golinux | I have no idea if anything important happed either because it's mostly chest-thumping and nostalgia and I couldn't wade through it either. | 00:01 |
---|---|---|
ShorTie | good luck on Tenkawa | 00:28 |
landley | The 4g desktop DVD does not appear to have a "live" option. | 00:35 |
landley | I can mount the big dvd from the little live dvd, but can't figure out how to tell sources.list to look at an iso on /mnt. | 00:35 |
landley | I'm booting the live dvd under kvm to have a reproducibly clean system, but having to install llvm via phone tether repeatedly eats my bandwidth when I'm out and about on my laptop. | 00:36 |
rrq | landley: I think it's like "deb [trusted=yes] file:///mnt daedalus main" | 00:39 |
rrq | haven't tried it though.. just reading "man sources.list" | 00:40 |
landley | Tried that (with chimaera), not finding anything. | 00:41 |
landley | I tried apt-cdrom which does not seem to understand devuan not quite being debian for repository format reasons. | 00:41 |
rrq | not forgetting "apt-get update" of course | 00:41 |
landley | (It finds the cdrom, names it with version info, and then says it contains 3 package indexes but no packages.) | 00:41 |
landley | Hmmm... the update ran but it's still not finding stuff. | 00:42 |
landley | Possibly it expects it to be mounted on /media/cdrom or something instead of /mnt? | 00:42 |
landley | The problem is I'm running from the live cd so I stuck it into -hdb for kvm. | 00:43 |
rrq | yes, check with "df" maybe | 00:45 |
rrq | apt definitley found the pool with "deb file:///home/ralph/... daedalus main" where I mounted it | 00:46 |
landley | Should I point it at the "pool" directory? | 00:49 |
rrq | no at the directory with pool and dist as subdirectories | 00:50 |
rrq | it also needs "[trusted=yes]" to be happy about it | 00:51 |
landley | Hmmm, it may be that the "desktop" iso just doesn't have the files I want because it "installed" them into the base image? | 00:51 |
landley | So they're not in the -livecd, and not available as packages into the "desktop" CD. | 00:52 |
landley | (The only *llvm* it has under pool is libllvm, and no *clang* | 00:52 |
landley | ) | 00:52 |
rrq | the "pool1" ISO seem to have clang | 00:58 |
landley | Where do I get the pool1 iso? | 00:58 |
rrq | only for daedalus https://files.devuan.org/devuan_daedalus/installer-iso/ | 01:00 |
landley | Is daedalus out yet, or still in development? | 01:00 |
rrq | not stable yet | 01:01 |
landley | Is this a new policy that they'll have pool1 iso now, or is it just a thing they do for dev nightlies but not stable releases? | 01:01 |
rrq | I added it to the automatic iso building after daedalus had started .. could be called "new policy" I suppose | 01:04 |
rrq | probably a good idea to make a chimaera one as well | 01:04 |
landley | Hmmm, there's cd2 cd3 and cd4 in the installer dir for chimaera. Maybe I could glue them together... | 01:05 |
landley | If you could make a pool1 dvd for chimaera and point me at the URL, I would be grateful. Or possibly greatful. Never can remember how to spell it. | 01:05 |
rrq | yeah, just started .. will take some hour though | 01:06 |
landley | (I could also move the tests to run from my server at home, but re-installing packages from the mirrors regularly seems... impolite.) | 01:06 |
landley | Thanks. I'll check back later. | 01:06 |
rrq | anything other than clang you'd need? | 01:07 |
landley | Not sure? I'm hoping to download the whole "standardish" package list once so I can install stuff into a fresh VM on the fly without bothering the servers or needing fast net. | 01:09 |
landley | Currently I'm testing http://lkml.iu.edu/hypermail/linux/kernel/2202.0/01505.html variations. | 01:09 |
rrq | right. pool1 has the 5000 "most popular" packages | 01:10 |
onefang | apt-cacher, apt-cacher-ng, or apt-mirror might help you be less impolite. Never tried them myself. | 01:10 |
landley | But I also have a tiny system builder (300 line bash script that creates bootable QEMU images for various architectures) that I portability test in various environments. With gcc and llvm toolchains using glibc and musl and bionic... | 01:11 |
landley | https://github.com/landley/toybox/blob/master/scripts/mkroot.sh | 01:11 |
onefang | I had a feeling you where that Landley. B-) | 01:11 |
landley | Testing it on a system that does _not_ have autotools and libtool and such installed is part of the point. | 01:11 |
landley | Ah, I'm infamous. Good to know. | 01:11 |
onefang | I'm the infamous onefang that gave you the x486 support. But that's offtopic for here. | 01:13 |
landley | Why are there ng and non-ng versions of apt-cacher surviving? | 01:13 |
onefang | Ask Debian. | 01:13 |
landley | the thing about having hung around at the fringes of a community for over 20 years is you recognize a whole bunch of handles and have only a vague idea why. :) | 01:14 |
onefang | We still use most of the Debian packages, and only fork a small number of packages. | 01:14 |
landley | I know. I switched to devuan as my desktop... in the before times. Pre pandemic. Years have lost their measure. | 01:14 |
onefang | Most of our package mirrors just bounce non Devuan specific packages to the Debian package mirror system. | 01:15 |
landley | I remember you'd written a complicated mirroring thing that was causing scalability issues back under ascii. It seemed to clear up by beowulf. | 01:15 |
landley | But it made me reluctant to try to set up my own mirror... | 01:15 |
onefang | That was before my time. Now I'm the mirror herder. And still I haven't poked at those apt cachers and mirror modules. | 01:16 |
landley | I still want to try to bootstrap musl based systems with devuan. (And eventually toybox ones...) | 01:16 |
landley | But every time I dig down into the instructions for that it gets eldrich fast. | 01:16 |
landley | (It's on the todo list...) | 01:16 |
landley | Hmmm, I guess apt-cache or apt-mirror would work fine with the devuan repo because the tower of redirects isn't end-user visible? | 01:17 |
landley | (That's mostly a question of staying up to date with what you can and updating locally what you need to delta, I guess. Sounded extremely complicated when I looked into it, and that was long enough ago I remember zero details.) | 01:18 |
onefang | Well unless you try to keep an eye on the 302s. | 01:18 |
landley | [whoosh over my head] | 01:18 |
landley | 302 redirects? Security information? | 01:19 |
onefang | HTTP return code 302 "Go ask Debian, we don't have that file here". | 01:19 |
landley | Google is not being informative. | 01:19 |
landley | Ah. | 01:19 |
landley | I'm old enough to remember when Mepis got sued by the FSF for doing that. | 01:20 |
* landley wanders dinnerwards. | 01:21 | |
landley | Thanks for starting a pool1 ISO. I look forward to downloading it later. | 01:21 |
c0rnelius | ShorTie: Thats rich. | 02:06 |
golinux | It's true . . . | 02:10 |
golinux | <onefang> 600 messages since I went to sleep, and lots of it off topic. Anything important I should know about? I'll skip reading it all. | 02:10 |
golinux | Clogging this channel with OT is not helpful to devuan | 02:11 |
golinux | c0rnelius: ^^^ Perhpas you can have a heart to heart with him . . . | 02:12 |
golinux | Him being Tenkawa . . . | 02:13 |
golinux | Thanks. | 02:13 |
c0rnelius | golinux: I'm not the boss of Tenkawa. People say stupid shit every where. I can't control that. Is he right or wrong..? Don't know, as I'm not apart of ur argument. Aren't you the boss..? | 02:18 |
golinux | I vote for respectful colloborators | 02:19 |
c0rnelius | I vote for less dicks regardless of said gender | 02:19 |
golinux | I just want developers to have clear access to issues that might be reported here and not buried in ego-driven banter | 02:20 |
golinux | OK How about this for neutral; IT being Tenkawa | 02:21 |
golinux | And let's stop here. | 02:22 |
c0rnelius | Regardless of Tenkawa I understand and respect the comment before last... and that one. | 02:22 |
golinux | I have an equal opportunity "no OT" policy to anyone who stuffs this channel with nonsense. | 02:23 |
golinux | Thanks. | 02:24 |
c0rnelius | golinux: Perfect | 02:27 |
c0rnelius | I need to interject though, I do believe you behave like a douche sometimes... golinux. | 02:32 |
c0rnelius | I don't know why ur so mad. But you do come across as a dick. | 02:34 |
golinux | PM | 02:34 |
golinux | c0rnelius: Feel better now? | 02:37 |
c0rnelius | calm down. I don't feel anything, relax. | 02:38 |
golinux | Feelings are irrelevant. And please stop this silliness or I will call up someone with a hammer | 02:40 |
WoC | For those who don't know this DICK is an army acronym; Dedicated Infantry Combat Killer... just saying... | 03:08 |
rrq | landley: chimaera pool1 isos are now published | 04:09 |
onefang | landley logged out, but he'll be back. | 04:27 |
onefang | Anyone used nwipe (ASCII version) on an SSD? Says it's successful, half a second later. My SSD isn't that fast. lol | 06:05 |
onefang | Currently just copying /dev/zero onto it, timed so I can see how fast writing the entire SSD should take. | 06:06 |
rwp | onefang, You might run strace on it and see what it is actually doing. Seems like whatever system call it tried failed immediately but then it didn't report it to you. | 06:31 |
ejjfunky | Hi all. When I tried sudo apt update, i got this: E: The repository 'https://packages.sury.org/php chimaera Release' does not have a Release file. | 06:50 |
onefang | packages.sury.org isn't one of our official package mirrors. | 06:52 |
onefang | So ask them. | 06:52 |
ejjfunky | ahh.. right. thanks | 06:53 |
ejjfunky | i tried installing it to get to php 8. but it didnt work. | 06:55 |
onefang | Or use an official package mirror. https://pkgmaster.devuan.org/mirror_list.txt lists them, http://veritas.devuan.org/apt-panopticon/results/Report-web.html tests them every ten minutes. | 06:55 |
ejjfunky | onefang, got it. | 06:55 |
onefang | Oh, that's arepo for PHP. It might not have Devuan release code names. | 06:55 |
ejjfunky | onefang, it looks like it. | 06:56 |
ejjfunky | i ended up using daedalus repo and it worked. i got my php 8 | 06:56 |
golinux | ejjfunky: Check this out for sury: https://dev1galaxy.org/viewtopic.php?id=3331 | 06:56 |
ejjfunky | hah, systemd creeping up everywhere! | 06:57 |
bb|hcb | It doesn't. Also I remember having problems with their packages back in Debian woody/jessie... | 06:58 |
golinux | tdrnetworks has been providing cleaned packages for years | 06:58 |
golinux | Last update here: https://dev1galaxy.org/viewtopic.php?pid=34499#p34499 | 06:59 |
bb|hcb | ejjfunky: Going to daedalus is much more stable than using 3rd party packages | 06:59 |
golinux | 2022-02-08 14:05:34 | 06:59 |
ejjfunky | bb|hcb, ic. cool! | 07:00 |
golinux | They have been providing packages since jessie.l | 07:00 |
ejjfunky | golinux, ic. | 07:00 |
ejjfunky | i upgraded it through chimaera. running pretty well so far. | 07:01 |
golinux | Your decision, of course | 07:01 |
ejjfunky | golinux, yea. im still learning and trying things out. | 07:01 |
ejjfunky | im installing java on devuan but i dont want it to run by default. i want to turn it on only when i need java. how do i do this in devuan? | 07:05 |
onefang | Java is a programming language. Things written in Java might run, but not Java itself. | 07:10 |
ejjfunky | onefang, got it. i just worry about the jvm running in the background. but i'll see once i get it installed. | 07:14 |
gnarface | ejjfunky: i don't use java but i also am pretty sure nothing stock starts the jvm by default | 07:19 |
ejjfunky | gnarface: i guess im used to running systemd and snap and see applications run by default | 07:20 |
gnarface | ejjfunky: what i mean is i'm pretty sure nothing starts it up at boot time. i'm sure there's all kinds of bad ideas floating around that might but snapd is also not in devuan | 07:22 |
gnarface | ejjfunky: it's actually super easy to check though; it would have to have put a script in /etc/init.d/ | 07:23 |
ejjfunky | gnarface: download is slow though, averaging 98 - 123k. how do i make downloads faster? | 07:23 |
gnarface | ejjfunky: be in the UK | 07:23 |
ejjfunky | gnarface, ic. thanks. will check on it | 07:23 |
gnarface | ejjfunky: oh wait, one of the mirrors might be faster than the others for US users, you could dial it in directly instead of using the dns round-robin (deb.devuan.org is a dns round-robin, so if you just retry, sometimes the situation will change immediately) | 07:24 |
gnarface | ejjfunky: lemme see, i think it was sledjhamr.org | 07:24 |
gnarface | ejjfunky: lots of us on this continent are having problems with the cross-connects to the others but it has been sporadic, i would recommend you run a caching proxy if you have multiple machines to update | 07:25 |
ejjfunky | gnarface, ahh ok. | 07:26 |
onefang | sledjhamr.org is my mirror, it's in Amsterdam. | 07:26 |
rwp | I am using apt-cacher-ng on a ceres unstable system here as a caching proxy and that works pretty well. | 07:26 |
ejjfunky | ok, will try to use that after this download. | 07:27 |
onefang | That list I linked to before includes the country the mirror is in. The apt-panopticon results include speeds. | 07:27 |
ejjfunky | rwp: ic. | 07:27 |
rwp | About once every 4-6 months it burps and hiccups and complains. I purge it and re-install it and then it is okay for another 4-6 months. | 07:27 |
rwp | I have tried all of the other proxies and I think apt-cacher-ng is the lessor of the evils. | 07:27 |
ejjfunky | rwp, ic. | 07:28 |
ejjfunky | rwp, ok. thanks. | 07:28 |
gnarface | rwp: i think on the admin console you can just purge the cache to fix that timetamp issue | 07:28 |
onefang | Often people say my mirror is quite fast, but those people might be in Europe. B-) | 07:28 |
gnarface | rwp: i don't think you have to reinstall the whole thing. for me, sometimes it clears itself after a while. | 07:28 |
rwp | gnarface, It's probably possible to fix and recover but I am mostly emphasizing that it's not too much trouble regardless. | 07:31 |
rwp | For me a US mirror is definitely better than an EU or UK mirror. :-) | 07:31 |
onefang | We do need some USA package mirrors. Convince someone to run one. B-) | 07:40 |
rwp | I see both Princeton and Berkeley in the mirror list. https://www.devuan.org/get-devuan | 07:41 |
rwp | I usually try for Berkeley since it is http and Princeton is only https. Belltower is also listed in the US. But we definitely could use more. | 07:43 |
onefang | And in the apt-panopticon results for Berkeley you'll see 13.49% updated. | 07:43 |
onefang | That mirror list you linked is the ISO mirror,s not the package mirrors. | 07:44 |
onefang | https://pkgmaster.devuan.org/mirror_list.txt is the package mirrors I linked before. | 07:44 |
rwp | Ah... I see. | 07:46 |
rwp | So that still includes Berkeley and beard.ly (which I didn't now about) in the US. | 07:50 |
onefang | 20.14% updated for beard.ly, so slightly better. Tends to time out a lot though. | 07:51 |
onefang | In general, according to apt-panopticon, the USA package mirrors tend to be the worst. We need a good mirror there. | 07:52 |
onefang | devuan.c3sl.ufpr.br might work well for USA for now. | 07:52 |
rwp | Is there a page for the mirror update and sync status? | 07:55 |
onefang | Or Like that apt-panopticon I keep mentioning? B-) | 07:55 |
rwp | I have no idea what apt-panopticon is and my web search fu and apt-cache search fu is failing me. | 07:55 |
onefang | https://sledjhamr.org/apt-panopticon/results/Report-web.html http://veritas.devuan.org/apt-panopticon/results/Report-web.html and https://borta.devuan.dev/apt-panopticon/results/Report-web.html that are checking from Netherlands, France, and Sydney respectively. | 07:56 |
onefang | Something I wrote to check on the health of our package mirrors. | 07:56 |
onefang | Take the speed measurement of sledjhamr on the sledjhamr instance with a big grain of salt. lol | 07:58 |
rwp | Thanks for those links. I will save them off. | 08:00 |
rwp | Well... For being in the backwater that is Colorado I don't seem to have any actual practical problems with the mirrors. | 08:00 |
onefang | To directly answer you question. A yellow warning in the Updated column means that mirror isn't fully up to date, but is still within it's sync window. Red error means it's outside of that sync window. | 08:00 |
rwp | But I run a cron.daily on all of the systems that will either upgrade or download pending upgrades. And use a central apt-cacher-ng proxy. So that probably hides problems and smooths things out. | 08:01 |
onefang | And green OK means it's fully up to date. | 08:01 |
ev_m | hello, need help finding installing manpages for gcc. uninstalling and reinstalling gcc doesn't help. found sth on gcc-doc, but it says no package to install | 10:47 |
rrq | appparently the actual documentaion is "non-free" .. gcc-9-doc or gcc-10-doc | 11:06 |
rrq | whereas the virtual package is "contrib" | 11:07 |
ev_m | rrq: does that mean i need to enable contrib and non-free in sources to access it? | 11:19 |
rrq | yes, that would do it | 11:20 |
ev_m | that did it. thanks | 11:22 |
blastwave | on the outside chance anyone was curious the NVidia drivers and the CUDA dev tools seem to work just fine on devuan daedalus. thus far | 11:54 |
Vall | blastwave> on the outside chance anyone was curious the NVidia drivers and the CUDA dev tools seem to work just fine on devuan daedalus. -> glad to hear it, I will be soon migrating my laptop from Ubuntu to Devuan and support for its NVidia GPU was a major concern | 15:26 |
blastwave | Vall: the real trick is to get the driver issues sorted out in advance | 15:27 |
blastwave | Vall: before you do anything, on a laptop, what does lspci say for the NVidia device ? | 15:28 |
blastwave | Vall: you may see something like --> 01:00.0 VGA compatible controller: NVIDIA Corporation GK107GLM [Quadro K1100M] (rev a1) | 15:29 |
blastwave | Vall: the laptop mobile quadro chips are a real concern as the NVidia corporate wonks are dropping support ALL over the place for those | 15:30 |
blastwave | marsen: good day | 15:30 |
marsen | blastwave, hey o/ | 15:31 |
blastwave | marsen: the devuan machine has no problems with CUDA and the NVidia profilers | 15:31 |
Vall | back | 15:52 |
Vall | blastwave: 01:00.0 VGA compatible controller: NVIDIA Corporation GP104GLM [Quadro P5000 Mobile] (rev a1) | 15:53 |
Vall | works perfectly with the nvidia-*-470 Ubuntu packages | 15:54 |
Vall | I think it will probably work great with Devuan Chimaera too | 15:55 |
Vall | blastwave> NVidia corporate wonks are dropping support ALL over the place for those [mobile Quadro chips] -> I haven't heard of it! Any reference/URL I can read up on that? | 15:56 |
blastwave | Vall: hold on a sec ... I am juggling a machine here | 16:24 |
blastwave | Vall: that must be some snazzy laptop. | 16:33 |
blastwave | Vall: turns out the driver released yesterday version 510.54 supports that chip | 16:34 |
blastwave | Vall: see https://www.nvidia.com/Download/driverResults.aspx/186996/en-us | 16:35 |
blastwave | your next problem is CUDA | 16:35 |
blastwave | funny .. that driver supports the Quadro M6000 but not the K6000 | 16:38 |
Vall | blastwave> Vall: that must be some snazzy laptop. -> yeah! it's my pride and joy: https://support.hp.com/gb-en/product/hp-zbook-17-g4-mobile-workstation/14840018/document/c05482346#AbT0 | 16:39 |
Vall | (mine is the model with Xeon processor -- ECC RAM! -- and the Quadro P5000) | 16:39 |
blastwave | something top of the line I guess | 16:39 |
blastwave | you are in luck | 16:40 |
blastwave | see https://docs.nvidia.com/cuda/cuda-toolkit-release-notes/index.html | 16:40 |
blastwave | CUDA 11.6 GA works with driver 510.39.01 or higher | 16:40 |
Vall | yeah, it was the absolute best I was able to buy with my budget at the time. Interestingly, it was not too expensive -- I bought it refurbished with 1 year of use, but it was practically in mint state -- and HP had just replaced the G4 line with the G5. | 16:41 |
Vall | Still had 1yr of warranty in it, which I fortunately haven | 16:41 |
blastwave | I am curious what nvidia-smi has to say about that chip | 16:41 |
Vall | *haven't had the opportunity to use | 16:41 |
blastwave | don't expect great performance from that GPU. It will work but it won't do great piles of CUDA tasks | 16:42 |
Vall | blastwave: https://0x0.st/o8Az.txt | 16:43 |
WoC | nice | 16:44 |
Vall | blastwave> don't expect great performance from that GPU. -> enough for my uses (some CUDA like f@h, but mainly games -- Doom Eternal runs perfectly at 1080p) | 16:44 |
blastwave | I am really baffled. the GPU has 16G of its own memory ? | 16:44 |
WoC | try blender ? | 16:44 |
Vall | WoC: never used blender | 16:45 |
WoC | it has cuda | 16:45 |
Vall | blastwave: yep, 16GB of in-GPU RAM happiness | 16:45 |
WoC | :) | 16:45 |
blastwave | Vall: do you have nvcc also ? | 16:45 |
blastwave | try nvcc --version | 16:46 |
Vall | blastwave: nvcc not installed, installing it right away as part of nvidia-cuda-toolkit, wait one | 16:47 |
blastwave | okay | 16:47 |
blastwave | oh .. you are using the default packages. fine | 16:47 |
blastwave | I build my own kernel and compiled the nvidia drivers and then installed the CUDA packages from NVidia | 16:47 |
blastwave | however that was all part of the test process | 16:48 |
Vall | blastwave: https://0x0.st/o8AX.txt | 16:51 |
blastwave | release 9.1 ??? | 16:51 |
blastwave | wtf | 16:51 |
Vall | blastwave: yeah, default kernel and most packages. I customized very little stuff in my machine these days | 16:51 |
Vall | too old? | 16:51 |
blastwave | very | 16:51 |
Vall | That's what's in the repo. | 16:52 |
blastwave | things move fast and the 2017 release | 16:52 |
blastwave | geez | 16:52 |
Vall | But then it's still Ubuntu *18.04*... | 16:52 |
blastwave | yeah well that is why most people that do this sort of thing with a new machine have to get the packages directly from NVidia | 16:52 |
Vall | That's one reason I'm upgrading to Devuan Chimaera | 16:52 |
blastwave | I installed daedalus for the test | 16:53 |
blastwave | see https://git.sr.ht/~blastwave/bw/blob/bw/nvidia_cuda/vector_add/readme.k6000 | 16:53 |
blastwave | Linux dev 5.17.0-rc3 #1 SMP PREEMPT Sat Feb 12 01:19:29 UTC 2022 x86_64 GNU/Linux | 16:53 |
Vall | wife calling... chores waiting... will be back later (sorry for leaving abruptly) | 16:53 |
blastwave | no problem ... I am dealing with confused people who need an IBM Informix database restored from tape | 16:53 |
RhineDevil | I created a virtual interface for LXC with appropriate rules. Everything's fine, but when I start a container, the system takes the virtual interface as default route, and this breaks my internet connection | 16:55 |
RhineDevil | Could someone help me? | 16:56 |
buZz | RhineDevil: disable the route? | 16:57 |
buZz | wait, the -system- ? whats the 'system' | 16:57 |
RhineDevil | buZz: what do you mean by disable the route? it shouldn't be here in first place, how can I prevent whatever puts the route from inserting it? ): | 17:23 |
buZz | inserting it where | 17:23 |
RhineDevil | buZz: the system is the devuan installation in his entirety | 17:23 |
buZz | what 'system' has no internet | 17:23 |
RhineDevil | inserting it in the routing table | 17:23 |
buZz | the host? the lxc? a random computer on your lan? | 17:24 |
RhineDevil | the system I'm using | 17:24 |
RhineDevil | the devuan installation | 17:24 |
buZz | the host? the lxc? a random computer on your lan? | 17:24 |
RhineDevil | the host | 17:24 |
buZz | eh | 17:24 |
buZz | maybe just dont overlap your ip ranges ; | 17:25 |
buZz | ;) | 17:25 |
buZz | pastebin your full route -n perhaps? | 17:25 |
buZz | starting a lxc should not affect the default routes (aka , the one route for 0.0.0.0) on your -host- | 17:26 |
RhineDevil | buZz: this is the malfunction status, in normal status the first route is missing https://paste.debian.net/hidden/dc29441b/ | 17:28 |
RhineDevil | system creates a veth interface for the container called `veth3ffF3' or something like that, and the first route appears | 17:29 |
RhineDevil | *host, sorry | 17:30 |
buZz | bizar, i have no clue why its considering your lxc a router, makes no sense | 17:30 |
RhineDevil | no idea either. I've also forced NetworkManager to consider them as unmanaged | 17:31 |
RhineDevil | but either way I need to solve this | 17:32 |
RhineDevil | otherwise no lxc and no party | 17:32 |
RhineDevil | so please whatever you can buZz throw it down | 17:34 |
RhineDevil | *whatever idea you have | 17:34 |
buZz | how are you starting the lxc? | 17:34 |
RhineDevil | buZz: lxc-start -n mycontainer | 17:34 |
buZz | maybe pastebin the lxc config? | 17:35 |
buZz | fyi, ever tried proxmox? :) | 17:35 |
RhineDevil | buZz: https://paste.debian.net/hidden/baf4d99c/ | 17:36 |
RhineDevil | proxmox has systemd and it isn't nice to use as workstation too | 17:37 |
buZz | i wouldnt recommend it as a workstation , no, its a server os | 17:37 |
buZz | RhineDevil: ahhh, i think i found some cause | 17:40 |
buZz | RhineDevil: if the LXC os is running a dhcp server, and the host is running a dhcp client, that might cause the host to grab a ip and route by itself | 17:40 |
buZz | hint from https://discuss.linuxcontainers.org/t/lxd-host-default-route-changed/10473/5 | 17:42 |
RhineDevil | buZz: strange, it didn't happen before | 17:47 |
RhineDevil | like, I made containers on beowulf and I never had that issue, then on chimaera suddenly happens | 17:48 |
buZz | maybe you didnt install a dhcp -server- on a lxc prior :) | 17:49 |
RhineDevil | buZz: nah I'm pretty sure I did | 17:49 |
buZz | tried removing that dhcp server yet? or eh , maybe make the dhcp CLIENT on your host only listen on eth0 | 17:49 |
buZz | instead of all interfaces | 17:49 |
RhineDevil | I think you gave me an idea buZz | 17:50 |
buZz | yay! | 17:50 |
RhineDevil | I'll try setting a static ip for the container | 17:50 |
buZz | that'll be 10 BTC plz | 17:50 |
buZz | :D | 17:51 |
buZz | ah hehe | 17:51 |
buZz | never a bad idea :) | 17:51 |
RhineDevil | and then figure out what happens | 17:51 |
RhineDevil | the container won't try to mess up things | 17:51 |
RhineDevil | so if I'll listen on the interface I can figure out what happens | 17:51 |
buZz | well, i'd just get into the lxc and check 'ps aux | grep dhcp' | 17:51 |
buZz | murder the dhcp server ;) | 17:51 |
RhineDevil | buZz: I like the idea of clients not having an idea of what happens and having the dhcp server telling them what to do, sorry but no | 17:52 |
RhineDevil | I just need to figure out the node that makes it all a mess | 17:53 |
RhineDevil | tiresome... | 17:53 |
buZz | why not listen to what i'm saying? the -OS INSIDE THE LXC- is probably running a DHCP SERVER | 17:53 |
buZz | which you dont need | 17:53 |
RhineDevil | buZz: I know it's running a dhcp server, I installed it on the host LOL | 17:53 |
RhineDevil | and that's exactly because I want containers to shut up and autoconfig | 17:54 |
buZz | fuck sake | 17:54 |
buZz | THE OS INSIDE THE LXC | 17:54 |
buZz | NOT THE HOST | 17:54 |
RhineDevil | oh uhm | 17:54 |
buZz | fuuuuck sake man | 17:54 |
RhineDevil | sorry I'm stupid | 17:54 |
RhineDevil | Guess I'll start toying around and figure it out | 17:55 |
buZz | :P | 18:00 |
RhineDevil | buZz: I properly investigated with wireshark and other stuff what was happening. There is no server in the lxc container, but something weird does happen. A client with a weird mac address (not present neither in the host's ifconfig or container's ifconfig) asks for a lease even BEFORE the dhcp client of the container starts | 18:39 |
RhineDevil | So this client gets the first lease and the container's client get the second | 18:40 |
RhineDevil | Then gajim and other stuff start listening on the address this dhcp client obtained | 18:41 |
buZz | check 'arp -a' perhaps to find what that address is? | 18:41 |
RhineDevil | weird stuff | 18:41 |
buZz | whats a gajim? are you running that? | 18:41 |
buZz | looks like a desktop thingy? | 18:41 |
RhineDevil | yes, on the host. an XMPP client | 18:41 |
RhineDevil | yes | 18:41 |
buZz | where did the dhcp lease originate from? | 18:42 |
buZz | the 'weird mac address' one | 18:42 |
RhineDevil | I have no idea D: | 18:42 |
buZz | didnt wireshark show the lease ? | 18:42 |
RhineDevil | yes | 18:42 |
buZz | what answer came to that question? | 18:42 |
buZz | from what ip/mac | 18:43 |
RhineDevil | the server accepted it | 18:43 |
buZz | 'the server' isnt a ip or mac | 18:43 |
RhineDevil | buZz: the isc-dhcp-server sitting on my virtual interface accepted it and gave the lease | 18:44 |
buZz | how about this , on your HOST , in /var/lib/dhcpcd/ , there should be files describing the lease | 18:44 |
buZz | incl the IP the dhcp server is running on | 18:44 |
buZz | look at the IP of that , thats the host that you need to murder the dhcp server on | 18:44 |
RhineDevil | buZz: the server still is the one I installed on the host, and it both accepted the weird mac address and the container's mac address | 18:47 |
RhineDevil | how do I decipher lease? it isn't readable | 18:47 |
buZz | oh, right, that dhcp client :P eh , dhcpcd -U interfacethatleaseison | 18:49 |
RhineDevil | dhcpd asked for that lease | 18:50 |
RhineDevil | I have no idea why it used that weird address | 18:51 |
RhineDevil | *that weird mac address | 18:51 |
buZz | dhcpcd , i assume | 18:52 |
RhineDevil | How could I tell dhcpcd to FUCKING STOP asking servers for leases on this interface? | 18:52 |
buZz | did you check arp -a on host yet? | 18:52 |
RhineDevil | yeah that | 18:52 |
buZz | by not having dhcpcd even do those interfaces? | 18:52 |
RhineDevil | only shows the lease I have with the wireless router | 18:52 |
buZz | wrong interface, i bet ;) | 18:52 |
buZz | do virbr0 | 18:52 |
RhineDevil | buZz: told dhcpcd to ignore the interface. Good news, internet doesn't crash, bad news, container is still nuts | 18:57 |
buZz | good :) progress | 18:57 |
RhineDevil | buZz: yeah guess so | 18:57 |
buZz | define 'nuts' ? | 18:57 |
RhineDevil | buZz: apt-update says "Ignored" on every stuff I download | 18:58 |
RhineDevil | not good | 18:58 |
buZz | does internet work on the container | 18:59 |
RhineDevil | buZz: sorry had a connection issue | 19:04 |
buZz | yeah several :P | 19:06 |
buZz | seems a theme ^_^ | 19:06 |
buZz | RhineDevil: and? :) does internet work on the container? | 19:16 |
RhineDevil | buZz: no. I've managed to fix firewall rules in order to get proper masquerading and forwarding between private and external interface, but packets don't return | 20:29 |
buZz | sounds like they arent proper then ;) | 20:29 |
buZz | RhineDevil: you know, stuff like proxmox can do -all that- for you ;) | 20:30 |
buZz | even works in virtualbox (for LXCs anyway, although KVM might work nowadays aswell) | 20:30 |
RhineDevil | buZz: they're the same rules used in /usr/libexec/lxc/lxc-net | 20:31 |
buZz | i dont have that file, so dont know what you mean by that (no i dont need to see em) | 20:31 |
buZz | RhineDevil: there is ##networking aswell for just general firewall/nat rule help | 20:32 |
buZz | i am convinced anywhere in there will know more about nat rules than me :D | 20:32 |
RhineDevil | buZz: I've set input and forwarding default policy as accept but still no packets being forwarded to my internal interface | 20:33 |
RhineDevil | by the way you can get this file by installing lxc package | 20:34 |
buZz | i figured | 20:34 |
buZz | but why would they even be needed to be set by -you- if those are already set by lxc? | 20:35 |
buZz | lxd* | 20:35 |
buZz | (i assume) | 20:35 |
RhineDevil | no, only lxc | 20:36 |
RhineDevil | well I don't like daemons coming up and doing their stuff so I made a static config | 20:37 |
RhineDevil | buZz: the weird thing is, I see the packet being changed for going back into the virtual interface, but it doesn't arrive on the target interface | 20:39 |
RhineDevil | buZz: seems it apparently was a netfilter problem. I thought my rules perfectly replicated the rules necessary for packet forwarding, but they don't. Gonna do some trial and error | 22:30 |
RhineDevil | Yeah if you haven't wondered I'm kinda stubborn | 22:32 |
buZz | ;) | 22:35 |
buZz | maybe should be RhinoDevil | 22:35 |
buZz | :D | 22:35 |
RhineDevil | buZz: lmao | 22:38 |
RhineDevil | now works btw | 22:38 |
buZz | \o/ huzzah! | 22:38 |
buZz | time to test some mines | 22:38 |
RhineDevil | buZz: how do you know it's a container for minetest | 22:38 |
RhineDevil | how | 22:38 |
RhineDevil | h o w | 22:38 |
* buZz is clairvoyant | 22:39 | |
buZz | also its on https://paste.debian.net/hidden/baf4d99c/ | 22:39 |
RhineDevil | buZz: embarassing | 22:40 |
RhineDevil | I outlined my cards without even knowing | 22:40 |
buZz | haha well | 22:40 |
buZz | at least its not a wifi password :P | 22:40 |
RhineDevil | btw buZz wanna know something funny | 22:40 |
RhineDevil | I'm using custom lxc templates, and by custom I mean they're from the lxc repo | 22:41 |
RhineDevil | buZz: https://github.com/lxc/lxc-templates I even bothered harmonizing everything and debian dudes didn't even bother adopting it and keeping it updated | 22:42 |
RhineDevil | commits from 2 yr ago | 22:43 |
buZz | well, you do know debian/devuan isnt a rolling release? :) | 22:45 |
buZz | its never the latest and greatest version, its a fixed version for a release, only security patches can come in, but no api changes etc | 22:45 |
RhineDevil | buZz: of course I know, but since these commits a major release has passed ;) | 22:46 |
buZz | right, but that doesnt affect debian's release model | 22:47 |
RhineDevil | buZz: wdym? these commits where already present when their equivalent was in sid | 22:48 |
RhineDevil | *were | 22:48 |
buZz | sid is unstable, not considered a debian release | 22:49 |
buZz | or testing even? hmm | 22:49 |
RhineDevil | buZz: sid and testing are almost the same thing | 22:50 |
buZz | right > a rolling development version of the Debian distribution containing the latest packages that have been introduced into Debian | 22:50 |
buZz | so you could help them push newer lxc stuff in | 22:50 |
buZz | you might just have to update -all- of em just to update that lxc-templates ;) | 22:51 |
RhineDevil | buZz: I emailed Pierre Elliot Becue at the time (lxc-templates mantainer) I'm still waiting an answer lmao | 22:51 |
RhineDevil | (the time was 2 years ago, when these commits were merged) | 22:51 |
buZz | probably better to use the MLs or whatever organisational method they use | 22:52 |
RhineDevil | well at this point I won't bother anymore, since in two years they didn't even bother updating their stuff | 22:58 |
DRX | Is anyone else having trouble resolving deb.devuan.org? | 23:45 |
golinux | DRX: Yes. Some glitches being resolved atm | 23:48 |
DRX | OK, thx. I couldn't resolve it via several different providers, so I thought something must be up. | 23:49 |
golinux | Yeah . . . stuff happens . . . | 23:49 |
DRX | DNS now resolves deb.devuan.org. All better now; thx. | 23:54 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!