| fsmithred | confirmed: plain 686 kernel sees ass 2G ram. -pae is not needed, jjakob rwp | 00:16 |
|---|---|---|
| jjakob | well I already selected the -pae kernel, now I'm trying to undo my stupidity again as I selected GPT and now grub won't install as there's no partition marked bootable | 00:16 |
| jjakob | I dunno if I should try figuring out how to set a bootable flag on a partition in GPT or just replace it with a MBR table | 00:17 |
| fsmithred | not bootable, you need a special partition for gpt with mbr | 00:17 |
| fsmithred | if you don't need more than 4 partitions, mbr makes sense | 00:18 |
| jjakob | special partition? I just created one partition, and set its type to "physical volume for encryption" | 00:18 |
| fsmithred | but gpt/legacay bios boot needs 1M partition with no filesystem | 00:18 |
| jjakob | then it's LUKS inside that, then LVM, then a root and a swap LV | 00:18 |
| fsmithred | and flag ef02 (gdisk) | 00:18 |
| fsmithred | or bios_grub in gparted | 00:18 |
| jjakob | I already had to set GRUB_ENABLE_CRYPTODISK=y manually | 00:19 |
| fsmithred | really 1MB | 00:19 |
| fsmithred | what you did should work for msdos partition table | 00:19 |
| jjakob | that's what I usually do for MBR, and it works there, just now I decided to try GPT for the heck of it. | 00:20 |
| fsmithred | that's a lotta layers for EEE | 00:20 |
| fsmithred | well, unless you left a tiny bit of free space to make a new partition, you'll have to redo it | 00:20 |
| jjakob | I think it left 1MB at the start free | 00:21 |
| fsmithred | I don't know if you can use that. I'd be interested to hear how it works. | 00:21 |
| jjakob | I want to try root on encrypted zfs dataset but I have no idea if devuan has a zfs installer | 00:23 |
| fsmithred | you might find instructions for zfs at the forum | 00:24 |
| fsmithred | mason might have his own installer | 00:25 |
| jjakob | I have servers with encrypted zvols that auto unlock with systemd but I haven't tried without systemd yet | 00:28 |
| jjakob | there's no gdisk in the mini iso, I downloaded fdisk and parted in the installer modules, probably with parted somehow but where to put the start of the partition? | 00:31 |
| rrq | mmm first partition usually starts at sector 2048 nowadays | 00:36 |
| fsmithred | that's probably where the first partition already starts. We were looking for a place to add a bios_gtub partition | 00:38 |
| fsmithred | is there a way to convert gpt to msdos without having to reinstall? | 00:40 |
| rrq | right I think an MBR partition table allows a sector start at 64, but GPT wants 2048 | 00:40 |
| rrq | if you copy all the numbers it should be fine with fdisk | 00:41 |
| fsmithred | the extra partition is only needed for gpt | 00:41 |
| rrq | (e.g.) | 00:42 |
| fsmithred | oh... | 00:42 |
| fsmithred | that just sunk in. I recall doing something like this with sfdisk a couple decades ago. | 00:43 |
| rrq | yes the partition tabls are just some data tables in the first few blocks (though GPT also has a "backup table" at the very end) | 00:44 |
| rrq | nothing intruding into the actual partitions | 00:44 |
| jjakob | I did this, grub loads and asks me for the luks password but then says invalid passphrase | 00:45 |
| jjakob | in parted: disk_set pmbr_boot on; mkpart grub 34s 2047s; set 2 bios_grub on | 00:46 |
| rrq | I was just about th\o add that some software do ugly things in the first few blocks too; I'm not sure what luks does. | 00:46 |
| jjakob | also says "no such cryptodisk found" | 00:46 |
| jjakob | luks is confined to the partition it's in, starting at 2048 | 00:47 |
| jjakob | rebooting to chroot back in | 00:47 |
| rrq | right. should be ok; you have verefied the start and end nmbers I gues hmm | 00:48 |
| rrq | the partition type? | 00:48 |
| jjakob | I was unsure about the start sector but 34 seems to work, found it here https://unix.stackexchange.com/questions/679971/how-to-create-a-grub-bios-boot-partition-using-sectors-34-to-2047 | 00:48 |
| jjakob | of what, the luks partition? | 00:49 |
| rrq | yes | 00:49 |
| jjakob | hmm whatever d-i picked, can't remember | 00:49 |
| rrq | what type is it now? | 00:50 |
| rrq | maybe someone with a luks partition that can tell what the partition type is for them? | 00:51 |
| fsmithred | what do you need? | 00:51 |
| rrq | the partition type | 00:52 |
| fsmithred | where do I find that? | 00:53 |
| fsmithred | my brain is not working right now | 00:53 |
| fsmithred | df -Th? | 00:53 |
| fsmithred | fdisk -l ? | 00:53 |
| rrq | fdisk -l is fine ... the "id" column | 00:54 |
| fsmithred | linux | 00:54 |
| fsmithred | 83 | 00:54 |
| rrq | ok hmm I thought there were a "Linux LUKS" type | 00:54 |
| jjakob | nope, not in MBR | 00:55 |
| jjakob | gdisk or parted show the same? you do have GPT, not MBR? | 00:55 |
| rrq | fine. I'm wrong on that | 00:56 |
| fsmithred | gpt machine is asleep | 00:56 |
| fsmithred | but I think it's just the same (linux 83) | 00:56 |
| fsmithred | oh that one is not luks | 00:57 |
| rrq | according to the bokk (wikipedia) luks wants type 0xe8 rather than 0x83 | 01:01 |
| rrq | https://en.wikipedia.org/wiki/Partition_type | 01:02 |
| rrq | worth a try I guess | 01:02 |
| jjakob | it says 8300 | 01:03 |
| jjakob | changed it to 8309, now fdisk says it's unknown | 01:05 |
| fsmithred | did you run grub-install again after creating the new partition? | 01:05 |
| rrq | why 0x8309 ? | 01:05 |
| jjakob | gdisk said that's the type for LUKS | 01:07 |
| jjakob | yeah I reran grub-install and update-grub, didn't change anything, still get invalid passhprase | 01:07 |
| jjakob | weird that it finds the right luks volume with the correct uuid and tries to open it | 01:08 |
| jjakob | grub.cfg had insmod luks2 | 01:08 |
| jjakob | probably wrong pbkdf or something that grub doesn't support yet https://wiki.archlinux.org/title/GRUB#LUKS2 | 01:10 |
| jjakob | booting into rescue again, seems like luksConvertKey should make this easy | 01:11 |
| rrq | sounds good :) | 01:19 |
| jjakob | still no luck | 01:35 |
| rrq | in the worst of worlds the luks password encryption is sensitive to the partition table data or location | 01:36 |
| jjakob | I think it's just GRUB not knowing how to decrypt LUKS2. | 01:39 |
| jjakob | that archwiki page says it partially works in grub 2.06 but some bugs were fixed later | 01:39 |
| jjakob | I did that luksConvertKey and it successfully converted it to pbkdf2 | 01:42 |
| jjakob | cryptsetup can convert from luks 2 to 1 now? I need to try that, tomorrow | 01:45 |
| mason | jjakob: I'm largely AFK today but scanning the conversation, something I find useful here is the keyutils package. With it, you can set groups of disks that share a key, so you unlock the key once and it'll unlock the rest. Example from local systems: https://bpa.st/LD5Q | 02:53 |
| mason | As for an installer, it's more just a script that drives debootstrap, although mmdebstrap seems the better choice rolling forward, so I might migrate at some point. It sets up LUKS and ZFS atop it and then runs debootstrap to populate. | 02:54 |
| onefang | I switched to mmdebstrap and I'm happy I did. | 03:35 |
| onefang | mason: ^^ | 03:35 |
| onefang | Do ALL the web browsers need pulse for microphones to work? Can't use ALSA or JACK? | 04:27 |
| onefang | Ah I can get firefox to use apulse apparently. | 04:35 |
| rwp | I normally run "apulse firefox" on my laptop. Works great. Recommended. | 04:48 |
| onefang | Progress, now it's actually asking for permission to use my microphone, but none of the microphone test sites hear it. | 05:01 |
| rwp | It's traditional that there are at least 37 mute controls in series and all must be enabled before anything can be heard through them all. | 05:17 |
| onefang | lol | 05:17 |
| onefang | Damn, I've only found 5 of them. | 05:18 |
| mason | onefang: Firefox can work with straight ALSA. | 05:19 |
| mason | onefang: https://bpa.st/GLLA | 05:19 |
| onefang | Including firefox-esr? | 05:20 |
| mason | onefang: If you say "aplay -l" or "arecord -l" you'll get the right numbers, should you need such a thing. | 05:20 |
| mason | yes | 05:20 |
| mason | At least, last time I tried, which I should repeat so I can continue saying this assertively. | 05:21 |
| onefang | I did the arecord thing to check ALSA was hearing my mic, it is. | 05:21 |
| * onefang waits for you to repeat. | 05:22 | |
| mason | hm, webrtc.org has evidently gone away | 05:23 |
| mason | Is there a useful test site? | 05:23 |
| onefang | That was gonna be my next question. A useful microphone test site, coz maybe the ones DDG found are useless? | 05:24 |
| mason | In the not-distant past test.webrtc.org was useful, but not any more. | 05:24 |
| mason | Ah, I have an idea. | 05:25 |
| mason | onefang: I'm going to have to dig into this. arecord can use the microphone as my user, but Firefox cannot, using the Devuan jitsi as a test case. | 05:30 |
| mason | I tried on my laptop, which has a simpler sound set-up. | 05:31 |
| gnarface | apulse has mic support now? that's news to me | 05:31 |
| onefang | man apulse - "dsnoop plugin allow multiple applications to capture from a single microphone;" | 05:32 |
| rwp | mason, Debian's compile of Firefox enables ALSA okay. Upstream Mozilla Firefox compiles it with ALSA disabled. So it all depends upon which version one is using. | 05:32 |
| onefang | I'm using Chimaera Devuan Firefox-ESR. | 05:32 |
| mason | rwp: I'm curious at it not finding my microphone. It's the Debian build, and output certainly works fine with bare ALSA. | 05:34 |
| gnarface | was device specified when alsa was tested? | 05:34 |
| gnarface | firefox probably isn't smart enough to find the mic unless it's on 0,0 | 05:34 |
| rwp | onefang, The Debian package works with ALSA okay. Should work. | 05:35 |
| onefang | With apulse it asks for permission to use the microphone, without it no asking. | 05:35 |
| rwp | mason, Hmm... I might try an alternative such as "mumble" (a voip voice conference program, I use it weekly) and see if the setup wizard there hears your microphone. | 05:35 |
| mason | Ah, I used to use Mumble a decade or so agfo. | 05:36 |
| mason | ago* - good idea | 05:36 |
| onefang | Think I've worked for the company that made Mumble. | 05:36 |
| rwp | onefang, The package version should work with either alsa or pulseaudio so running it with apulse would just cause it to use that interface. But that's a clue. Maybe something on your alsa is not yet enabled. | 05:36 |
| onefang | My .asoundrc doesn't have this dsnoop thing in it. | 05:39 |
| onefang | It does have dmix. | 05:41 |
| rwp | I have never needed to set up a .asoundrc file before. | 05:44 |
| mason | I use it on my workstation because input is USB. | 05:44 |
| onefang | I have complex sound setup and use JACK for most things. Things like my graphics card has 6 sound output devices, though only four video output plugs. Motherboard has two sound chips. I use two USB devices for my music. | 05:46 |
| onefang | I'm trying a different approach. using the second option on this page https://jackaudio.org/faq/routing_alsa.html any ALSA application is now being routed via JACK, and the inputs and outputs get routed automatically. | 06:23 |
| onefang | Plus I can route multiple things at once, coz it's JACK. B-) | 06:24 |
| onefang | Everything works fine, except firefox-esr still can't hear my microphone. A JACK device for capturing it pops up. | 06:25 |
| gnarface | onefang: you mentioned that your .asoundrc is missing a capture definition; that's probably what it would take to make firefox work right | 06:29 |
| onefang | It has a capture, works fine with arecord using the default ALSA device. I said it has no dsnoop. | 06:30 |
| gnarface | oh, well i assumed that you'd omitted capture too when you said that, but i think my general point still stands | 06:30 |
| gnarface | if arecord worked then it's not a driver issue | 06:30 |
| gnarface | and there's only so much that can be wrong | 06:31 |
| gnarface | without dsnoop, only one thing can record at once, and with some devices they can't even do duplex right without a .asoundrc | 06:33 |
| gnarface | still, i don't recall firefox ever working right with regards to mic input, so maybe it's just a bug in firefox | 06:34 |
| onefang | Did you miss the "I'm routing ALSA apps through JACK now" bit? | 06:34 |
| gnarface | no, no i didn't miss it i was just trying to talk you down off that ledge, since you also mentioned that firefox still doesn't work | 06:35 |
| onefang | And NOW firefox-ESR AND Falkon hear my mic. B-) | 06:35 |
| gnarface | ah | 06:35 |
| gnarface | <onefang> Everything works fine, except firefox-esr still can't hear my microphone. A JACK device for capturing it pops up. | 06:35 |
| gnarface | this was your last statement, so i was just trying to find you a solution since jack wasn't apparently it | 06:36 |
| onefang | I was still playing with stuff. | 06:36 |
| gnarface | if you got jack working well... whatever works, but for the record i still think a correction to your ~/.asoundrc would have done the trick with less complication | 06:36 |
| onefang | https://www.onlinemictest.com/ is hearing my mic on both browser at the same time. I like this solution. | 06:37 |
| onefang | Now I can put EVERYTHING through JACK. Loving it. Thanks everyone for helping out. | 06:41 |
| onefang | gnarface: I have a complex JACK setup for reasons. Now it is a complete solution for all my sound stuff. Even console beeps go through it, routed to my main monitor. | 06:42 |
| onefang | I have a telehealth appointment on Thursday, and it was their test web page that isn't hearing the microphone after I got the JACK thing working. The web site I mentioned above does work. Same on all the browsers I've been testing. Not sure if they'll use the telehealth web site or phone me, they told me both. | 06:55 |
| onefang | So I'm just gonna assume they only tested their telehealth page under Windows and Mac. | 06:56 |
| systemdlete | whence come this: https://i.imgur.com/oAsNQgo.png | 09:38 |
| systemdlete | the system had been idle for a long time, then this suddenly occurred. I did remove a USB from the VM almost an hour before, but I don't know if that is related or not. I have been noticing these rcu traces increasingly lately. | 09:40 |
| systemdlete | read-copy-update | 09:40 |
| systemdlete | I don't recall seeing these until rather recently. | 09:41 |
| systemdlete | btw, this happened to be a daedalus VM, but I am still running chimaera VMs and hosts and have seen them there as well | 09:41 |
| systemdlete | the host system was also daedalus. | 09:43 |
| systemdlete | I sense that the references to "GP" and the like might be hinting at the video drivers? | 09:44 |
| systemdlete | (only using radeon here as nvidia/nouveau doesn't seem to play well these days) | 09:44 |
| gnarface | hard to say, not sure about the GP thing | 09:56 |
| gnarface | what it looks like is that it was idling and something couldn't wake up the cpu core it wanted fast enough then freaked out | 09:57 |
| gnarface | seems like a problem with power management | 09:58 |
| gnarface | unless you're right about the "GP" thing i don't see any reason to pin this on the video card | 09:59 |
| gnarface | but if you're running without the appropriate cpu microcode package, you might want to try adding it | 09:59 |
| gnarface | you may also want to check the bios for a setting called something like "c1e" power saving | 10:00 |
| gnarface | (then disable it) | 10:00 |
| gnarface | obviously make sure acpid is installed and running too, but i'm not sure this could even happen without it | 10:00 |
| gnarface | heh, and look up what "rcu" means in this context, i think it was important | 10:02 |
| gnarface | hmm, not sure but maybe also try adding the package "irqbalance" if you don't have it? | 10:02 |
| al1r4d | Folks, what codename of Devuan are you using? I am Ceres | 10:14 |
| al1r4d | =) | 10:14 |
| gnarface | i have literally every single release installed somewhere or another | 10:15 |
| gnarface | daedalus on this machine | 10:16 |
| gnarface | systemdlete: it's some kernel process, not sure if the stack trace could lead you to whether it's a driver for a particular piece of hardware or not but maybe worth looking into too | 10:20 |
| al1r4d | I have a problem | 14:00 |
| al1r4d | Somehow the volume is always mute when I pull the headset when my laptop is turned off in the headset is connected | 14:00 |
| |cos| | with this being devuan, i presume the first culprit is not around? systemd's older brother pulseaudio? killing that one tends to solve 90% of audio problems. | 14:03 |
| |cos| | given that the powered-on status is a factor, i'm thinking the vendor and model of the computer might play in. do you know if it happens on a similar setup with some other machine? | 14:07 |
| gnarface | al1r4d: is "Auto-Mute Mode" enabled in alsamixer? | 15:00 |
| gnarface | i think pulseaudio has an equivalent setting | 15:00 |
| gnarface | the first question is whether one of those is working normally | 15:00 |
| joerg | al1r4d: changes in audio config (like muting internal speaker and setting a dedicated audio volume etc for [wired?] headphones) are most likely event driven, no matter if polypaudio, ALSA, pipewire whatever. When you unplug the headphones while system shut down, it will resume on next boot with the config it saved at shutdown time. Which in your case probably means: the system - or parts of the audio stack - _thinks_ your headphones are still plugged in, since | 17:11 |
| joerg | no unplug event happened | 17:11 |
| joerg | simplest solution: do not unplug headphones while system down. probably most simple "fix": plug and unplug again those headphones when system has that "audio muted" faulty state, which usually should result in audio settings syncing with reality again | 17:15 |
| n4dir | not sure if it helps, probably not, if i got the piano-keyboard plugged during boot, alsamixer sees it as an external soundcard, and unplugging won't help to make alsamixer not having it as default | 17:17 |
| n4dir | well, it wants to confirm that plugging and unplugging might be taken a bit care of | 17:18 |
| joerg | probably there are tools that could synthesize a virtual unplug event. sth along sysfs and/or kevents? | 17:20 |
| joerg | udevadm trigger ? | 17:21 |
| joerg | dang. I thought I had a remark about how to get rid of, in my https://termbin.com/h43j asoundrc config patch script | 17:36 |
| joerg | n4dir: ^^^ | 17:36 |
| n4dir | joerg: let me look at it | 17:37 |
| joerg | I think you basically need to do the reverse: stop alsa, remove the device from asoundrc, start alsa again | 17:38 |
| n4dir | joerg: well, i don't really understand it. To me audio is a big riddle in general | 17:38 |
| joerg | look into ~/.asoundrc. Does it exists? does it contain something that looks like a reference to your USB external card? | 17:39 |
| n4dir | joerg: when it happened i didn't even have an asoundrc file | 17:39 |
| n4dir | now i am used to take care of the plug in and out, i got an asoundrc file from fsmithred, it mainly seems to take care firefox has audio if jack is running | 17:40 |
| n4dir | joerg: i think the problem is during boot, if keyboard or midi-keyboard is connected, alsa sets it as device number 1 | 17:40 |
| n4dir | i also figured out i can use "alsamixer -c <number>" to make it use the internal soundcard, not the default one | 17:41 |
| joerg | hmm, a pity. There is an equivalent file in /etc/alsa* or somesuch iirc | 17:41 |
| n4dir | as said, i found my ways to workaround it. But i can't claim i really understand | 17:42 |
| n4dir | using runit doesn't really make it better ... as i don't understand that either, and searching the web, say how to restart alsa, seldom relates to runit | 17:43 |
| fsmithred | You can set the order of the sound cards in /etc/modprobe.d/alsa.conf | 17:49 |
| n4dir | fsmithred: that would assume that always all devices are connected. | 17:50 |
| n4dir | or? | 17:51 |
| n4dir | the real fun info was that alsa thinks midi-keyboards and the piano-keyboard are a soundcard. :-) | 17:51 |
| fsmithred | are they usb? | 17:52 |
| n4dir | most are, the piano keyboard is usb at the laptop/PC, but has a different, weird plug at the keyboard itself | 17:52 |
| n4dir | so probably yes, usb | 17:52 |
| onefang | Might be one of the other USB connectors. Mine has a squarish USB connector on the MIDI keyboard. | 17:53 |
| n4dir | probably | 17:54 |
| n4dir | to me all hardware is just "stuff" | 17:54 |
| joerg | n4dir: man alsactl >>-f Select the configuration file to use. ***The default is /var/lib/alsa/asound.state***<< | 17:54 |
| fsmithred | I'm not sure how to differentiate between different usb sound cards when there's more than one | 17:54 |
| onefang | Type B https://en.wikipedia.org/wiki/USB_hardware?useskin=vector#/media/File:USB.svg | 17:54 |
| n4dir | joerg: thanks | 17:55 |
| n4dir | in general at least for me learning about audio in linux is way more difficult than when i learned linux/computing in general | 17:55 |
| fsmithred | https://termbin.com/dsba <- example /etc/modprobe.d/alsa-base.conf from lubuntu | 17:56 |
| joerg | yes, audio is very obscure and usually poorly documented | 17:56 |
| n4dir | joerg: "obscure". That perfectl describes it at least for me | 17:56 |
| onefang | fsmithred: in .asoundrc have a "pcm.X .... card ALC1220VBDT ...." for example. | 17:57 |
| onefang | Oops and "device 3" | 17:57 |
| n4dir | if i am not wrong Charles Mingus even wrote a song about Audio in Linux: | 18:01 |
| n4dir | Moaning ! | 18:01 |
| onefang | I should document my audio setup on untalenz.rocks. It's very educational. | 18:07 |
| n4dir | could also do it in the forum, if you are a member | 18:08 |
| n4dir | audio documentation sure is too little | 18:08 |
| onefang | Part of the reason for untalenz.rocks is to document and write tutorials for all this music software I'm learning this year. The other reason is to publish the results. | 18:12 |
| onefang | First song should go up this week, or maybe next week. | 18:12 |
| joerg | >><n4dir> fsmithred: that would assume that always all devices are connected. or?<< I don't think so | 18:12 |
| n4dir | joerg: aha. | 18:13 |
| n4dir | Well, i could look into it. Right now i am in a big audio mess in general (mainly midi related, the general audio setup works) | 18:13 |
| n4dir | thanks for any input, it helps a lot | 18:13 |
| onefang | What MIDI mess? | 18:14 |
| joerg | see in the example fsmithred linked: | 18:14 |
| joerg | # Prevent abnormal drivers from grabbing index 0 | 18:14 |
| joerg | options bt87x index=-2 | 18:14 |
| joerg | options cx88_alsa index=-2 | 18:14 |
| joerg | I _guess_ this is relevant. And the first 10 lines tryong to load snd-card-0 to snd-card-7 won't fail when only one or two actually are present | 18:16 |
| onefang | In my JACK startup I use "a2j_control --ehw && a2j_control --start" which lets JACK use ALSA MIDI stuff. In my ~/.asoundrc I specify the actual sound cards. I'm not getting "MIDI device is a sound card" issues. | 18:18 |
| n4dir | onefang: if using the sequencer seq24 and use fluidsynth, then make seq24 choose a different instrument than the default one | 18:19 |
| n4dir | to me it is not too clear what is a channel, what is a bank, what is a program, what is an instrument | 18:20 |
| n4dir | say in ardour i get a gui and just choose an instrument from the soundfont | 18:20 |
| onefang | Ah MIDI basics. | 18:20 |
| onefang | I could explain them, but that's off topic. | 18:21 |
| n4dir | if there only was a channel for offtopic | 18:25 |
| onefang | There is, I try to avoid it, and I really should concentrate on my work now. | 18:28 |
| yuuri | Hello people! I've just installed minimal Devuan, and I can't find where wpa_supplicant takes its configuration from | 21:28 |
| rwp | Hello yuuri. Normally that would be the /etc/wpa_supplicant/wpa_supplicant.conf file. But it may be relocated. Network-Manager for example relocates it. | 21:29 |
| yuuri | There's no wpa_supplicant.conf, and I don't have NetworkManager installed | 21:30 |
| n4dir | might it be as simple as /etc/network/interfaces? | 21:30 |
| rwp | Everything written here https://wiki.debian.org/WiFi/HowToUse#wpa_supplicant applies to Devuan too as Devuan is an overlay fork on top of Debian. | 21:30 |
| yuuri | n4dir: yes, it's here! What service uses this file? | 21:31 |
| n4dir | perhaps dhclient? | 21:32 |
| rwp | Most of the network programs at least observe the contents of /etc/network/interfaces file. The ifupdown package makes the most use of it. | 21:32 |
| rwp | The manager programs network-manager, wicd (now dead rip), connman read that file and ignore any interface listed in it leaving that interface to ifupdown to manage. | 21:33 |
| n4dir | gotta admit that now that you get wireless during installation, i was astonished how all that works, but didn't really look in it | 21:33 |
| n4dir | rwp: they are not even installed | 21:33 |
| rwp | What is installed? | 21:35 |
| n4dir | wpa_dingens | 21:36 |
| rwp | Okay. However I assume you have things working. But it was yuuri who had asked. And I assume that if asking yuuri is not using any of those either and is doing everything manually. | 21:37 |
| yuuri | I have dhclient installed | 21:38 |
| yuuri | Trying to modify existing wireless network description in /etc/network/interfaces, how do I write WPA3 password here? | 21:39 |
| yuuri | wpa-psk doesn't cut it | 21:39 |
| rwp | The process is described here: https://wiki.debian.org/WiFi/HowToUse#wpa_supplicant | 21:40 |
| n4dir | yuuri: you have wpa-ssid in interfaces file too? | 21:41 |
| yuuri | yes | 21:41 |
| rwp | Use wpa_passphrase to create the wpa_supplicant.conf entry. Then add the section described lower "iface wlan0 inet dhcp\n wpa-conf /etc/wpa_supplicant/wpa_supplicant.conf" | 21:41 |
| yuuri | let me see | 21:41 |
| rwp | I do not recommend using the old wireless-tools package way of putting the passphrase in the interfaces file. But if you do then you need to have the wireless-tools package installed. But I think it is better to move forward to the normal wpa_supplicant tool way. | 21:43 |
| yuuri | Also, it tried get DHCP at boot (and couldn't because of outdated wifi password) and it significantly delayed the boot | 21:43 |
| yuuri | how can I make it ask for network configuration post-boot? | 21:44 |
| n4dir | yuuri: during the installation process? | 21:44 |
| yuuri | after the installation | 21:44 |
| rwp | The "auto wlan0" config causes synchronous boot action. The "allow-hotplug wlan0" config uses the event driven path in the background. | 21:45 |
| n4dir | after configuring of the network during installation fails or succeeds, hit tab to get to "go back", then choose "don't configure network" | 21:45 |
| rwp | If the network is required then use auto. If the network is only needed opportunistically later then use allow-hotplug. | 21:45 |
| n4dir | how to avoid the installer trying to configure the network at all, not having to "go back", i wouldn't know | 21:46 |
| yuuri | There's no "go back", only an already installed system with no graphics | 21:46 |
| yuuri | (It's going to be a homeserver) | 21:46 |
| rwp | yuuri, I wrote up a blog article on setting up a Banana Pi which includes me taking about this type of WiFi configuration. https://www.proulx.com/~bob/doc/banana-pi/ Look at the section "Option 2 using wpa_supplicant" and ignore the rest but that covers what you are trying to do. | 21:47 |
| yuuri | rwp: will see about hotplug, thanks; adding wpa-conf to interfaces makes ifup wlan0 fail to start | 21:47 |
| rwp | Does it produce an error message about why it fails to start? | 21:48 |
| yuuri | wpa_supplicant: /sbin/wpa_supplicant daemon failed to start | 21:49 |
| yuuri | run-parts: /etc/network/if-pre-up.d/wpasupplicant exited with return code 1 | 21:49 |
| yuuri | ifup: failed to bring up wlan0 | 21:49 |
| yuuri | that's all | 21:49 |
| rwp | Look in /var/log/syslog and see if there are more messages from wpa_supplicate there. | 21:49 |
| rwp | I am suspecting either permissions problems on files or contents of /etc/wpa_supplicant/wpa_supplicant.conf being incorrect. | 21:50 |
| yuuri | incorrect as in bad format or incorrect password | 21:51 |
| rwp | I would suspect bad format. | 21:51 |
| rwp | But I expect it would print an error in the /var/log/syslog in that case too. | 21:52 |
| yuuri | yes, it generated a WPA2 config by default | 21:53 |
| yuuri | var/log/syslog is empty | 21:53 |
| yuuri | like, completely | 21:53 |
| yuuri | I fixed the config to use WPA3 and ifup seems to work | 21:53 |
| rwp | Completely empty? You said you installed minimal Devuan. Perhaps it was too minimal? :-) | 21:54 |
| yuuri | rebooting | 21:54 |
| rwp | I guess I would suggest installing "apt-get install rsyslog cron logrotate" so that system logs will exist and be expired automatically. :-) | 21:55 |
| yuuri | ...network is unreachable, and I saw a little "failed to bring up wlan0" | 21:55 |
| yuuri | that's after I changed it to hotplug | 21:55 |
| yuuri | and rebooted | 21:55 |
| yuuri | I have rsyslog service installed and active in /etc/service | 21:56 |
| rwp | The change to hotplug was so that the boot would be asynchronous and the network run in the background asynchronously. | 21:56 |
| rwp | Which init uses /etc/service? I don't recall. | 21:57 |
| yuuri | runit | 21:57 |
| rwp | Runit. Gotcha. Should work okay no problem. | 21:57 |
| yuuri | yeah it borken now | 21:57 |
| rwp | It sounded like it was borken before too though. So actually just no change. | 21:57 |
| yuuri | well, it couldn't connect to a network | 21:58 |
| yuuri | now it did but ping shows "From (an address I have no idea about) ... Destination Host Unreachable" | 21:58 |
| yuuri | can't even ping my router now | 21:58 |
| rwp | Wait... You could ping your router before? Then it was all working before! | 21:59 |
| yuuri | After fixing the config but before rebooting | 21:59 |
| yuuri | now it works after I manually ifup | 21:59 |
| yuuri | what gives | 22:00 |
| yuuri | now let's remove hotplug and reboot again | 22:01 |
| rwp | Good idea. | 22:01 |
| yuuri | alright, now it bounded but it took precious boot time again | 22:02 |
| yuuri | *and also* I've got a couple "ok: run: (service)" messages flowing over login prompt | 22:02 |
| rwp | The two paths through the network code are different paths. There is the synchronous path (auto) and the asynchronously event driven path (allow-hotplug) and the code is different between them. | 22:03 |
| n4dir | yuuri: same here regarding the flowing over messages | 22:03 |
| rwp | The console log message overwriting is typical on consoles. | 22:03 |
| yuuri | never happened to me on other distros :( | 22:04 |
| rwp | To avoid that you can set "dmesg -n5" to prune that down somewhat. | 22:04 |
| yuuri | where do I stick it | 22:04 |
| rwp | Just run that command interactively initially to protect your sanity. | 22:04 |
| yuuri | nothing | 22:05 |
| rwp | The Linux kernel default is 8 so that all messages are logged to the console. Red Hat sets it to 3 so users never see console messages. Debian/Devuan don't touch the kernel default leaving it at 8. | 22:05 |
| rwp | Here is the listing of numbers and levels: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/include/linux/kern_levels.h | 22:06 |
| rwp | This reference will explain it: http://tuxthink.blogspot.com/2012/07/printk-and-console-log-level.html | 22:06 |
| rwp | I normally set the log level in my firewall configuration startup for machines on the hostile Internet that both should have a firewall and are constantly probed. | 22:07 |
| rwp | If "dmesg -n5" still has annoying messages overwriting the console then increase the limitation to "dmesg -n3" and that should trim out the rest of the noise while still leaving critical messages. | 22:07 |
| rwp | Summary: You have things working, synchronously at boot with "auto", it takes a moment to associate and dhcp an address but works. Right? | 22:08 |
| yuuri | Yes | 22:09 |
| rwp | I think you are still missing a system logging daemon. I recommend "apt-get install rsyslog cron logrotate" so that system logs will exist and be expired automatically. :-) | 22:09 |
| yuuri | ...all are installed | 22:09 |
| rwp | So what does "sv status rsyslog" say? (I think that is the right runit syntax.) | 22:10 |
| yuuri | I have a service called svlogd that's not currently enabled, shall I? | 22:10 |
| yuuri | it says "run: rsyslog: (pid 1553) 549s" | 22:11 |
| yuuri | which I assume means that it's running | 22:11 |
| rwp | That says that rsyslogd is running. Good. Strange that the /var/log/syslog file is zero sized empty. | 22:11 |
| yuuri | oh, now it's full | 22:12 |
| yuuri | maybe I made a typo before | 22:12 |
| rwp | There is a /etc/rsyslog.conf file and it has config stuff in it? | 22:12 |
| rwp | Oh! Good! Easier than debugging it. Probably good useful debug messages in there from wpa_supplicant. | 22:12 |
| yuuri | yeah, it's full of something | 22:12 |
| rwp | But wpa_supplicant is working so no problem now. But earlier it probably woudl have been useful. | 22:13 |
| yuuri | well, I want it to connect to the network in the background | 22:13 |
| rwp | How often do you expect to be rebooting? I would say reliably working is better than booting faster without a network. | 22:14 |
| rwp | I think it likely that the network using hotplug would /eventually/ wake up. I think. And then eventually it would connect. | 22:14 |
| yuuri | let's test that | 22:14 |
| rwp | Also hotplug is mostly required on USB network devices which can be detached and attached dynamically. | 22:14 |
| rwp | If it were me and it were working at boot synchronously with "auto" then I would stop there, use it, and move on to the next problem. | 22:15 |
| yuuri | is there a better way to set up networking than ifupdown? Void Linux provides a runit servicre for wpa_supplicant | 22:16 |
| yuuri | I mean I guess I can just move on yeah | 22:17 |
| rwp | Hmm... "Better" means personal preference. I personally like ifupdown best for WiFi devices that are not mobile. | 22:17 |
| rwp | The ifupdown manager has been most reliable for me. The main alternative is Network-Manager which most people use on laptops and mobile devices. | 22:18 |
| rwp | I think it is fine there for laptops which are always occupied by a person driving the keyboard. | 22:18 |
| rwp | But for IoT devices and servers I have had NM drop connections and never reconnect requiring me to go get on the console and manually connect too often. I really despise NM. But on a laptop always being used by someone at the keyboard then NM is the simplest way to pick WiFi networks and connect to them. | 22:19 |
| yuuri | It's a laptop but it's going to be stationary | 22:19 |
| rwp | The only alternative to NM is "connman" which is also quite often used. It's okay. To my thinking the connman UI is a little strange. It works after you figure it out though. | 22:19 |
| rwp | The old previously most reliable manager was WICD but that is a python2 program, now orphaned, and with the deprecation of python2 it is now no longer available. So far no one has ported wicd to python3 and so it is no longer available. | 22:20 |
| rwp | For a stationary laptop doing what I assume is going to be dedicated things like a personal server I think ifupdown is most reliable. But you could make NM work too. With the exception that sometimes, not always but sometimes, you would need to get on the console and tell it manually to connect. | 22:22 |
| yuuri | after fedora experience with its painfully slow dnf, which is written in python, I have prejudice against python system utilities | 22:22 |
| rwp | I can understand that. My python experience is always looking at hugely long backtraces instead of reasonable user presentable error messages. But I blame the programmer not the language for both of those things. | 22:23 |
| rwp | If you decide you want a deeper dive into working wpa_supplicant and dhclient manually step by step ping me and I can give a walk through of those step by step. | 22:23 |
| yuuri | well, they work for now I guess | 22:24 |
| yuuri | thanks for the offer | 22:24 |
| rwp | I need to drop afk for a while. And we have been hogging the channel for a bit so it is good to let others rotate in too. | 22:24 |
| rwp | Good luck with the rest of your setup! | 22:24 |
| yuuri | aw, can't get openjdk-21 on stable | 22:25 |
| tocsa | How can I see and potentially flush the DNS cache? | 22:26 |
| tocsa | All web search comes up with something systemd based, but as we know Devuan is systemdless | 22:27 |
| tocsa | I don't have a reoslvd start script in /etc/init.d (trying to find things from my memory decades before), I don't see any daemons or processes running by the name of "resolve" or "dns" in their name. | 22:28 |
| tocsa | My main problem is this https://www.reddit.com/r/devuan/comments/17v8td2/strange_network_problem_cannot_connect_to_google/ | 22:28 |
| jjakob | depending on who you ask, not being able to connect to google might be a good thing | 22:30 |
| tocsa | I know. However various services/platforms which use GCP storage under the hood could fail unexpectedly because of this. (They do in my case) | 22:32 |
| tocsa | I don't even know what exactly provides the dns right now in my installation | 22:33 |
| yuuri | dhclient mayhaps | 22:33 |
| yuuri | tocsa: is there /etc/resolv.conf? | 22:42 |
| tocsa | @yuuri There is. This is a ~5 year old installation | 23:01 |
| fsmithred | Have you looked in /etc/hosts.deny? | 23:02 |
| tocsa | As far as dhcp goes, I don't see a daemon running, I have isc-dhcp-client install version 4.4.3-P1-4 | 23:04 |
| tocsa | @fsmithred I have an /etc/hosts.deny but it's all commented out | 23:06 |
| tocsa | One root cause could be this: https://imgur.com/35zbbcs | 23:11 |
| tocsa | I perform an nslookup / dig. Since GCP (or any Amazon S3 or Azure storage, etc) is load balanced I get back a dozen of IP addresses. However when I try to connect or ping my machine tries to use an IP which was not among those IPs | 23:12 |
| tocsa | I suspect that IP is some leftover from the past. But where did it get stuck, where is it coming from? | 23:12 |
| fsmithred | bad dns records? | 23:16 |
| fsmithred | maybe try setting a nameserver in /etc/resolv.conf | 23:17 |
| fsmithred | Your router might be using your ISP and it might not be right. I've seen that happen. | 23:17 |
| xisop | greetings. is it possible to use a newer kernel than what is provided by apt? | 23:21 |
| brocashelm | xisop: yes, but it could break your system if you do not do it correctly | 23:21 |
| brocashelm | xisop: if you need a newer kernel than 6.1 (assuming you are on daedalus), consider adding daedalus-backports to your sources.list file | 23:22 |
| xisop | i'm on chimera actually. i guess i should upgrade to daedalus first | 23:22 |
| brocashelm | worth it to do so | 23:22 |
| brocashelm | don't create a frankendevuan! | 23:22 |
| xisop | apt-get dist-upgrade? | 23:22 |
| brocashelm | if you use any non-free drivers, append non-free-firmware to your repo | 23:23 |
| xisop | ah, i'll follow this https://www.devuan.org/os/documentation/install-guides/daedalus/upgrade-to-daedalus | 23:23 |
| brocashelm | example: deb http://deb.devuan.org/merged/ daedalus main contrib non-free non-free-firmware | 23:24 |
| n4dir | xisop: at least for debian there always have been some non-repo kernels around, claiming to do this and that better. Say liquorix was one of them | 23:24 |
| brocashelm | then run apt update -> apt dist-upgrade -> apt upgrade -> reboot | 23:24 |
| xisop | all my chimera sources were just `main` | 23:25 |
| brocashelm | ok, then just keep it that way | 23:25 |
| xisop | k | 23:25 |
| xisop | i love this distro | 23:25 |
| xisop | nothing like it | 23:26 |
| brocashelm | if you use firmware, a change happened where the drivers from non-free got placed in non-free-firmware | 23:26 |
| brocashelm | cool, enjoy | 23:26 |
| xisop | how did that happen? | 23:26 |
| brocashelm | i even used ceres (unstable) for years and enjoyed it | 23:26 |
| brocashelm | it's a debian decision thing | 23:26 |
| xisop | i think aside from systemd, debian is a solid distro | 23:26 |
| brocashelm | https://lists.debian.org/debian-devel-announce/2022/10/msg00001.html | 23:26 |
| brocashelm | devuan is basically debian but with sysvinit/runit/openrc instead of systemd | 23:27 |
| brocashelm | everything else stays the same | 23:28 |
| brocashelm | so whatever debian does that's not systemd-centric, devuan mirrors that | 23:28 |
| n4dir | most of the time i couldn't even say if i am using debian or devuan at the moment. Now i purged all debian installations, but before, i was never too sure | 23:29 |
| xisop | i find it hard to accept systemd as a linux package. if it stopped at being just an init system, i'd probably have no problem with it | 23:30 |
| xisop | it's just this giant black box of always changing code | 23:31 |
| tocsa | @fsmithred The DHCP results in setting my wifi router (which is an OpenWRT flashed one) as the DNS server. 192.168.1.1. I tried the same thing on a Windows 11 backup laptop, and that is able to ping, because it doesn't use that rouge IP address | 23:41 |
| tocsa | @fsmithred @yuuri My whole network issue is a PEBKAC! So this is a more than 5 year old installation, and I don't know how many years ago I must have had some DNS problem, and so I "temporary" "poisoned" my /etc/hosts file with some IP addresses, one of which is this GCP storage IP. I peeked into the /etc/host and I just realized that. Sorry to bug you with this. | 23:48 |
| tocsa | Wow I kicked myself in the nuts with this... | 23:54 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
