in the eudev init script | 04:03 |
fsmithred | that sounds vaguely familiar | 04:03 |
mason | I'm wondering... It appears maybe (still need to dig/verify) to have an implicit assumption that resolvconf exists. | 04:04 |
fsmithred | I'm trying to find the change that was made (or will be made, I guess) | 04:05 |
mason | It's also getting some apparmour denials that I'm trying to understand. | 04:06 |
mason | start-stop-daemon says (in its man page) that it won't honor non-root-owned pidfiles, and the unbound pidfile is owned by the unbound user. It's empty and hence won't work, but now it has two reasons why it shouldn't work. | 04:11 |
mason | Hm, changing it to root, it gets changed back and now the daemon stop and starts. | 04:12 |
mason | Weird. | 04:12 |
mason | Runnig the daemon, /run/unbound.pid changes back to unbound:unbound, which won't work with start-stop-daemon. | 04:14 |
fsmithred | mine is not empty and is owned by root | 04:17 |
mason | Is this after a fresh start? | 04:17 |
fsmithred | I assume it started last reboot | 04:17 |
fsmithred | days ago | 04:17 |
fsmithred | yeah, it's empty after a restart | 04:19 |
mason | and unbound:unbound? | 04:19 |
mason | And do you have multiple unbound daemons running after restart? | 04:19 |
mason | Or maybe after a subsequent restart? | 04:19 |
fsmithred | yeah, multiple processes | 04:21 |
fsmithred | and they're still running after attempting to stop the service | 04:21 |
mason | Yeah. man start-stop-daemon and the ownership will jump out as a problem. | 04:22 |
fsmithred | delete the pidfile and I can start it again and get a non-empty pidfile | 04:34 |
mason | Right. But only if it's root:root, and then it'll change to unbound:unbound. | 04:36 |
mason | I haven't quite figured out what's going on with it yet. | 04:36 |
fsmithred | shit, I just shut it down | 04:36 |
gnarface | does unbound have a setting? | 04:39 |
gnarface | i note that apparently all my pid files are owned by root | 04:40 |
gnarface | i assume for a good reason | 04:40 |
gnarface | the daemon probably shouldn't own it's pid file for security reasons, i think | 04:41 |
gnarface | like for the same reason they typically don't own their log files either | 04:42 |
mason | gnarface: Right. This is discussed in start-stop-daemon. | 04:42 |
mason | In particular, if a daemon owns its own pidfile, then if it's compromised, it can rewrite that and cause any other process to be targeted instead. | 04:42 |
gnarface | so i would further presume that unbound would have to be pretty silly not to include a way to specify ownership and permissions of the pid file | 04:43 |
mason | My suspicion is that they've called it in for sysvinit and run it as a simple service under systemd. | 04:43 |
mason | gnarface: I think the answer is that it's pretty silly. It needs to write a pidfile before it drops root, and there's no obvious knob to control that. | 04:44 |
gnarface | though, i guess it may not be strictly not necessary for it to write it's own pid file to begin with. you could do it with bash right? | 04:44 |
gnarface | its* | 04:44 |
mason | Well. The thing launching it will get a pidfile back. Anything involving a search is decidedly wearing the big mouse ears. | 04:45 |
fsmithred | sleep time here. Let me know if you figure it out. | 04:46 |
mason | kk, g'night | 04:46 |
mason | hrm: https://bpa.st/W5UQ | 05:00 |
mason | so it really wants to chown it | 05:00 |
mason | Oh, gods, apt build-dep unbound is getting into a fight with itself over libsystemd0 | 05:04 |
mason | This is all gross. :( | 05:04 |
mason | I think having had a look at the source, I'm going to stop using unbound. | 05:06 |
mason | But, for now, testing a quick fix. | 05:06 |
rennj | heh | 05:06 |
mason | Probably be worth my while to learn more dnsmasq. | 05:08 |
mason | Or just go straight back to BIND for this. | 05:08 |
rennj | dnsmasq has the bonus dhcp and tftp built vs bind. | 05:10 |
rennj | pxe/ipxe booting helper | 05:10 |
mason | Yeah, that comes up at work a fair amount. | 05:10 |
rennj | course 1 busybox binary provide all that also | 05:12 |
rennj | along with sshd ftpd httpd telnetd and such | 05:13 |
mason | busybox is useful | 05:13 |
rennj | why android has the toybox is it | 05:13 |
mason | fsmithred, gnarface: So, I installed build deps (will undo that horror soon) and yanked out the ifdef HAVE_CHOWN block shown in https://bpa.st/W5UQ and rebuild, installed my new versions of unbound, unbound-anchor, libunbound8, and now root continues to own /run/unbound.pid, which means it can start and stop normally without failing and tripping over itself in service-disrupting ways. I didn't see a | 05:18 |
mason | configure knob for this. Is this something where we want to fork and ship our own? I'll be testing the patch I've just described until I ditch unbound entirely, as I depend on it for daily work. | 05:18 |
mason | s/rebuild/rebuilt/ | 05:18 |
golinux | mason: Perhaps encourage upstream to provide the missing cog? | 05:22 |
golinux | unbound on jessie has worked wonderfully well. | 05:23 |
golinux | Hoping it will remain that way. | 05:23 |
mason | golinux: Here's the problem with that. This would clearly be broken on Debian without systemd. Hard to imagine getting the Debian package maintainer to care. | 05:24 |
mason | Maybe they'd care. Dunno. But the conflict is with the sysvinit scaffolding. | 05:24 |
mason | The problem is in the source, but the conflict is there. | 05:24 |
golinux | Maybe they are unaware that they broke it. Couldn't hurt to bring it to their attention. | 05:27 |
golinux | If they refuse we can add it to the list of "non-cooperation" packages. Someday we should publish the list to big the magnitude of the attitude to light (if there is anyone who still cares). | 05:29 |
golinux | to big > to bring | 05:29 |
mason | Alright, I guess that's the reasonable direction, yeah. | 05:29 |
golinux | Have a documentary trail could be useful at some point. | 05:30 |
rennj | they broke their own /etc/rc.d/init.d/unbound | 05:30 |
rennj | works on the bsd's | 05:30 |
golinux | Well, double team them! | 05:30 |
mason | rennj: It's chowning the pidfile that's the issue though. The init script is unremarkable. | 05:30 |
rennj | so it wants to run as a user/group | 05:31 |
mason | rennj: Right, but it's chowning its pidfile before dropping root. | 05:31 |
golinux | Better yet, provide a patch for them | 05:31 |
mason | I'd need to read more of the code to make sure there's no case where they legitimately need to chown the pidfile in a chroot. I can't imagine such a case, but that'd be necessary legwork. | 05:33 |
mason | I'll see if I can find an upstream to open a bug asking why they want to chown it, anyway. That'll be a good way to test the water to see if it's worth pursuing. | 05:34 |
rennj | heh fbsd manual $ kill `cat/etc/unbound/unbound.pid` | 05:36 |
mason | rennj: That's actually the upstream manual. | 05:36 |
rennj | it under /var/run on linux i assume | 05:37 |
mason | Configurable. They even have examples with it in /etc/unbound | 05:37 |
rennj | damn its in base fbsd,open | 05:39 |
mason | fsmithred, gnarface, rennj, golinux: https://github.com/NLnetLabs/unbound/issues/303 | 05:43 |
mason | rennj: The unbound in base on FreeBSD is weird. It barfs on my standard config, and I end up having to use the one from ports instead. | 05:44 |
rennj | https://bazaar.launchpad.net/~sdeziel/apparmor-profiles/unbound-chown/files/169 | 05:44 |
rennj | did you say apparmor before | 05:44 |
mason | Yeah, it wants unlimited privs. Reading. | 05:44 |
mason | haha, nice | 05:45 |
mason | That's one way to do it. | 05:45 |
mason | golinux: So, if you look at, say, https://bazaar.launchpad.net/~sdeziel/apparmor-profiles/unbound-chown/revision/169/ubuntu/16.10/usr.sbin.unbound you can see how Canonical fixes it. They just force the chown to fail with AppArmor. :P | 05:45 |
mason | rennj: Nice catch. | 05:45 |
mason | Makes me wonder why I'm not seeing that same behaviour from Beowulf. | 05:46 |
rennj | and its old | 05:47 |
rennj | this problem | 05:47 |
golinux | Some good sleuthing there. | 05:48 |
mason | rennj: Weird. Beowulf, /etc/apparmor.d/usr.sbin/unbound, we explicitly have apparmor *allow* chown. | 05:48 |
golinux | All I know is that it seems to work without issue on jessie. | 05:48 |
rennj | https://bugs.launchpad.net/ubuntu/+source/unbound/+bug/1749931 i was reading prior to that | 05:49 |
* golinux doesn't understand the finer points of unbound. | 05:49 | |
golinux | Sometimes knowing too much gets you into trouble . . . | 05:50 |
rennj | Simon Déziel (sdeziel) wrote on 2018-02-23:"deny capability chown" was initially added for the PID file, see [1]. Failing to chown the PID or the control socket is only logged at higher log level specifically to not generate noise when the chown capability isn't available, see [2,3]. | 05:50 |
mason | golinux: The thing to read here is "man start-stop-daemon" and in particular the paragraph describing the "-p, --pidfile pid-file" option | 05:51 |
mason | rennj: sdeziel is a good guy. | 05:51 |
mason | He's helpful. | 05:51 |
mason | Anyway, for tonight I've got patched packages running and they run properly, so good enough for now. We'll see what comes back from the bug I filed. | 05:52 |
mason | I'm soon to become a pumpkin, so, g'night all. | 05:53 |
golinux | Thank you for doing that! If sdeziel is a good guy, maybe there is hope . . . | 05:55 |
rennj | the cynic! | 06:50 |
golinux | Lived too long not to be | 06:52 |
brocashelm | nice | 06:52 |
GNUmoon2 | Hi, I've installed Beowulf on to a i7 10510 with thunderbolt support (Titan Ridge, JHL7540) and trying to get it to work with a dell wd19tb thuderbolt dock. USB peripherals are not work, no external DP screens, no internet via dock. | 07:40 |
GNUmoon2 | I've updated the kernal to backports 5.7. | 07:40 |
GNUmoon2 | add misc-firmware (yuke) just to get stuff working. | 07:40 |
GNUmoon2 | I've activated the dock via bolt adm tool. The logs show the dock being connected with a bunch of peripherals. | 07:42 |
GNUmoon2 | Just not sure what else I need to get it to work. | 07:42 |
GNUmoon2 | I used the original installed Pop!OS and that detected two external screens once the dock was authorised. | 07:43 |
GNUmoon2 | typo *USB peripherals are working. | 07:45 |
GNUmoon2 | Possibly relevant error in logs: "Broken atomic modeset userspace detected, disabling atomic" occurs after plugging in dock. | 07:47 |
gnarface | GNUmoon2: maybe video driver related, you might have to specify a video driver in xorg | 08:09 |
gnarface | GNUmoon2: also i'm not sure about misc-firmware, did that come from the distro repos or some 3rd party? | 08:09 |
GNUmoon2 | I used this command: apt install -t beowulf-backports firmware-misc-nonfree | 08:11 |
GNUmoon2 | gnarface: sorry, I should have typed the actual command. | 08:12 |
gnarface | GNUmoon2: ah, i see. well then it's probably fine. check the xorg log to see if it's loading the modesetting driver or the intel driver, and force it to load whichever the other one is instead | 08:14 |
GNUmoon2 | yeah, I suspect it's the driver level problem, but I don't understand how thunderbolt dock works. Because thunderbolt is effectively a PCI bus, the integrated graphics would just see new connectors. | 08:14 |
gnarface | GNUmoon2: unfortunately i'm not familiar with that hardware either but first i would just try treating it regularly as a second display in xorg | 08:15 |
GNUmoon2 | gnarface: xorg load modesetting. | 08:15 |
gnarface | GNUmoon2: (which wouldn't work for me with the modesetting driver even without a dock) | 08:15 |
GNUmoon2 | I read that modesetting was preferred to the xorg-video-intel driver. | 08:16 |
gnarface | GNUmoon2: yea, it supports more resolutions and hardware in general, and it is way more stable. but unfortunately for you that doesn't matter | 08:17 |
gnarface | GNUmoon2: it's the default for a reason, but if you want *any* chance of vendor-specific hardware features to work, you're generally going to have to use their own driver | 08:18 |
GNUmoon2 | okay | 08:18 |
GNUmoon2 | mmm xserver-xorg-video-intel is installed according to apt. | 08:18 |
gnarface | yea you'll have to craft a small xorg.conf snippet to override the driver load | 08:19 |
gnarface | the entirety of what you need is at the top of "man intel" but you might need to check "man xorg.conf" for details if you're not familiar with the syntax overall | 08:20 |
GNUmoon2 | okay, I'll create xorg.conf and see if that works. Would I need to disable modesetting in grub (is it event set)? | 08:20 |
gnarface | hrmmm. i forget | 08:20 |
GNUmoon2 | yeah, I've created xorg.conf before...long time ago :) | 08:20 |
gnarface | you don't need the whole xorg.conf anymore, you can just have the Device stanza | 08:20 |
gnarface | that makes things simpler | 08:21 |
gnarface | and i forget if you need to disable modesetting in grub, i know you used to have to when switching from nouveau to nvidia but i don't recall if that was an issue for modesetting/intel (i thought not though) | 08:21 |
GNUmoon2 | gnarface: thanks for the help. I'll try it out. | 08:22 |
gnarface | no problem | 08:22 |
GNUmoon | Well, I switched to the intel driver via xorg.conf .....and nothing. Still no external displays detected via dock. | 09:42 |
gnarface | damn, sorry. well that's a bummer. | 09:48 |
gnarface | you said there were messages about it in dmesg though? | 09:49 |
gnarface | any errors? | 09:49 |
gnarface | GNUmoon^ | 09:49 |
gnarface | i wonder if it's just something simple like you have to also load a thunderbolt driver | 09:50 |
GNUmoon | gnarface: I installed the bolt package and authorised the dock. | 09:52 |
gnarface | GNUmoon: how about the thunderbolt-tools package? and can you confirm there's a kernel module loaded for it? i assume there would be one. | 09:53 |
gnarface | GNUmoon: (but the kernel module might not necessarily have auto-loaded correctly) | 09:53 |
GNUmoon | I had the thunderbolt-tools installed but it does the same thing as bolt, does it not? | 09:54 |
gnarface | i honestly don't know. i don't have any firsthand experience with this. | 09:55 |
gnarface | but there's certain common problems that fit a pattern | 09:55 |
GNUmoon | lsmod shows thunderbolt module is loaded. | 09:55 |
gnarface | oh, well that's good | 09:56 |
gnarface | maybe the module has some options? like debug options? | 09:56 |
GNUmoon | Logs: xhci_hcd 000:06:00.0: Host halt failed, -19 ... Host not accessible, reset failed. | 09:57 |
gnarface | heh | 09:57 |
gnarface | isn't xhci the USB though? | 09:57 |
gnarface | definitely doesn't look good either | 09:57 |
GNUmoon | yeah, that's what I thought. | 09:57 |
GNUmoon | I can still plug in a usb camera and it works via the dock. | 09:58 |
GNUmoon | red herring? | 09:58 |
gnarface | well, or another symptom of a shared cause | 09:59 |
gnarface | maybe | 09:59 |
GNUmoon | I wish I copied the log from Pop!OS to see what it is expected when it works. | 09:59 |
xinomilo | iirc usb3, not very stable | 09:59 |
xinomilo | try usb2 | 09:59 |
gnarface | looks like my version of the thunderbolt module only has one parameter: start_icm | 10:01 |
gnarface | what's ICM? | 10:01 |
gnarface | "start ICM firmware if it is not running (default: false) (bool)" | 10:01 |
GNUmoon | Just looking through logs, nothing else really stands out. | 10:04 |
gnarface | GNUmoon: do you power the dock up before you plug into it? | 10:06 |
GNUmoon | Yes, it's plugged into power before I plug into laptop. The laptop is charged/powered by the dock. | 10:07 |
gnarface | GNUmoon: it isn't a Mac though? | 10:07 |
GNUmoon | No, System76 | 10:07 |
gnarface | maybe the ICM thing needs to be enabled anyway | 10:09 |
gnarface | i'm seeing maybe a weird dependency on ACPI too in some cases? | 10:09 |
GNUmoon | I'm just looking up what it actually is... | 10:09 |
gnarface | you have acpid installed, right? | 10:09 |
GNUmoon | mmm nope...installing... | 10:10 |
gnarface | i would think you'd want that for a laptop anyway | 10:10 |
GNUmoon | gee, would have thought that was standard. | 10:10 |
gnarface | depends on how you installed | 10:10 |
GNUmoon | yeah! | 10:10 |
GNUmoon | minimal iso ... (very minimal then :) ) | 10:11 |
gnarface | oh, yea | 10:11 |
gnarface | you probably want to make sure you have pm-utils, too, if it didn't install automatically | 10:12 |
gnarface | there's a meta-package called task-laptop that should grab all sorts of stuff automatically if you want... | 10:12 |
GNUmoon | gnarface: thanks for the tip. | 10:13 |
GNUmoon | ICM = Internal Connection Manager (https://elixir.bootlin.com/linux/latest/source/drivers/thunderbolt/icm.c) | 10:15 |
gnarface | ah! | 10:15 |
gnarface | well that sound suspiciously important | 10:16 |
gnarface | ... to be off by default | 10:16 |
gnarface | hmmm | 10:16 |
GNUmoon | ...soooo might be important ;) | 10:16 |
ottavio | https://devuan.org/os/documentation/dev1fanboy/en/stretch-to-beowulf.html | 10:41 |
ottavio | How do I migrate from Stretch to Beowulf but with openrc instead of sysvinit? | 10:42 |
ottavio | The guide above mentions sysvinit-core but I'm interested in having openrc instead. | 10:42 |
gnarface | you can install openrc at any time | 10:47 |
gnarface | the default install might still require sysvinit parts | 10:48 |
gnarface | some people choose to replace it | 10:49 |
ottavio | gnarface: my question is: is it safe to install openrc and then remove systemd or do I still have to install sysvinit-core first before I migrate? | 10:51 |
gnarface | ottavio: i don't know for sure if you need to manually request sysvinit-core or if openrc will include it | 10:53 |
gnarface | ottavio: it would be easy to tell. | 10:54 |
ottavio | gnarface: Do I need sysvinit-core or not? | 10:55 |
gnarface | ottavio: yes | 10:55 |
ottavio | Why? | 10:55 |
gnarface | ottavio: that's the way debian has it | 10:55 |
ottavio | $ apt-get -s install openrc |tb | 10:56 |
ottavio | https://termbin.com/oqax | 10:56 |
ottavio | It's only recommended. | 10:56 |
gnarface | ottavio: well, i know before beowulf it required sysvinit-core | 10:57 |
gnarface | ottavio: i don't think the work to change it to the gentoo setup which does not got pushed through | 10:58 |
ottavio | In any case, the guide should include instructions for openrc. Can I open a bug about it? | 10:58 |
gnarface | i assume so | 10:59 |
gnarface | but to get the openrc setup the way you want it might still require some change to the openrc package | 10:59 |
gnarface | someone has already done that work, ask around here | 10:59 |
gnarface | it might be in the forums somewhere too | 11:00 |
gnarface | almost certainly is i would think | 11:00 |
ottavio | So, why is openrc offered as option in the installer? | 11:00 |
gnarface | popular demand | 11:00 |
ottavio | I asked on the forum. | 11:00 |
ottavio | Why is openrc offered as an option, yet it requires "changes to the package"? | 11:01 |
gnarface | what? | 11:01 |
gnarface | that's not what i said | 11:01 |
ottavio | Then I don't get it. | 11:01 |
gnarface | i said that if you want to fucking throw sysvinit-core overboard, it requires changes | 11:01 |
gnarface | it absolutely does not require changes to just use the way it is | 11:01 |
ottavio | ok calm down. | 11:01 |
gnarface | there is a historical reasoning to Debian's openrc setup that Devuan inherited | 11:02 |
gnarface | it requires some sysvinit stuff for compatibility | 11:02 |
gnarface | it's not popular amongst people jumping ship from Debian to use openrc with Devuan, so i mentioned that you can still change it | 11:03 |
gnarface | incidentally, every single other one of the openrc refugees went through this same series of panic attacks | 11:03 |
ottavio | That doesn't really answer my question, though. | 11:03 |
gnarface | (in the same order) | 11:03 |
gnarface | which one exactly didn't i answer? | 11:04 |
ottavio | <ottavio> gnarface: my question is: is it safe to install openrc and then remove systemd or do I still have to install sysvinit-core first before I migrate? | 11:04 |
gnarface | ottavio: i answered that. i said i assume openrc will include sysvinit-core, obviating the question. | 11:05 |
xinomilo | http://linuxmafia.com/kb/Debian/openrc-conversion.html | 11:05 |
ottavio | But openrc does not require sysvinit-core on Debian. | 11:05 |
gnarface | ottavio: to be absolutely clear, if it doesn't, then i would not assume it is safe to ignore. i would still install it if that guide says. you can always remove it later. | 11:06 |
gnarface | ottavio: but i didn't try this myself and it's not the majority of users who has. if you're patient one of them will show up eventually. | 11:07 |
gnarface | ottavio: if you don't want to add risk, just do a regular upgrade and add openrc afterwards | 11:08 |
gnarface | ottavio: it won't matter whether you add openrc before or after the upgrade | 11:12 |
ottavio | gnarface: I've read a few threads and I now understand what you meant. Openrc is not used as init by Devuan. | 11:33 |
mason | GNUmoon: I'm late to this, but I've got one system where I need to explicitly trust the dock in BIOS, explicitly outside of the OS, before all of it works. | 15:14 |
fsmithred | mason, do you have a rebuilt unbound you would like me to test? | 15:18 |
fsmithred | I can play with it later today. | 15:18 |
mason | fsmithred: If you want to test, sure. If you want to build your own, I chopped out that one ifdef from the pastebin last night and rebuilt, no other changes. | 15:19 |
mason | half a sec and I'll whip up a patch. | 15:19 |
fsmithred | if the packages aren't big, you could send them in email | 15:20 |
mason | Not too big for an email, but I'll whip up a patch too. | 15:21 |
fsmithred | thanks | 15:24 |
fsmithred | heading out now - bbl | 15:24 |
mason | fsmithred: gmail has it | 15:26 |
mason | For folks following along at home: https://bpa.st/KVNA | 15:30 |
mason | Might be worth doing this in a throwaway VM as I had to scrape the systemd-logind off my system again when it was done. | 15:31 |
fsmithred | Damn, maybe USPS just took over gmail. I don't see it yet. | 15:31 |
mason | Sep 9 09:26:21 phlegethon postfix/smtp[20208]: EB5E1194CFD: to=<fsmithred@gmail.com>, relay=gmail-smtp-in.l.google.com[74.125.132.26]:25, delay=11, delays=7.7/0.04/0.35/3, dsn=2.0.0, status=sent (250 2.0.0 OK 1599657981 u5si1970540ilk.143 - gsmtp) | 15:31 |
fsmithred | ah, there it is | 15:31 |
mason | maybe check spam? | 15:31 |
mason | kk | 15:31 |
fsmithred | tubes must be clogged | 15:31 |
mason | Google could use a good purgative. | 15:31 |
ottavio | Is the DNG mailing list available on Gmane.io ? | 15:43 |
markizano | https://hub.docker.com/r/markizano/devuan | 19:15 |
markizano | a tee hee | 19:15 |
markizano | <: | 19:15 |
markizano | Is there officially supported Docker images already / am I just re-inventing the wheel here? Didn't see one in Docker hub when searching... | 19:16 |
dormito | I recently download a devuan live image, however "apt-get update" complains about "Repository [...] InRealse' changed its 'Suite' value from 'testing' to 'stable' ", it referes to a man page, but that man page doesn't seem to talk about what ever that means | 20:09 |
dormito | how do I use the repos? | 20:09 |
markizano | dormito, if it's auto.mirrors.devuan.org - change to deb.devuan.org if you are >jessie; use archive.devuan.org if == jessie | 20:12 |
golinux | Guess you haven't read the Release Notes. https://files.devuan.org/devuan_beowulf/Release_notes.txt | 20:12 |
golinux | Always a good idea to do so;. | 20:13 |
golinux | "The first time you do this after Beowulf has gone Stable, you will get a warning that the repository changed its 'Suite' value from 'testing' to 'stable'. Answer 'yes' to proceed. Note that aptitude or apt-get will fail this first time. Only apt will ask you to accept the change" | 20:14 |
dormito | I guess nobody bothered to post an updated livecd image after it went stable | 20:17 |
dormito | so I have to use the apt cmd? | 20:18 |
brocashelm | yes | 20:18 |
yanmaani | why did you deprecate the old domain? | 20:19 |
yanmaani | Couldn't you just have set it up to point to the new one? | 20:19 |
dormito | hmmm is the amdgpu firmware in nonfree or something? it seems kinda odd to include the amdgpu kernel module, and xorg driver but not firmware in the livecd | 20:22 |
fsmithred | dormito, firmware-amd-graphics is the package in non-free | 20:24 |
SuaveDandy | So I'm installing Devuan through the CLI. | 20:24 |
SuaveDandy | How do I use debootstrap? | 20:24 |
fsmithred | SuaveDandy, are you doing it from an installed system or from a live-CD | 20:25 |
fsmithred | ? | 20:25 |
SuaveDandy | Live CD. | 20:25 |
fsmithred | which one? | 20:25 |
SuaveDandy | The minimal one with your installer. | 20:25 |
SuaveDandy | The one that's GPT. | 20:26 |
fsmithred | oh, the uefi iso? | 20:26 |
SuaveDandy | From the Refracta website. | 20:26 |
SuaveDandy | Yes. | 20:26 |
fsmithred | yeah, I remember that | 20:26 |
fsmithred | hang on, I need to look at the debootstrap command | 20:26 |
SuaveDandy | I want to install OpenRC instead of sysvinit as well. | 20:27 |
fsmithred | debootstrap --arch amd64 beowulf /mnt http://deb.devuan.org/merged | 20:28 |
fsmithred | you might try adding --include=openrc | 20:28 |
fsmithred | not sure if that will work | 20:28 |
dormito | fsmithred: thanks. its a royal PITA to get this live cd to working when the gpu driver won't load, and X wont start because it hates FB mode for some reason | 20:28 |
fsmithred | and that command assumes you mounted the target partition on /mnt | 20:28 |
fsmithred | dormito, is that the desktop-live? | 20:29 |
fsmithred | I thought the firmware was installed | 20:29 |
SuaveDandy | Is it required to be installed from the Internet? I wanted to have a minimal system just like the live ISO. | 20:29 |
fsmithred | debootstrap requires a repository to pull from. That's usually our repo, but you could use a local one if you have it. | 20:30 |
dormito | fsmithred: yup | 20:31 |
SuaveDandy | The only thing important is that install the OS without a desktop. | 20:31 |
SuaveDandy | As in CLI only. Want to install a specific window manager. | 20:32 |
fsmithred | why not just install the minimal-live system? You don't need internet for that. | 20:32 |
SuaveDandy | That's why I'm asking if debootstrap can install the live system. | 20:33 |
SuaveDandy | The thing is that I'm installing on ZFS. | 20:33 |
fsmithred | no, debootstrap installs just the base system | 20:33 |
SuaveDandy | And the guide told me to use debootstrap. | 20:33 |
fsmithred | slightly less than what you get with the devuan-installer if you un-check all the boxes at tasksel | 20:33 |
fsmithred | which guide? | 20:33 |
SuaveDandy | https://openzfs.github.io/openzfs-docs/Getting Started/Debian/Debian Buster Root on ZFS.html#step-1-prepare-the-install-environment | 20:34 |
SuaveDandy | Sadly, IRC doesn't include spaces in the link. | 20:34 |
SuaveDandy | It's a mess. | 20:34 |
fsmithred | wow, ok. I didn't look at the directions. | 20:35 |
fsmithred | If you can create and format partitions manually, you should be able to use refractainstaller | 20:35 |
fsmithred | probably easier to install as is and then switch to openrc after you reboot into the new system | 20:36 |
SuaveDandy | I've already set up ZFS pools. They are mounted to /mnt. | 20:36 |
fsmithred | do they show up as /dev/something? | 20:37 |
SuaveDandy | The partitions? Yeah. | 20:37 |
SuaveDandy | 3 partitions: ESP, the boot pool and the root pool. | 20:37 |
SuaveDandy | With the root pool being encrypted with the native ZFS encryption. | 20:38 |
fsmithred | dormito, I will be sure to include firmware-amd-graphics in the next desktop-live isos. I thought I already did that, but I'm not seeing it. | 20:38 |
rwp | JFTR but I see spaces in the posted IRC okay. No problem. But the spec says they should be encoded with %20 encoding. | 20:40 |
SuaveDandy | I don't think you can install on the partitions directly. The guide had a specific mounting process. | 20:40 |
SuaveDandy | The pools are all mounted to /mnt. I think you can install to /mnt only. | 20:41 |
fsmithred | dormito, if you can boot to console, you can install the package with 'dpkg -i /firmware-amd-graphics' | 20:41 |
fsmithred | oops | 20:41 |
fsmithred | that path is /firmware/firmware-amd-graphics | 20:41 |
fsmithred | that path is /firmware/firmware-amd-graphics*.deb | 20:42 |
dormito | fsmithred: yeah I did that, and then restarted slim. but I had to also enable non-free | 20:44 |
fsmithred | oh, does it need some other package to work? | 20:45 |
dormito | (well there was also a correctly times modprobe -r amdgpu; modprobe amdgpu) | 20:45 |
SuaveDandy | fsmithred: Can I install to /mnt with Refracta installer? | 20:45 |
fsmithred | SuaveDandy, yes | 20:45 |
fsmithred | oh | 20:45 |
SuaveDandy | Good. | 20:45 |
fsmithred | no | 20:45 |
SuaveDandy | Not good. | 20:45 |
fsmithred | it wants a partition | 20:45 |
SuaveDandy | Oof. | 20:45 |
SuaveDandy | That's gonna be problematic. | 20:45 |
dormito | oh, are you saying that the dpkg is included on the image, but not installed? I used the repos to install it (I didn't know the deb was present) | 20:45 |
fsmithred | like /dev/sda1 or /dev/mmcblk0p1 or /dev/mapper/mypart | 20:46 |
fsmithred | yes, dpkg is installed in the live isos | 20:46 |
SuaveDandy | The system needs to be installed into mounted ZFS pools. Plus I have ZFS encryption. | 20:47 |
fsmithred | dpkg Priority is "required" | 20:47 |
SuaveDandy | Not LUKS. | 20:47 |
SuaveDandy | The native one. | 20:47 |
golinux | <dormito> I guess nobody bothered to post an updated livecd image after it went stable | 20:47 |
golinux | You are always welcome to join the release team to streamline the process to perfection | 20:48 |
mason | SuaveDandy: What issues are you encountering, if you could summarize? | 20:48 |
dormito | golinux: lol I would consider it... if I didn't prefer gentoo for most of my machines (but sometimes I have to setup stuff for other people) | 20:48 |
fsmithred | SuaveDandy, maybe take a look at this to see how I used refractainstaller: https://dev1galaxy.org/viewtopic.php?id=2323 | 20:49 |
SuaveDandy | I created the pools, mounted them, created datasets and now I need to install the system. | 20:49 |
fsmithred | yeah, and I don't really remember what I did. I wrote it down because I knew I'd forget. | 20:50 |
mason | SuaveDandy: I need to write this up as docs rather than a script, but all the concepts you want are here: https://bpa.st/X3OA | 20:50 |
fsmithred | I think there's a way to do what you want. | 20:50 |
mason | SuaveDandy: Don't just use the script intact unless you're willing to fiddle with GRUB a bit. I wasn't using GRUB with Devuan for a while, and I need to figure out some install ordering issues. | 20:51 |
mason | SuaveDandy: But the upshot is that if you pick a path to adventure, you can observe a native-ZFS-encrypted install (potentially on a mirror) there. | 20:51 |
mason | Also, FWIW, I find it convenient to partition to accomodate both UEFI and legacy at once, so your layout will likely be mildly different. | 20:52 |
mason | (And the Debian bits are misleading. There'll be more pain than indicated in forceing debootstrap to not install sysvinit.) | 20:54 |
fsmithred | but it's easy to switch over after the install | 20:55 |
mason | That's true. | 20:56 |
mason | Just nettlesome that it's a fight to have the stuff never installed in the first place. | 20:56 |
fsmithred | SuaveDandy, I think you should stick with the debootstrap instructions | 20:58 |
SuaveDandy | Got disconnected because I have to use a VPN to access Dev1 Galaxy for whatever reason. | 20:58 |
dormito | Hmmm. to install openrc you need to do an advanced install right? I dont see an advanced button, or an equivelent check box in the installer | 20:58 |
fsmithred | dormito, in beowulf it's in the regular install | 20:59 |
fsmithred | expert install is in the Advanced option in the installer boot menu | 20:59 |
SuaveDandy | Reporting that Dev1 Galaxy isn't loaded when you're in Russia for God knows what reason. | 20:59 |
fsmithred | weird | 20:59 |
SuaveDandy | It may be blocked by IP. | 21:00 |
fsmithred | we're considered terrorists? | 21:00 |
fsmithred | oh, you might be blocked | 21:00 |
SuaveDandy | Our censorship… is a funny thing. | 21:00 |
fsmithred | any censorship is weird | 21:00 |
SuaveDandy | It blocks all kinds of things. The things that I like, especially. We have an abusive relationship. | 21:01 |
SuaveDandy | Might as well just subscribe to a VPN provider. | 21:01 |
fsmithred | yeah | 21:01 |
golinux | There is also the snaplock on the forum that denies access to funky IP addresses | 21:03 |
SuaveDandy | Ohhhhhhhhhh. | 21:03 |
SuaveDandy | I have an adblocking DNS. | 21:03 |
SuaveDandy | AdGuard, to be precise. | 21:04 |
fsmithred | golinux, do we ban any groups of IP addresses, or is it all individual addresses? | 21:04 |
golinux | If you land on a tainted exit node in a VPN you will be denied access to the forum | 21:04 |
dormito | I dont see anything about the init/rc system in the installer (and it already finished installing) | 21:04 |
fsmithred | dormito, are you installing beowulf or ascii? | 21:04 |
dormito | beowulf | 21:04 |
golinux | Should be in expert install | 21:04 |
fsmithred | choose_init is in the regular install | 21:04 |
SuaveDandy | Maybe your forum doesn't like my DNS much. | 21:05 |
fsmithred | well, if you already installed the system, just install openrc now. That's easier than reinstall the whole system. | 21:05 |
golinux | Could be. Even I have gotten slammed a few times. | 21:05 |
SuaveDandy | Ouch. | 21:05 |
dormito | how do you launch the expert install? or should i just chroot into the system and run the apt-get... is that safe? | 21:06 |
golinux | dormito: Screenshots of the installer https://devuan.org/os/documentation/install-guides/beowulf/install-devuan | 21:06 |
SuaveDandy | It's getting very late. I must wrap up with my installation. | 21:06 |
fsmithred | step 19 | 21:06 |
dormito | so do I need to use a cli installer? (is the option not in w/e gui install was present) | 21:07 |
fsmithred | you found a gui installer? | 21:07 |
fsmithred | I thought we left that out | 21:07 |
dormito | yeah. its... barely existant, but yes | 21:07 |
fsmithred | what iso are you using? | 21:07 |
golinux | There is no gui debian installer for beowulf | 21:08 |
dormito | it requires running gparted/gdisk directly (but it will launch them for you) | 21:08 |
fsmithred | oh | 21:08 |
fsmithred | you have a live iso | 21:08 |
dormito | yeah | 21:08 |
fsmithred | that's refractainstaller | 21:08 |
fsmithred | that just installs what is on the iso | 21:09 |
fsmithred | so reboot into the new system, get root terminal, apt update, apt install openrc | 21:09 |
dormito | ah. ok. The page you linked said you must choose the expert install from the boot screen (grub,sysiso or w/e it is). since my systems already booted, is it possible to luanch with a switch or do I have to reboot? | 21:10 |
fsmithred | yeah, that was for the installer isos, not the live isos | 21:10 |
fsmithred | no expert install in the live isos | 21:10 |
fsmithred | different installer | 21:10 |
golinux | Jeesh . . . took forever to figure that out . . . | 21:11 |
fsmithred | dormito, did you already finish the install? | 21:11 |
SuaveDandy | THERE we go. http://manpages.ubuntu.com/manpages/trusty/man1/cdebootstrap.1.html | 21:14 |
SuaveDandy | This one has flavors like build, minimal and standard. | 21:16 |
SuaveDandy | Wonder if it works. | 21:17 |
fsmithred | try man debootstrap in devuan or debian | 21:18 |
SuaveDandy | "man debootstrap" or "man cdebootstrap?" | 21:18 |
fsmithred | I know a lot of people like to use minbase but that excludes some command commands | 21:19 |
fsmithred | I've never used cdebootstrap. Don't know it. | 21:19 |
SuaveDandy | minbase? | 21:19 |
fsmithred | command commands/important commands | 21:19 |
fsmithred | --variant=minbase|buildd|fakechroot | 21:20 |
fsmithred | without a variant, you get all required and important packages | 21:21 |
SuaveDandy | Will it install Xorg and desktop as well? | 21:22 |
fsmithred | LOL! | 21:22 |
fsmithred | please remember to create a root password | 21:22 |
fsmithred | and remember to install a kernel | 21:22 |
fsmithred | without those two things, the first reboot is really difficult | 21:22 |
SuaveDandy | Talking about the base install. | 21:22 |
fsmithred | what base install? | 21:23 |
fsmithred | debootstrap? | 21:23 |
SuaveDandy | debootstrap without --variant=minbase. | 21:23 |
fsmithred | no xorg, but you will get some things like 'less' | 21:23 |
SuaveDandy | Nice. | 21:23 |
fsmithred | and some other commands you expect to be there | 21:23 |
SuaveDandy | That's all I needed, actually. | 21:23 |
SuaveDandy | Was scared it'll install Xorg. | 21:24 |
fsmithred | not even close | 21:24 |
SuaveDandy | Fewh. | 21:24 |
SuaveDandy | debootstrap it is then. | 21:24 |
fsmithred | it will install libsystemd0, but you can replace that with libelogind0 | 21:24 |
SuaveDandy | Can it actually install libsystemd from Devuan's repos tho? | 21:25 |
SuaveDandy | I remember it's blacklisted. | 21:25 |
fsmithred | what do you mean? | 21:25 |
SuaveDandy | All the Systemd packages are blacklisted. | 21:25 |
fsmithred | no, libsystemd0 is not blacklisted. It's just a library that will say "systemd is not installed" when apps ask about that. | 21:25 |
SuaveDandy | So how do I replace it with libelogind exactly? | 21:26 |
fsmithred | apt install libelogind0 | 21:26 |
SuaveDandy | Can't you just --exclude=libsystemd --include=libelogind? | 21:27 |
fsmithred | don't know. Try it and let us know if it works. | 21:27 |
fsmithred | must add the 0 to the name | 21:27 |
SuaveDandy | Alright. | 21:27 |
SuaveDandy | Is elogind installed with debootstrap tho? | 21:35 |
gnarface | probably not unless you ask for it too | 21:35 |
gnarface | debootstrap won't even install a kernel unless you remember to ask it | 21:36 |
gnarface | it does not leave you with a bootable system on it's own | 21:36 |
gnarface | (people usually chroot into the debootstrap target to finish setup manually) | 21:37 |
SuaveDandy | Is the kernel package called linux? | 21:41 |
gnarface | SuaveDandy: linux-image-* and linux-headers-* | 21:42 |
SuaveDandy | linux-headers-$(uname -r)? | 21:43 |
SuaveDandy | Already installed. | 21:43 |
SuaveDandy | On the live system tho. | 21:43 |
SuaveDandy | But is it the same? | 21:43 |
gnarface | SuaveDandy: yes, but inside the debootstrap target you're going to need one as well | 21:43 |
gnarface | SuaveDandy: try this one: apt-cache search '^linux-(headers|image)' | 21:44 |
SuaveDandy | Or you do apt install linux-headers-$[uname -r] after chroot? | 21:44 |
gnarface | SuaveDandy: yes, i meant after chroot | 21:44 |
gnarface | SuaveDandy: don't forget linux-image-* | 21:45 |
gnarface | SuaveDandy: you only might need a linux-headers-* but you absolutely will need a linux-image-* | 21:45 |
SuaveDandy | Is it installed by using $(uname -r) variable as well? | 21:45 |
ranix | gnuface | 21:45 |
SuaveDandy | Do I do this after installing GRUB and setting up the users? | 21:46 |
gnarface | SuaveDandy: you're on a amd64 system? uname -r might or might not match the current version. if it is a amd64 system though, "linux-image-amd64" always points to the latest | 21:46 |
gnarface | SuaveDandy: *before* installing GRUB. won't matter what order you add the users. | 21:47 |
ranix | debootstrap basically installs the userspace on a disk | 21:47 |
ranix | then you install the kernel and install grub separately | 21:48 |
SuaveDandy | And I've already used linux-headers-(uname -r) for the live system. | 21:48 |
ranix | I usually debootstrap then chroot into the userspace to set up the rest of the packages | 21:48 |
ranix | or just use a usb netinst | 21:48 |
gnarface | ranix: SuaveDandy is paranoid about unnecessary packages and is trying to accomplish a minimal debootstrap for his first time | 21:49 |
ranix | first time installing linux ever? | 21:49 |
SuaveDandy | Yep. | 21:49 |
ranix | you should just install off a usb stick and take what you get until you're more familiar with it | 21:50 |
fsmithred | omg, I didn't know that | 21:50 |
fsmithred | ranix, he's doing this on zfs | 21:50 |
fsmithred | you are bold, dude | 21:50 |
SuaveDandy | Well, I've installed Void Linux before. | 21:51 |
SuaveDandy | Yes, I'm bold. | 21:51 |
SuaveDandy | A Windows user installing Void Linux from CLI with LUKS on root. | 21:51 |
ranix | the kind of person who installs void linux can usually just change their desktop to a skull and crossbones and open up a few cmd prompts before setting the text color to green and be happy with just that | 21:51 |
SuaveDandy | And Wayland's Sway. | 21:51 |
fsmithred | lol | 21:52 |
SuaveDandy | It wasn't too hard. Just needed to follow instructions. | 21:52 |
fsmithred | SuaveDandy, do your debootstrap and chroot instructions include bind-mounts for /dev /proc and /sys? | 21:52 |
fsmithred | you'll need that to do grub-install in chroot | 21:52 |
SuaveDandy | Yes. | 21:52 |
SuaveDandy | Of course. | 21:53 |
SuaveDandy | How can it be otherwise? | 21:53 |
ranix | https://wiki.debian.org/Debootstrap is the cheat sheet I use basically every time | 21:53 |
SuaveDandy | I mean, /dev is like all the connected devices. | 21:54 |
fsmithred | hey, my first kernel compile was so I could compile sound in the kernel, and I ended up with a kernel that only had sound | 21:54 |
SuaveDandy | You obv need that in chroot. | 21:54 |
ranix | fsmithred: No screens found. | 21:54 |
ranix | because your mouse wasn't plugged in | 21:54 |
ranix | those were the days | 21:54 |
fsmithred | no filesystems found | 21:54 |
fsmithred | or known | 21:54 |
fsmithred | I need to go outside. biab | 21:56 |
SuaveDandy | Yeah, I hopped on lots of distros. Alpine included. | 21:57 |
SuaveDandy | Alpine was a mistake. | 21:57 |
SuaveDandy | Void is rolling. And I am lazy. | 21:57 |
SuaveDandy | Elementary had a bug with Cerbere. | 21:57 |
SuaveDandy | Regolith's installer crashed on me. ArchLabs broke my PC. It's a mess. | 21:58 |
SuaveDandy | Guys, Debian is EASY… I just wanted to install it on ZFS. That's why I'm fiddling around with debootstrap. | 22:02 |
SuaveDandy | Emmmm, can't cd to http://deb.devuan.org | 22:08 |
SuaveDandy | Weird. | 22:08 |
SuaveDandy | AHHHHHHH. | 22:08 |
SuaveDandy | Damn, forgot to add "beowulf." | 22:09 |
SuaveDandy | Ouch. | 22:09 |
SuaveDandy | Yep, it's bootstraping. | 22:10 |
SuaveDandy | Look at it rolling. | 22:12 |
ranix | it's debootstrapping | 22:23 |
ranix | you thought it was bootstrapping but it was actually the opposite | 22:23 |
yanmaani | im booooooooooootstrapping | 22:23 |
SuaveDandy | Ohhhh, so that's what "de-" means. | 22:31 |
SuaveDandy | I thought it stands for Debian. | 22:32 |
SuaveDandy | So, as for rbinding. | 22:34 |
SuaveDandy | Do you do it like "for dir in dev proc sys: do?" | 22:35 |
SuaveDandy | Wait a sec. Let me check. | 22:35 |
MinceR | bian is the opposite of debian | 22:35 |
SuaveDandy | That's logical. | 22:36 |
SuaveDandy | Except that bian doesn't exist. | 22:36 |
ranix | deb exists but ian was murdered by the police | 22:37 |
SuaveDandy | I was wrong. | 22:37 |
SuaveDandy | for dir in dev proc sys run; do | 22:38 |
SuaveDandy | fsmithred: You know, --include and --exclude do indeed work. | 22:53 |
fsmithred | you got openrc without going through sysvinit first? | 22:53 |
SuaveDandy | Yeah, it's here. | 22:55 |
SuaveDandy | On the chrooted system. | 22:55 |
SuaveDandy | Was I not supposed to include it? | 22:56 |
fsmithred | huh? I'm just verifying that I you got what you wanted. | 22:58 |
ranix | <3 sysvinit | 22:59 |
ranix | I sucked it up and allowed upstart and learned it and even made a product with it | 22:59 |
ranix | I figured if so many people were going to start that much shit over an init system it wasn't worth fighting against | 23:00 |
SuaveDandy | OpenRC still relies on SysVinit, right? | 23:00 |
ranix | and then as soon as they were done raising hell they did it again | 23:00 |
ranix | give these people an inch | 23:00 |
SuaveDandy | Only Gentoo and Alpine use OpenRC-init? | 23:01 |
ranix | no idea, I'm never learning another init system | 23:01 |
fsmithred | yes, openrc uses sysvinit scripts | 23:01 |
ranix | experience has taught me that everyone who has a problem with sysvinit is retarded | 23:02 |
SuaveDandy | Oof. | 23:02 |
ranix | you might say no, that's not the case in all situations | 23:02 |
ranix | and yes sysvinit has problems | 23:02 |
ranix | but the real world has spoken | 23:02 |
ranix | just try to imagine | 23:03 |
SuaveDandy | I guess OpenRC is still good for service management. Tho it is technically not an init system then. On Devuan, at least. | 23:04 |
ranix | someone, probably sloshed out of his fucking mind, proposed with complete seriousness that the idiot who wrote pulseaudio be allowed to author yet another init system and everyone should switch to it | 23:04 |
ranix | and everyone said wow what a great idea | 23:04 |
SuaveDandy | ranix: Having a bad day? | 23:05 |
ranix | then they went back to arguing about whether or not users of python should be allowed to use the terms "master" and "slave" in man files | 23:05 |
SuaveDandy | Ah, that dum-dum thing. | 23:05 |
ranix | hasn't anyone considered whether man files themselves are inherently toxic and masculine | 23:06 |
SuaveDandy | Yeah, not something I like either. | 23:06 |
SuaveDandy | Actually, I remember people talking about man being toxic. | 23:07 |
yanmaani | no | 23:07 |
SuaveDandy | So there's that. | 23:07 |
yanmaani | you've got to be kidding me | 23:07 |
SuaveDandy | No. | 23:07 |
yanmaani | this must be a joke | 23:07 |
SuaveDandy | Ah, yes. | 23:07 |
SuaveDandy | Yes, it is. | 23:07 |
ranix | this is why I was against removing bitchx | 23:07 |
yanmaani | surely people can't be this stupid | 23:07 |
SuaveDandy | People were joking about replacing the name of man due to it being toxic. | 23:08 |
SuaveDandy | Due to all this happening. | 23:08 |
yanmaani | You have to fight back. If you have 10 guys saying "nuke every trace of master/slave" and 10 guys saying "nah it's fine as it is", it'll end up somewhere in the middle - "well it's a bit problematic but whatever" | 23:08 |
SuaveDandy | I guess so. But how? | 23:09 |
ranix | no they'll never stop | 23:09 |
yanmaani | yes, that's why you have to push against them | 23:09 |
yanmaani | there has to be an equal and opposite force | 23:09 |
ranix | they will use systemd to get rid of you | 23:09 |
ranix | and then when you're gone they'll change upstream however they want | 23:10 |
yanmaani | SuaveDandy: use more offensive and less anodyne terms | 23:10 |
SuaveDandy | Ah, that makes sense. | 23:10 |
SuaveDandy | I wasn't even considering using anything other than master. | 23:10 |
luser977 | inclusiveness requires 33% man pages, 33% fem pages and 33% oth(er) pages | 23:10 |
yanmaani | you should obviously keep using master. But if their guys are sending in 30 PRs a day to change terms to less offensive terms, you have to send in 30 PRs a day to change terms to more offensive terms. | 23:11 |
yanmaani | That's how it works. | 23:11 |
SuaveDandy | I can imagine fem being for configs. | 23:11 |
ranix | this page intentionally left gender-neutral | 23:11 |
SuaveDandy | That would be kinda funny. fem as man 5. | 23:11 |
ranix | be sneaky and make the default visualizer for man more and fem less | 23:12 |
SuaveDandy | So we have man for man 8, fem for man 5… but what would nonbin act as? | 23:12 |
yanmaani | UB | 23:13 |
yanmaani | picks one at random | 23:13 |
luser977 | oth | 23:14 |
GNUmoon | mason: I'm late getting back to you :) The System76 laptop has coreboot from System76, so there are no settings to change, unless I rebuild coreboot. That said, the dock worked using Pop!OS after I authorizedd the dock in Gnome gui. | 23:15 |
GNUmoon | So theoretically the bios is not the problem. | 23:15 |
mason | GNUmoon: Hrm. Unless there's some special support needed that doesn't exist in coreboot. I'm not clear on what's needed to enable the dock. Mine was on a ThinkPad X1C6. | 23:16 |
SuaveDandy | Heeheheheeeeeh. SysV-RC isn't installed. Good sign of OpenRC installing correctly. | 23:17 |
SuaveDandy | fsmithred: Yeah, debootstrap with --include and --exclude is pretty much everything one needs. No chroot required. | 23:19 |
SuaveDandy | I mean, it's still required for setting up things but still. | 23:19 |
fsmithred | how you gonna make a root password? | 23:19 |
fsmithred | oh, ok | 23:20 |
SuaveDandy | Legit method of installing the kernel and stuff. | 23:20 |
SuaveDandy | I'm setting up the locales. | 23:21 |
SuaveDandy | Do you think I should add my native language? | 23:21 |
SuaveDandy | Or should I just leave English and not bother? | 23:22 |
ranix | depends on what your native language is | 23:22 |
ranix | and how you type it | 23:22 |
ranix | you can always dpkg-reconfigure locales later | 23:23 |
SuaveDandy | I mean, I would still need the keyboard language. | 23:23 |
SuaveDandy | Hm, I don't seem to be able to dpkg-reconfigure the locales. | 23:25 |
ranix | are you doing it as root | 23:26 |
ranix | sudo | 23:26 |
SuaveDandy | Of course. | 23:26 |
SuaveDandy | I've chrooted into /mnt | 23:26 |
SuaveDandy | Maybe that's why I'm not able to. | 23:26 |
gnarface | locales is a package. it's probably not installed by default | 23:27 |
gnarface | s/probably// | 23:27 |
SuaveDandy | I've installed it just now. | 23:27 |
ranix | I just assumed "setting up the locales" meant setting up the locales package | 23:27 |
SuaveDandy | No. | 23:27 |
SuaveDandy | dpkg-reconfigure locales. | 23:28 |
fsmithred | doesn't it go through that when you install the package? | 23:28 |
gnarface | SuaveDandy: it is supposedly safe to add multiple locales, but to avoid bugs i would recommend you do not omit utf8 and latin1 | 23:28 |
gnarface | fsmithred: i think the debconf priority is different in a debootstrap install | 23:28 |
SuaveDandy | No, it does not. | 23:28 |
SuaveDandy | So you think I should've chrooted before installing the kernel? | 23:30 |
ranix | look at this guy chrooting around all over the place | 23:30 |
SuaveDandy | I mean, I installed the kernel through debootstrapping and then chrooted into /mnt. | 23:31 |
SuaveDandy | I'm starting reconsidering using ZFS on Debian. | 23:33 |
ranix | the kernel doesn't really affect your ability to chroot | 23:33 |
ranix | presence or absense | 23:33 |
ranix | zfs is not usually worthwhile | 23:34 |
ranix | use ext4 if it's your first time | 23:34 |
SuaveDandy | Really? | 23:34 |
ranix | it's worth experimenting with btrfs or zfs on a secondary drive or partition | 23:34 |
SuaveDandy | Ah. | 23:34 |
SuaveDandy | But the snapshots. | 23:34 |
ranix | snapshots of a bunch of shit logrotate deletes anyway | 23:35 |
specing | btrfs is great | 23:35 |
specing | you can delete the snapshots whenever you want | 23:35 |
specing | keep in mind that ext4 has no data integrity assurance | 23:36 |
SuaveDandy | The ZFS manual told me how to not snapshot stuff like cache and tmp. | 23:36 |
specing | a bad disk can change your data with impunity | 23:36 |
specing | tmp should be on tmpfs, anyway | 23:36 |
specing | unless the temporaries are big, they should never hit the disks | 23:36 |
SuaveDandy | One person says to me tmpfs eats RAM. | 23:37 |
SuaveDandy | You tell me /tmp is supposed to be on tmpfsd | 23:37 |
SuaveDandy | What am I even supposed to do at this point? | 23:37 |
gnarface | SuaveDandy: tmpfs doesn't eat much ram | 23:38 |
gnarface | SuaveDandy: what you're supposed to do at this point is decide for yourself | 23:38 |
mason | Use it for builds and it can. :) | 23:38 |
gnarface | right, which is partially why i have a 5GB physical /tmp partition on my workstation but the virtual servers all have between 32MB and 64MB of tmpfs for /tmp/ instead | 23:39 |
gnarface | (most people are not going to use more than 16MB but some stuff like firefox and mysql, and certain kernel packages can get messy) | 23:40 |
mason | Rust | 23:40 |
SuaveDandy | Maybe I should instead use BtrFS? Or is it even harder to maintain? | 23:40 |
mason | SuaveDandy: I'd stick with ZFS, but you can use whichever. If you use BtrFS, stick to a very simple config and make sure you have back-ups. | 23:41 |
gnarface | SuaveDandy: ext4 is the most common choice so that's the one that's gonna generally get the best support, but i would recommend XFS anyway | 23:43 |
gnarface | SuaveDandy: everyone has a different opinion about this but it's more important you practice with all of them than you decide now | 23:43 |
mason | Once XFS grows self-healing by way of Stratis I'll start recommending it. | 23:44 |
SuaveDandy | Can XFS do snapshots? | 23:44 |
gnarface | SuaveDandy: i think there is some well-documented snapshot mechanism for it yes | 23:45 |
gnarface | SuaveDandy: not sure if it's a built-in or a bolt-on though (suspect it's more of a bolt-on) | 23:45 |
SuaveDandy | mason: can you install the system on ZFS through fsmithred's installer? | 23:45 |
syco- | you might try NTFS integrated right into the kernel now :p | 23:46 |
mason | SuaveDandy: I use his live images now to install Devuan on ZFS. That's not the same as an installer. | 23:46 |
SuaveDandy | Wow, XFS having snapshots? Did not expect that. | 23:46 |
syco- | (kidding) | 23:46 |
SuaveDandy | I'm talking about Refracta installer. | 23:47 |
fsmithred | SuaveDandy, what's the mount command for whatever you mount at /mnt? | 23:47 |
gnarface | SuaveDandy: https://en.wikipedia.org/wiki/XFS#Snapshots | 23:47 |
fsmithred | what's the name of the device you mount? | 23:48 |
GNUmoon | mason: Well, I don't think the problem is in coreboot, given Pop!OS worked previously. When I get a chance, I'll live boot Pop!OS again and see what is going on in the logs and versions of packages...cannot think what else to do at this stage. | 23:48 |
gnarface | SuaveDandy: (short version is it doesn't support snapshots itself, but it has xfs_freeze, after using which you can snapshot with basically anything) | 23:48 |
GNUmoon | I'll also check xorg logs, lsmod, etc... | 23:49 |
gnarface | SuaveDandy: it's probably worth mentioning that the practical value of snapshots over tar is in most cases entirely hypothetical | 23:49 |
SuaveDandy | Gimme a sec. | 23:50 |
syco- | rsync | 23:50 |
SuaveDandy | fsmithred: rpool/ROOT/devuan and bpool/BOOT/devuan | 23:53 |
SuaveDandy | The filesystem datasets. | 23:53 |
fsmithred | wow, I have to look at the code. I think it expects /dev/something | 23:54 |
fsmithred | ok, $install_dev is whatever you type in when it asks (in the cli installer) | 23:56 |
fsmithred | and then it gets tested to see if it's a block device | 23:56 |
fsmithred | will that test pass or fail? | 23:56 |
fsmithred | same for $boot_dev if you're using separate /boot | 23:57 |
SuaveDandy | You first create rpool and bpool ZFS pools, then you create rpool/ROOT and bpool/BOOT container datasets and after that you create and mount rpool/ROOT/devuan and bpool/BOOT/devuan | 23:57 |
fsmithred | so... | 23:57 |
fsmithred | if [ -b bpool/BOOT/devuan ] ; then echo "yeah, it's a block device." ; fi | 23:58 |
SuaveDandy | It's, eh… It's a weird process. | 23:58 |
fsmithred | can you run that line? | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!