[00:30] <mattfly> sarnold: I installed with apt
[00:30] <mattfly> No GUI on the server
[00:30] <mattfly> i didn't even configure it, i just wanted to safely get rid of it
[02:25] <subsume> apparently i am too stupid to follow the basic steps necessary to add a deploy key to github.com
[02:26] <subsume> or the documentation is plainly lying
[02:48] <tdannecy> What is tar doing?? https://usercontent.irccloud-cdn.com/file/aPD82KPL/tar%20output
[02:49] <tdannecy> My command was `tar -c out.tar /link/to/directory`
[02:50] <tdannecy> Aha I need `-cf`. Whoops.
[03:02] <Intelo> When I connect DP cable to my m6700 laptop, it says no signals. When I type xrandr while the cable is connected, it says this https://paste.ubuntu.com/p/z3yqFHrsTY/ . my hardware drivers are these https://imgur.com/qAH7Ngx.png        Any clues?
[03:03] <Intelo> lotuspsychje: leftyfb  ^
[03:26] <Kraus> Hey all, I'm curious, is PipeWire going to be part of 21.04 by default? I've heard yes and no from different sources.
[03:27] <Intelo> solved
[03:34] <Intelo> Which GPU might be a little better than quadro k3000m  but for desktop? I am asking because my laptop has quadro k3000m and it just do fine. If I buy a desktop, I don't want to spend a huge amount on gpu but I do want to play mid level games.  any suggestions for linux compatible gpus in that category?
[03:34] <Bashing-om> Kraus: Yes: per - https://packages.ubuntu.com/search?searchon=contents&keywords=pipewire&mode=exactfilename&suite=hirsute&arch=any :D
[03:37] <Kraus> Bashing-om: Cannot wait. :) It's two days until release, right?
[03:38] <arther181> Hello guys! I can mount rootfs by root=/dev/sdb3 (grub) it's a squshfs disk. Is there a way to mount this disk by label.
[03:39] <Bashing-om> Kraus: Yup - still scheduled for release on the 22nd :D
[04:20] <Arc1> In a shared "/home" situation why do configs get changed if the same aps in partitions with diff OS ver's have exactly the same settings?
[04:25] <ninev> boooooooon...tooooooo
[04:25] <ninev> sup
[04:26] <oerheks> a shared home can be a disaster
[04:26] <ninev> y?
[04:28] <Arc1> oerheks: you mean because of configs or something else?
[04:30] <oerheks> browsers probably
[04:30] <Arc1> firefox requests a separate profile...
[04:31] <oerheks> and maybe snaps and any app that needa auth with keyring
[04:32] <Arc1> but if all the pathwords and user names are the same, why would things get changed?
[04:33] <Arc1> *passwords...
[04:38] <Woodonia> So, I'm stupid, and pretty sure I know the answer to this, but:  I was playing around with 21.04 daily images, and went with encrypted ZFS.  Unfortunately, my witty attempt at a good passphrase (in XKCD style) has bitten my behind, as I can't remember all of the passphrase.  For some reason I did remember to record the recovery key during
[04:38] <Woodonia> installation... but didn't write down my passphrase.  Is there any way to recover the data?  (the data is all backed up and all readily replaceable, but I've been using this as a learning exercise, and have hit nothing but dead ends)
[04:41] <Arc1> ok, another question: For the latest Xubuntu, what is the minimum safe partition size for: 1. OS with "/home", 2. OS without "/home" and 3. "/home" by itself?
[04:43] <Bashing-om> Arc1: Well ,, root can install to 5 gigs - and a /home of 10 gigs leaves just a bit of wiggle room.
[04:45] <Arc1> Bashing-om: what are the current recomendations to operate with a reasonable safety margin?
[04:47] <Bashing-om> Arc1: "YOU" must define what in your use case what is "a reasonable safety margin" . Want to see my partitioning set up that works well for my uses ?
[04:47] <Arc1> Bashing-om: sure
[04:49] <Bashing-om> Arc1: See mine: https://termbin.com/8hlp with cirrently just the 2 drives active.
[04:57] <Arc1> Bashing-om: I see. Regarding the "safety margin" perhaps there is a general recomendation like "keep at least 5% of disk free" or something like that and perhaps that % varies depending on the purpose/type of the partition?
[04:59] <Bashing-om> Arc1: thing is no one can tell how much sugar you want to put into any bag. There is no use recomendation for how "YOU" will use the system.
[05:02] <Bashing-om> Arc1: Seen: https://help.ubuntu.com/community/Installation/SystemRequirements/ ?
[05:04] <Arc1> Bashing-om: while that makes sense, I recently ran into a problem with a system not leting me log in because it was out of space. Just want to avoid this situation in the future.
[05:05] <Arc1> regarding recomendations on that page: 25GB, it is for Ubuntu not Xubuntu and I want to differentiate between separate /home or not
[05:07] <Woodonia> Arc1 There are definitely recommendations depending on different use cases, file system types, etc - but as Bashing-om said, there really isn't a solid guideline because different people use things differently.  I'd highly recommend having separate /root and /home partitions (or data sets), as that will greatly limit situations like you mention
[05:07] <Woodonia> you just experienced
[05:10] <Arc1> runing out of space on which partition is likely to cause more of a problem on /root or on /home?
[05:11] <Woodonia> Most definitely / (root)
[05:15] <Arc1> if root (xub20.04) is ext4, no new aps are to be intalled, /home and /data are on separate partitions, would 10% empty space on /root be enough for regular updates and maintenance?
[05:16] <Bashing-om> Arc1: If one is going to consistenly run the system out of space - then one creates a seperate parition to contain what runs the system out of space. That way the system does not crash.
[05:18] <Arc1> Bashing-om: good advice; but this only happened once and unfortunately I couldn't figure out what did it...
[05:19] <Bashing-om> Arc1: As "these things happen" one keeps a rescue CD - mounts the relevant partitions from rescue and looks to see what happened.
[05:22] <Bashing-om> Arc1: I am out for this session - Keep plugging away and we see what the result is :P
[05:22] <Arc1> Bashing-om: makes sense; I'll be a more patient detective next time :)
[05:23] <Bashing-om> Arc1: As they say "patience is a virtue" :P
[05:23] <Arc1> Bashing-om: al right, all right, talk to you later :)
[05:24] <Bashing-om> \o laters all
[05:26] <Arc1> ok, a diff question, perhaps the issue is my reading comprehension but: For the partition of the new OS, installer says: "no root file system is defined" (partition is already formated as ext4) - is it asking for a mount point? if so, should the answer for all booting OS partitions be the same - "/"?
[05:28] <ducasse> morning folks
[05:30] <Woodonia> I'm not sure exactly what you mean for all booting OS partitions... but you'll need to make sure that there is a root (/) mountpoint assigned on one of the partitions (as this is where all of the programs / OS will live).  Most times you'll have /boot on the same root file system (/), but you sometimes don't, if you have EFI, or booting multiple
[05:30] <Woodonia> OSs, etc.  - but those get a bit more complicated quite quickly
[05:31] <ducasse> you can still have /boot on the root fs, just not /boot/efi
[05:37] <Arc1> so in a multiboot system with two OS's each on a separate ext4 partition, there is no problem with indicating that the mount point for each is "/"?
[05:41] <Arc1> I mean, it is a situation where either one boots or the other but not both simultaneously
[05:43] <Arc1> * the two are not mounted simulteneously...
[05:45] <Arc1> sorry if the question is too idiotic... it's been a long day...
[06:09] <unixbsd> maybe anyone interested in learning programming, free, course to make Graphical applications for starters and beginners using the Ubuntu os?
[06:10] <pony> I'm not sure what you're asking there
[06:10] <pony> are you looking for such a coursre?
[06:11] <Maik> !discuss | unixbsd
[06:12] <unixbsd> pony: other way around, to teach people for GUI.
[06:14] <Maik> Mind taking it to discuss or offtopic please? It's not a support question.
[06:54] <ducasse> unixbsd: #ubuntu-discuss would be a good place to ask, but it's quiet right now
[06:58] <unixbsd> thx ok
[07:04] <zamba> is it time to upgrade to 20.04?
[07:16] <rfm> zamba, if you're on the LTS train I'd say yes,I have
[07:17] <zamba> yeah, currently running 18.04
[07:18] <guiverc> zamba, if using 'main' software (ubuntu desktop & provided apps) then you have two years of support life, if using flavor or 'universe' software than that's EOL this month (except studio which wasn't a LTS for 18.04)
[07:18] <zamba> the issue we're having is that the kernel mainline is pretty old for 18.04
[07:19] <guiverc> zamba, there are two kernel stacks available on Ubuntu for LTS releases.  Ubuntu 18.04 using GA stack uses 4.15. but with HWE enabled it's 5.4, the identical stack to 20.04 (GA)
[07:20] <zamba> guiverc: is it recommended to just turn on GA stack?
[07:20] <zamba> (and how do i do that?)
[07:21] <guiverc> see https://wiki.ubuntu.com/Kernel/LTSEnablementStack and https://wiki.ubuntu.com/Kernel/RollingLTSEnablementStack  (the ISO used to install controls the default, but it can be changed post-install)
[07:23] <rfm> zamba, HWE stack is only needed if you have new hardware;  If you have a system that's happily running 18.04 you care not
[07:23] <guiverc> zamba, I don't know what ISO you used to install, nor if desktop or server, so i cannot know which stack you're using.. GA = general or the initial stack, HWE = hardware enablement, ie. 18.04 bumped to 18.10 stack, then 19.04, 19.10 before final stack is the 20.04 GA stack
[07:25] <zamba> guiverc: this is the one supplied from the OVH cloud provider
[07:25] <zamba> guiverc: as part of their deployment
[07:27] <zamba> the reason we need to upgrade is because we're having periodic kernel panics
[07:28] <zamba> https://dpaste.org/DA8t
[07:29] <rfm> zamba, if OVH is offering 20.04 vms then they should be happy with it.  Shouldn't it be easy to spin up a new vm?
[09:30] <kristian_on_linu> cheers
[09:31] <kristian_on_linu> I just got 20.04, and Brave refuses to work with Keepassxc, any idea what I should do?
[09:44] <amuro> Why are there so many ass hole in Windows channel? lol
[09:44] <cbreak> amuro: ask in windows channel
[09:45] <amuro> I was asking wifi problem on my windows machine and then I may have some off topic conversation about windows laptops then I get warning LOL
[09:45] <amuro> dumbass windows fanboy
[09:47] <DJones> misc
[09:48] <amuro> kristian_on_linu: try firefox
[10:07] <p0a> Hello when I use `apt search pegtl' (for the library), I see two results, pegtl-dev 1.3.1-1 and tao-pegtl-dev 2.8.1-1
[10:08] <p0a> Why are is such a big version difference? Or is it simply supporting both the major 1 and 2 versions?
[10:15] <handlebar> pOa: Did you try to get a solution from Freenode #ubuntu
[10:16] <handlebar> pardon, P0a
[10:17] <handlebar> sorry. my mistake. I thought I was on DALnet, my apologies
[10:18] <handlebar> only 0618 EST here in the USA
[10:19] <handlebar> East Coast to be exact
[10:20] <handlebar> ikonia: is a guru
[10:21] <handlebar> and pici =)
[10:25] <p0a> It's alright I went with cmake's git fetch so I'm using the latest 3.2.0
[10:57] <linuxmint> test
[11:03] <linuxmint> Any help with hardware for my Ubuntu Desktop?
[11:04] <linuxmint> Monitors are black and keyboard shows no LED lights on boot?
[11:04] <linuxmint> Tried a new computer and same issue.
[11:04] <linuxmint> Monitors and HDMI cable works via MacMini.
[11:05] <linuxmint> Computer powers up, but maybe the SSD is corrupt.  However, no BIOS flash screen appears on black monitors?
[11:06] <linuxmint> New computer: I installed MOBO, CPU, RAM and PSU but same issue as with old computers with black monitors?
[11:06] <linuxmint> HDMI from monitor to MOBO HDMI port..
[11:08] <ogra> if you do not see BIOS this isnt a problem with the OS at all
[11:08] <ogra> (or UEFI)
[11:26] <cbreak> linuxmint: which CPU?
[11:26] <cbreak> which mainboard?
[11:28] <linuxmint> cbreak, Intel core i3-10105F LGA1200 and Asrock H470M-HVS
[11:29] <cbreak> https://ark.intel.com/content/www/us/en/ark/products/203474/intel-core-i3-10105f-processor-6m-cache-up-to-4-40-ghz.html ?
[11:30] <cbreak> you won't get graphics output without a GPU of some sort
[11:31] <cbreak> the F variant is without built-in GPU
[11:34] <linuxmint> cbreak, wow ok.  I did first try with the GPU Asus Gt710-Sl-2Gd5-Brk Graphics Card
[11:35] <linuxmint> Connected HDMI, vga and display port, but black screens.
[11:35] <cbreak> to use your GPU, attach your screen to the GPU
[11:35] <cbreak> not the mainboard ports
[11:35] <cbreak> those are only for GPUs that are built-in to CPUs
[11:37] <linuxmint> cbreak, yes, I tried 1st with the GPU and the 3 monitors only into the GPU, but black screens.  Now I have removed the GPU and plugged a VGA to the MOBO, then tried only HDMI to the mob.
[11:37] <linuxmint> *mobo
[11:49] <BluesKaj> Hi folks
[12:01] <sacarde> hi
[12:02] <sacarde> how I can to have a keyboard for /dev/tty.. and another for /dev/pts/..
[14:19] <larkfisherman> Hey guys, does anyone know whether I can somehow (cgroups, maybe?) reserve some small percentage (5%?) of my CPU to be *always* available for gnome-shell, no matter how many other processes are running?
[14:21] <larkfisherman> Related question, how exactly does Ubuntu *start* gnome-shell? If I figure that out, maybe I can figure out the answer to the original question myself...
[14:26] <cbreak> larkfisherman: probably via systemd somehow.
[14:27] <cbreak> larkfisherman: systemctl status will show it in the tree, probably
[14:28] <larkfisherman> cbreak: Thanks, I forgot about that. It's apparently called "gnome-shell-x11.service". Now, only to figure out how to add a constant cpu slice...
[14:28] <cbreak> nice is not enough? :) not sure if nice works on this...
[14:32] <larkfisherman> Wait, does nice guarantee a cpu slice? I though it only specified some abstract "priority" thingy - and when there's a fork bomb, "high priority process" among tens of thousands of fork bomb processes is not much of CPU...
[14:51] <ducasse> larkfisherman: nope, nice does not guarantee anything
[14:54] <kristian_on_linu> can any of you ppl recommend a good antivirus program that's easy to use?
[14:54] <lotuspsychje> !info clamav
[14:54] <leftyfb> kristian_on_linu: for ubuntu, no. You don't need it. Unless you're scanning a shared filesystem to be accessed by Windows machines
[14:55] <kristian_on_linu> lotuspsychje, I'm aware of clamav, but I get a headache from trying to use it
[14:55] <kristian_on_linu> leftyfb, I'm contractually obliged to have antivirus
[14:56] <leftyfb> kristian_on_linu: ah, good ole security policies based on 1 OS
[14:57] <leftyfb> kristian_on_linu: install clamav and call the contract satisfied
[14:57] <ducasse> kristian_on_linu: there are a few commercial options, but we can't recommend any
[16:29] <BlackComb> Hi Guys,
[16:29] <BlackComb> I'm currently running Ubuntu 20.04. Everything is working file until I Rebooted my virtualbox now i'm getting a black screen with blinking cursor
[16:29] <BlackComb> any help would be great
[16:37] <kristian_on_linu> BlackComb, is the black screen for the virtual machine or the system as such?
[16:38] <kristian_on_linu> and what OS is the vb running?
[16:38] <kristian_on_linu> I don't think I can help, but that info would probably help the brighter users here
[16:38] <JoeBk> I'm installing 20.04 in a virtialboc now.
[16:39] <JoeBk> I'll see what happens.
[16:39] <BlackComb> kristian_on_linu the system os ubuntu 20.04
[16:39] <BlackComb> it was working file like yesterday
[16:40] <BlackComb> then turn back on again it goes into the blackscreen of death with cursor blinking
[16:40] <JoeBk> did you try powering off the virtualbox and restarting.
[16:40] <BlackComb> yeap
[16:41] <BlackComb> however I am able to access the terminal with Ctrl + ALT + F3
[16:42] <BlackComb> F2*.
[16:42] <JoeBk> sounds like X is not starting.
[16:43] <JoeBk> I tried installing 21.04 on a usb drive and got an error.
[16:45] <BlackComb> JoeBk
[16:45] <BlackComb> The os was working fine yesterday
[16:48] <[VMGuy23]> Did it restart yesterday? did you change anything the last time it was on and working?
[16:48] <BlackComb> nope
[16:48] <BlackComb> it just crash
[16:50] <leftyfb> !ubuntu+1 | JoeBk
[17:02] <sneakyimp> Is it possible to have two *different* nvidia cards on your ubuntu machine each use a different proprietary driver? I have a GeForce GTX 470 and a new GeForce GTX 1050 Ti and the 470 won't work with the nvidia-driver-460 that seems unavoidable when you try to install CUDA. I posted askbuntu forum some time ago, even added a bounty, but no answers: https://askubuntu.com/questions/1331925/possible-to-install-distinc
[17:02] <sneakyimp> ilar-nvidia-gpus-on-ubuntu-20
[17:02] <sneakyimp> the `nvidia-smi` command doesn't even acknowledge the 470 currently
[17:03] <lotuspsychje> sneakyimp: sudo lshw -C video to see your card(s)
[17:05] <sneakyimp> lotuspsychje: that command reports both cards, but if you check that forum post, there's a lot more detail. I cannot alter the driver in use with the 470 using the Software control panel, and it would appear that the driver is suboptimal
[17:06] <sneakyimp> lotuspsychje: I'm wondering if the 470 is even being used at all or whether I should remove it and sell it on ebay
[17:07] <lotuspsychje> sneakyimp: lshw should show if your card(s) have loaded drivers or unclaimed
[17:09] <sneakyimp> lotuspsychje: the lshw command output is not particularly informative, nor does it tell me how to get the 470 working with the nvidia-driver-390 while allowing the GTX 1050 to work with the more modern nvidia-driver-460. See for yourself: https://paste.ubuntu.com/p/rdmYr6k9Gd/
[17:10] <sneakyimp> lotuspsychje: note how the 1050 has this but, but the 470 does not: driver=nvidia
[17:10] <lotuspsychje> ah there we go sneakyimp second card is 'unclaimed' means driver not loaded, lets have a look in your dmesg plz?
[17:11] <cbreak> sneakyimp: do you want to use the card?
[17:11] <sneakyimp> lotuspsychje: not sure what you mean, happy to supply dmesg output
[17:11] <cbreak> maybe you can give it to some VM via an IOMMU thing
[17:11] <cbreak> and inside the VM you install the other drivers
[17:12] <sneakyimp> cbreak: YES i'd like the card to assist in any way possible in speeding blender rendering. although i doubt it will do so.
[17:12] <cbreak> (if your mainboard supports IOMMU)
[17:12] <lotuspsychje> sneakyimp: full dmesg in a paste would be nice, maybe volunteers can find something
[17:12] <cbreak> I can't imagine such an ancient card helping in any way
[17:12] <sneakyimp> cbreak: the reason i bought the new 1050 was because i couldn't get CUDA working with the 470. the 470 doesn't even appear as an option for GPU rendering in blender. See https://askubuntu.com/questions/1331925/possible-to-install-distinct-drivers-for-two-dissimilar-nvidia-gpus-on-ubuntu-20
[17:13] <cbreak> at least not in relation to the amount of power it sucks
[17:13] <sneakyimp> cbreak: if the old card won't help, that is useful information. i was imagining it might at least handle display duty, leaving the 1050 entirely to blender, but after installing the 1050, the 470 display stoppe dworking
[17:14] <cbreak> sneakyimp: seems that thing has compute capability 2.0
[17:14] <cbreak> maybe you can use nouveau on it?
[17:14] <cbreak> or what ever that thing's called
[17:15] <lotuspsychje> yeah ubuntu should at least be able to use 2 screens on a system right
[17:15] <cbreak> it can. I've done it with a single GPU :)
[17:15] <cbreak> all my multi-gpu ubuntu machines don't have screens attached at all...
[17:16] <lotuspsychje> maybe we will see a driver conflict in his dmesg
[17:16] <sneakyimp> cbreak: the Additional Drivers control panel one typically uses to choose the card driver is *stick* on 'continue using proprietary driver' and will not allow me to select any other option.
[17:16] <sneakyimp> lotuspsychje: here's the dmesg. Note in particular line 1053: https://paste.ubuntu.com/p/xphJNwtyGH/
[17:17] <sneakyimp> the issue seems to be that the newer nvidia-driver-460 is hogging all the video card interactions
[17:19] <lotuspsychje> sneakyimp: ok the 460 ignoring your older gpu makes sense right
[17:20] <lotuspsychje> maybe its a good idea as cbreak mentioned, check out software&sources see if you can set your older card on nouveau?
[17:20] <[VMGuy23]> how do check apt log
[17:24] <leftyfb> [VMGuy23]: the apt log is in /var/log/apt/history.log or /var/log/dpkg.log
[17:24] <sneakyimp> lotuspsychje: not sure what you mean by 'check out software & sources?' As I described in that forum post, I attempted using the control panel and it will NOT let me change the driver for the card. Given that the 1050 is helping speed up blender (a LOT), I'm worried about screwing up its configuration. Is it even worth the trouble to get the 470 working again?
[17:24] <[VMGuy23]> leftyfb: thanks
[17:24] <lotuspsychje> sneakyimp: software&sources/tab additional drivers check
[17:25] <sneakyimp> lotuspsychje: also, the nvcc --version is reported as 10.1.243, whereas the `nvidia-smi` output reports CUDA Version: 11.2 -- should I bother trying to improve this config?
[17:26] <lotuspsychje> sneakyimp: did you install nvidia driver/cuda from nvidia website or something?
[17:26] <sneakyimp> lotuspsychje: as I described in the forum post linked above, the software/additional drivers check *cannot* change the driver. the radio buttons are stuck on 'Continue using a manually installed river' https://i.stack.imgur.com/Y0QMC.png
[17:28] <lotuspsychje> sneakyimp: if it was me, i would start experimenting all combo's to make alive cbreak idea
[17:29] <nicomachus> hi all. I just did a fresh install of 20.10 and don't have any video in Firefox. I installed ffmpeg but still no video at all. Twitch, Youtube, DRM sites like Netflix, nothing. Any simple plugin/driver I might be overlooking?
[17:29] <sneakyimp> lotuspsychje: before i installed the 1050 card, i tried installing the CUDA toolkit for the 470 in a dozen different ways and every single way ended up installing the nvidia-driver-460, which hates the 470 and won't work with it. i wiped every thing and restored the 470 to the nvidia-driver-390 which is happy and bought the 1050. I rebooted the machine with both cards in it and the monitor plugged into the 470 and
[17:29] <sneakyimp> . I plugged the display into the 1050 and the display showed the login screen.
[17:30] <lotuspsychje> sneakyimp: the bigger GTX cards usualy benefit the latest drivers
[17:33] <lotuspsychje> sneakyimp: im curious what shows in ubuntu-drivers devices
[17:33] <nicomachus> oh hi lotuspsychje. nice to see you're still hanging around. :)
[17:33] <lotuspsychje> hey nicomachus long time no see mate
[17:33] <sneakyimp> lotuspsychje: i'm pretty sure I installed cuda toolkit using apt
[17:34] <lotuspsychje> sneakyimp: before playing around with cuda, i would focus on make both drivers work first
[17:35] <lotuspsychje> sneakyimp: in your case id purge all and try again vanilla, and test out the combos on both cards
[17:35] <sneakyimp> lotuspsychje: the entire point of the cards is to support blender rendering -- and the 470 has not been useful in that regard
[17:35] <lotuspsychje> nicomachus: is your video driver installed? might also need ubuntu-restricted-extras
[17:36] <sneakyimp> sneakyimp: i suspect, and would appreciate advice on this matter, that the best I can hope for from the 470 is to handle display duty while the 1050 actually does useful work helping blender.
[17:38] <sneakyimp> lotuspsychje: also 'purge all' and 'try again vanilla' are not very specific. remove the cards? try one first then the other? go back to nouveau? I expect that all the possible branches of installation sequences could be days of effort -- and I've already tried quite a few with the 470 by itself. Before installing the 1050, I had removed cuda and all the nvidia drivers, gone back to just the 390 proprietary driver
[17:38] <sneakyimp> tuck in the 1050 card, which immediately caused the 470 to stop working.
[17:39] <sneakyimp> can Ubuntu even support nvidia-driver-390 and nvidia-driver-460 at the same time?
[17:39] <sneakyimp> or are they mutually exclusive?
[17:39] <lotuspsychje> sneakyimp: every hardware 'should' load its own module i assume
[17:42] <sneakyimp> lotuspsychje: i suspect -- but don't know for sure -- that the apt/deb/ppa packages are constructed to replace the old nvidia-driver-390 with nvidia-driver-460, and other configuration says that cuda-toolkit needs 460.
[17:42] <nicomachus> lotuspsychje: yea, video drivers are installed. checking restricted extras
[17:43] <nicomachus> lotuspsychje: yep, I was missing restricted-extras. resolved now.
[17:43] <nicomachus> thanks pal
[17:43] <nicomachus> !cookie lotuspsychje
[17:43] <nicomachus> aw, who disabled that?
[17:44] <lotuspsychje> !cookie
[17:44] <lotuspsychje> tnx nicomachus
[17:44] <mort> so this is technically about ubuntu+1, but since 21.04 releases tomorrow, this feels relevant here: the nextcloud client (or qt libraries) was updated earlier this month, it made nextcloud segfault on startup, and it's still not fixed: https://bugs.launchpad.net/ubuntu/+source/nextcloud-desktop/+bug/1923053
[17:44] <lotuspsychje> sneakyimp: can you pastebin; ubuntu-drivers devices plz
[17:45] <mort> it seems like 21.04 will release with a straight-up broken crashes-at-startup nextcloud client? That's gonna impact a lot of people
[17:45] <lotuspsychje> mort: until final release please use #ubuntu+1 and #ubuntu-quality please
[17:46] <mort> alright, I'll come back here tomorrow if it's still an issue then :p
[17:49] <bigode> Hello folks!
[17:50] <bigode> Could someone help me list the available drives on my computer from command line please?
[17:50] <mort> You can run `df`
[17:51] <cbreak> bigode: type `lsblk`
[17:51] <cbreak> bigode: or `findmnt` if you only want mounted filesystems instead of actual drives
[17:51] <mort> right, df justs lists mounted drives
[17:53] <sneakyimp> lotuspsychje: ok this is pretty interesting. somehow old driver 340 is in use: https://paste.ubuntu.com/p/DFB5Qcbw2w/
[17:53] <bigode> I have a mini sd card which does't show
[17:57] <leftyfb> bigode: run: dmesg -Tw # in a terminal and watch it as you insert the SD card
[17:58] <bigode> leftyfb: dmesg: read kernel buffer failed: Operation not permitted
[17:58] <leftyfb> bigode: what version of ubuntu is this?
[17:59] <bigode> The last one
[17:59] <cbreak> bigode: sudo
[17:59] <leftyfb> bigode: please pastebin the output of:   cat /etc/os-release
[17:59] <leftyfb> bigode: cat /etc/os-release | nc termbin.com 9999
[18:01] <leftyfb> cbreak: I just checked on a fresh 20.04 desktop install and a 20.04 server install. dmesg shouldn't require sudo by default
[18:01] <bigode> leftyfb: https://paste.ubuntu.com/p/XhXybd5jtB/
[18:02] <bigode> leftyfb: it connects then disconnects without being mounted ever
[18:02] <leftyfb> bigode: os-release output?
[18:03] <cbreak> leftyfb: so normal users can access kernel log output? :O
[18:04] <bigode> leftyfb: https://paste.ubuntu.com/p/KmyCXvtKPW/
[18:04] <leftyfb> cbreak: feel free to try it on your machine
[18:05] <leftyfb> bigode: it says it's sdb. sudo disk -l /dev/sdb
[18:05] <bigode> Wow.. or https://termbin.com/45st
[18:06] <bigode> leftyfb: just disk command?
[18:06] <leftyfb> ugh
[18:07] <leftyfb> bigode: sorry sudo fdisk -l /dev/sdb
[18:07] <bigode> leftyfb: fdisk: cannot open /dev/sdb: No such file or directory
[18:07] <cbreak> if it's in dmesg, then lsblk should show it
[18:08] <bigode> leftyfb: I think it got disconnected
[18:09] <bigode> leftyfb: as USB disconnect, device number 8
[18:10] <leftyfb> bigode: if it's disconnecting on it's own, it's probably a hardware issue
[18:12] <bigode> leftyfb: when I use a old 1gb mini sd card it works just fine
[18:14] <aaroncossey> I encountered a weird security bug which I cant reproduce
[18:14] <leftyfb> bigode: ok, so definitely a hardware issue then
[18:15] <aaroncossey> from locked, when I typed my password into the login, the characters were visible as plain text instead of showing as dots
[18:16] <aaroncossey> any idea how I can try to reproduce/investigate to create some kind of bug report?
[18:16] <bigode> leftyfb: I have one old 8gb sd card which doesn`t mount and this new 64gb mini sd card. Same thing. But I have several old 1gb or 2gb which works great.
[18:17] <bigode> When you say it is a hardware issue it doesn`t mean there is a problem with the hardware but a certain limitation from sd card size?
[18:17] <leftyfb> bigode: ok, so maybe the card reader you have there doesn't support larger storage devices
[18:18] <bigode> leftyfb: yeah. Probably.. On windows the same happens. :/
[18:18] <bigode> Can I figure the hardware manufactor to check it?
[18:19] <leftyfb> bigode: sudo lshw
[18:23] <bigode> ould you help me identify it please? https://paste.ubuntu.com/p/3wCVd4qCVt/
[18:23] <bigode> Would*
[18:28] <p9> i've just updated docker.io and its service was renabled and started automatically despite I had it disabled
[18:28] <p9> why such things happen?
[18:28] <p9> should I mask everything to be sure?
[18:29] <Nelson89> Hello!
[18:30] <Nelson89> Have sameone idea, how fix it?
[18:30] <Nelson89> https://pasteimg.com/image/screenshot-2021-04-21-20-27-10.c8JkM
[18:30] <Nelson89> '-fix--broken install' doesn't work
[18:31] <Nelson89> Ubuntu 18.04.5
[18:34] <BlackComb>      https://itectec.com/ubuntu/ubuntu-ubuntu-boots-to-black-screen-with-cursor-after-upgrade-from-18-04-to-20-04/
[18:34] <sarnold> Nelson89: pastebin that error, too?
[18:34] <BlackComb> does this remove all the items on the desktop?
[18:34] <BlackComb> will i lose all my stuff on desktop
[18:36] <Nelson89> sarnold https://pastebin.com/bRMmPZKC
[18:36] <BlackComb> when I do journalctrl -xe. I got the following error: fail to start tracker metadata database store and look up manager
[18:36] <BlackComb> is that normal?
[18:37] <sarnold> Nelson89: heh, dang, I was hoping it'd give more details. have you do an apt update recently?
[18:37] <Nelson89> sarnold yes, I did
[18:38] <Nelson89> I installed today AnyLogic, and after this isntallation I have this issue
[18:38] <Nelson89> https://www.anylogic.com/
[18:38] <bigode> Thank you leftyfb! o/
[18:39] <sarnold> Nelson89: try apt-get install libicu60 ?
[18:40] <BlackComb> I'm using Ubuntu 20.04>
[18:40] <BlackComb> after reboot / crash my computer load into a black screen with blinking cursor
[18:40] <BlackComb> how can i fix this?
[18:41] <Nelson89> sarnold, yes
[18:41] <Nelson89> https://pastebin.com/X3ZQnBqz
[18:41] <Nelson89> last two positions
[18:43] <sarnold> Nelson89: I wanted to see what the error message would be from trying the install command
[18:44] <Nelson89> sarnold https://pastebin.com/Xt4EjPKn
[18:45] <sarnold> Nelson89: drat, going around in circles with apt, heh. okay, how about apt policy libicu60 ?
[18:46] <Nelson89> xD
[18:46] <Nelson89> https://pastebin.com/jhSvrxaS
[18:48] <sneakyimp> so is there no simple command to tell ubuntu to use a different driver for one of its video cards?
[18:48] <sneakyimp> while leaving the other video card unmolested?
[18:48] <sarnold> sneakyimp: most people use ubuntu-drivers install to install the video driver for their card when they install a different one
[18:49] <sarnold> Nelson89: try this then :) dpkg --get-selections | grep icu
[18:49] <sneakyimp> sarnold: my situation is more complex -- i have two dissimilar video cards ,one of which requires an older driver
[18:49] <Nelson89> sarnold https://pastebin.com/1fxW44ja
[18:50] <sneakyimp> sarnold: the whole sordid story is here: https://askubuntu.com/questions/1331925/possible-to-install-distinct-drivers-for-two-dissimilar-nvidia-gpus-on-ubuntu-20
[18:50] <BlackComb> I'm getting a message saying you don't have enough free space in /var/cache/apt/archives/
[18:51] <BlackComb> I've try doing sudo apt-get autoclean and sudo apt-get clean
[18:51] <sneakyimp> sarnold: TLDR version is that installing new 1050 card caused old 470 card to use an old driver, and I cannot change it in the software->additional drivers control panel: https://paste.ubuntu.com/p/DFB5Qcbw2w/
[18:51] <BlackComb> but still working
[18:52] <BlackComb> still not working*
[18:53] <sarnold> Nelson89: hrmph. I expected to see a 'hold' in there. :( I'm running out of ideas :(
[18:53] <sneakyimp> sarnold: before i installed the new card, the GTX 470 card was happily using nvidia-driver-390 (the latest recommended one for this rather old card) and the GTX 470 will NOT work with the latest nvidia-driver-460.
[18:54] <Nelson89> sarnold, Thanks for your efforts :)
[18:54] <sarnold> sneakyimp: nice question
[18:54] <sarnold> sneakyimp: I don';t know much about the nvidia drivers besides the baffling number of packages they've got :(
[18:55] <sarnold> sneakyimp: can you manually apt-get install both the -390 and the -460 packges at once?
[18:56] <sneakyimp> sarnold: it's a mess, and I've already spent two entire days. I'm thinking the old 470 might just be more trouble than it's worth. it's currently in the machine doing absolutely nothing. the whole point is to get blender working faster, and at best the 470 would handle display duties, allowing the GTX 1050 to handle rendering
[18:56] <sarnold> sneakyimp: it feels like an obvious use case to me, use the good card for rendering, use the okay card for web etc :)
[18:57] <sneakyimp> sarnold: that is precisely my question! that last paste i linked suggests that Ubuntu wants the old card to use nvidia-driver-340 which is REALLY OLD. the software->additional drivers control panel is *stuck* on 'continue using a manually installed driver' https://i.stack.imgur.com/Y0QMC.png
[18:57] <sneakyimp> sarnold: i'm glad to hear you think it sounds like an obvious use case. the 470 seems less than useless at the moment, just sucking up electricity
[19:04] <lotuspsychje> sneakyimp: try a few tests like we discussed before
[19:04] <lotuspsychje> sneakyimp: like what happens when you physically plug out your 1050, does it load the right driver for the older card?
[19:08] <sneakyimp> lotuspsychje: I was truly hoping we could just try some command to change the driver for the 470 card. The other card is doing an OK job. The 470 is useless at the moment. Did you see the ubuntu-drivers devices output? IT's noteworthy that it mentions the ancient 340 driver, but not the 390 driver. https://paste.ubuntu.com/p/DFB5Qcbw2w/
[19:09] <lotuspsychje> sneakyimp: apt is always right, if it wants to load 340 for that card
[19:11] <nukke> quick question: is it possible to use Xenial repositories in Bionic/18.04? I am currently upgrading some servers and have an application that hard-depends on Postgres 9.5, which is only available on 16.04.
[19:11] <sneakyimp> lotuspsyche: before the 1050 was in the machine, it used the nouveau driver by default. trying to install nvidia-cuda-toolkit using apt ended up loading the 460 driver, which *hates* the 470 card. They won't work together. That's what led to the purchase of the 1050 in the first place.
[19:11] <nukke> I was planning on adding the `xenial-security` repository once the server is upgraded to 18.04
[19:12] <sneakyimp> lotuspsychje: and I want to say that apt previously chose the 390 driver for that card -- I certainly didn't.
[19:13] <lotuspsychje> sneakyimp: but why are you using cuda for?
[19:13] <sneakyimp> lotuspsychje: blender
[19:14] <lotuspsychje> sneakyimp: i would first try make both drivers work before playing with cuda
[19:14] <sneakyimp> lotuspsychje: and the 1050 is playing nicely with blender -- however, apt apparently chose CUDA 10.1 for nvcc and 11.2 for the video card (as described in my askubuntu post)
[19:14] <lotuspsychje> sneakyimp: in an ideal world, ubuntu should install and at least recognize both cards right?
[19:15] <sarnold> nukke: you'd probably be better suited to spinning up xenial in a vm or lxd instance or something similar
[19:15] <summonner> nukke, I don't have an answer for you, but if you could get your hands on the .deb file then you could manually install via  sudo dpkg -i <file>.deb
[19:16] <sneakyimp> lotuspsychje: yes, in an ideal world, but i would also point out that my primary goal is to get blender working as fast as possible. trying every conceivable server configuration is starting to sound like diminishing returns
[19:16] <lotuspsychje> sneakyimp: try to work systematic, before thinking of blender getting drivers to work first
[19:17] <sneakyimp> lotuspsychje: btw, i am currently working on the machine in question...if i have to start shutting it down, swapping cards, removing drivers, etc. (the commands of which are not something i'm very familiar with)...you can imagine that it starts to become a giant chore
[19:19] <sneakyimp> surely there is some command which will instruct ubuntu to use a different video driver for the 470 card, while leaving the 1050 card untouched?
[19:26] <Nelson89> sarnold, I found the solution xD
[19:27] <Nelson89> I have to use this "sudo apt-get install libicu60=60.2-3ubuntu3.1
[19:27] <Nelson89> "
[19:28] <Nelson89> thanks and bye!
[19:31] <sneakyimp> lotuspsychje: OK so I took out the 1050 card, plugged the monitor into the gtx 470, rebooted the machine, and now my screen resolution is stuck at 1024x768.
[19:32] <sneakyimp> lotuspsycheje: the nvidia-smi command doesn't recognize the 470 card
[19:33] <sarnold> o_O
[19:33] <sneakyimp> lotuspsychje: ubuntu-drivers devices now shows the 390 driver https://paste.ubuntu.com/p/7M6VQKdXNt/
[19:35] <sneakyimp> lotuspsychje: lshw indicates the video card is unclaimed. https://paste.ubuntu.com/p/SVQN7CMyTC/
[19:36] <sneakyimp> lotuspsychje: the software->additional drivers control panel is *still* stuck on 'Continue using a manually installed driver'
[19:38] <sneakyimp> calling `sudo ubuntu-drivers autoinstall` just says unmet dependencies, broken packages blah blah: https://paste.ubuntu.com/p/C8XzR3Km8S/
[19:38] <sneakyimp> so that's what happens when i take out the 1050 card
[19:41] <oerheks> GeForce GTX 470 is pretty old, use the 304driver?
[19:41] <sneakyimp> lotuspsychje: sudo apt install nvidia-driver-390 REMOVES the 460 driver, as I speculated before: https://paste.ubuntu.com/p/yGcjhvKCMn/
[19:41] <oerheks> and one wants to be updated before that, sudo apt update && sudo apt full-upgrade # and reboot
[19:43] <sarnold> sneakyimp: argh, with a single 'nvidia.ko' kernel module, it seems unlikely to me that they're prepared for two versions of drivers to live side-by-side :(
[19:43] <sneakyimp> sarnold: this is essentially my question
[19:43] <sneakyimp> 'can you have two distinct nvidia drivers for dissimilar video cards at the same time'
[19:44] <sneakyimp> i have rebooted now, after installing 390, and the old 470 card is happily displaying at the proper 1920x1080 resolution
[19:44] <sneakyimp> lshw no longer looks unclaimed
[19:44] <nukke> summonner, sarnold: the upgrades will be done in-place with `do-release-upgrade`. my fear is that apt will remove that obsolete package even thoguh it's being used by an application
[19:44] <sneakyimp> nvidia-smi reports the 470 as using the 390 driver. no CUDA version is reported
[19:44] <sarnold> nukke: yeah, it probably will
[19:45] <sneakyimp> `nvcc --version` complains that the command is not found
[19:45] <nukke> sarnold: does apt have a "keep $package indefinitely" option?
[19:45] <mmlj4> is this normal on a new install? I don't see why k8ts, micro or not, needs to run on my linux box   # Apr 21 14:19:50 aslab1 microk8s.daemon-containerd[1945]: time="2021-04-21T14:19:50.638402165-05:00" level=info msg="Exec process \"a0b2e7892b18e251f36d74f0f4dfa8fe45ac53f92dac4f0c26ef6ab9f669bc7b\" exits with exit code 0 and error <nil>"
[19:45] <nukke> nevermind, seems like `apt-mark hold` would do the trick perhaps
[19:47] <sneakyimp> OK so here's a more general question: what is the advantage, if any of using a proprietary nvidia video driver versus using nouveau?
[19:47] <sarnold> nukke: it does, but it blows apart your upgrade path
[19:48] <sarnold> nukke: which might not be so bad, you can just leave it on 16.04 and get an ESM subscription in a few days or something
[19:48] <sarnold> mmlj4: if you don't need it, uninstall it? how'd you get it?
[19:48] <nukke> sarnold: there is no upgrade path for now, unless you mean it would hold up the entire 18.04 upgrade. ESM would be an option though.
[19:49] <mmlj4> sarnold: a junior tech did the install, I have no idea what he clicked on
[19:49] <sarnold> nukke: yeah, I'm pretty sure I've seen bugreports filed by folks who had held packages that scuttled their do-release-upgrade
[19:50] <sarnold> mmlj4: ah; try snap list, that'll probably show microk8s; if it does, snap remove or snap delete or something similar can get rid of it
[19:52] <mmlj4> ok, so what is snap? sorry, I'm not a group hug user
[19:52] <summonner> a do-release-upgrade is like a brand new install. There are c libraries which are updated, your apps most likely won't work against the previous c library
[19:52] <summonner> *without the previous c library
[19:53] <sarnold> mmlj4: snaps are kinda like dockers, for applications
[19:53] <sneakyimp> summonner: not exactly...your sources list seems to keep some cruft from the old OS
[19:53] <summonner> nukke, you'd do better testing this against a non-production system where total loss is not an issue
[19:53] <sneakyimp> probably other cruft, too
[19:53] <sarnold> mmlj4: the server installer offers a bunch of click boxes when installing to add a few of them, and microk8s is one of them
[19:54] <nukke> summonner: I will be taking snapshots, and we have daily backups. I'm still scared about the upgrade.
[19:54] <mmlj4> sarnold: ah, yes, docker/k-hates containers, gotcha
[19:54] <summonner> nukke, have you tested the restore before? restore to another piece of hardware. if that works, go for it.
[19:55] <sneakyimp> is there a log of this irc chat somewhere on the internet?
[19:55] <nukke> summonner: restore what? the database?
[19:55] <summonner> nukke, whatever you're going to upgrade
[19:55] <sarnold> sneakyimp: https://irclogs.ubuntu.com/2021/04/21/%23ubuntu.html
[19:56] <sneakyimp> sarnold: so tell me straight? you think it's a lost cause trying to get both of these cards working with different proprietary drivers?
[19:56] <sneakyimp> sarnold: and thanks for your kind help btw
[19:56] <sarnold> sneakyimp: well.. you *could* try running a VM for blender, and do pci pass through of the card to the VM
[19:57] <sarnold> sneakyimp: nvidia goes to some effort to prevent that from working, but arch folks have some hints on how to make it work
[19:57] <sneakyimp> sarnold: ARG lol what a mess
[19:57] <sarnold> sneakyimp: https://wiki.archlinux.org/index.php/PCI_passthrough_via_OVMF#Video_card_driver_virtualisation_detection
[19:57] <sneakyimp> sarnold: all that overhead of the VM just so i can use the 470 to handle display
[19:57] <sarnold> yeah :/
[19:57] <sneakyimp> sarnold: somehow i think the overhead of a VM would cancel out any potential gains
[19:58] <sneakyimp> sarnold: i'm not attached to the 470. i could probably get another 1050 for $300
[19:58] <sarnold> sneakyimp: hard to guess, but the time to find out if it works.. *sigh
[19:58] <freebench__> hey can anyone advice simple smtp to run locally ?
[19:59] <sneakyimp> sarnold: the time i've wasted trying to get them to play nice together is already past the $300 mark ;)
[19:59] <freebench__> for email  sending testing
[20:00] <sneakyimp> freebench__ : it's been my experience that mail is NOT simple...this is based on postfix configuration attempts. why do you want to 'run stmp?' are you trying to send mail from somewhere on your machine?
[20:00] <troozers> Does anyone know of a good tutorial on how to implement gpu/egpu switching for Wayland - preparing for installation of Ubutu 21.04
[20:01] <freebench__> sneakyimp, agreed, not simple. it would do like this - an app use local smtp server to send email
[20:01] <sneakyimp> freebench__: is the app trying to use sendmail or something? you're getting an error?
[20:01] <freebench__> no authentication, nothing advanced, simplier - better
[20:02] <sneakyimp> the app wants to send outgoing mail but there's no MTA there to provide the outgoing mail service, eys?
[20:02] <sneakyimp> yes?
[20:03] <freebench__> the app just needs to send email to a recipient. that is it.
[20:03] <freebench__> via python's libsmtp
[20:03] <sarnold> sneakyimp: yeah, that was a fantastic lesson from my first manager.. the amount of time it takes to do something vs buying your way out of the problem..
[20:04] <sarnold> freebench__: look at msmtp
[20:04] <sneakyimp> can you configure the app with credentials to use some smtp gateway? e.g., if you were writing a PHP project you could use PHPMailer to connect to gmail to send the app
[20:04] <sarnold> freebench__: it requires an online mail server, and it will *block* while sending mail, so it might not be a great choice for an API like that, but it works great for me with mutt
[20:05] <sneakyimp> sarnold: the fact of the matter is I don't really *need* both these cards to work. the new 1050 has sped up rendering about 4x faster already. i just wanted to know if i could keep the 470 working to make things a tiny bit faster perhaps
[20:05] <sarnold> sneakyimp: and hey if you already have it, let it keep helping out :)
[20:05] <sarnold> bummer that it didn't work :(
[20:06] <sneakyimp> sarnold: i'm gonna try one last thing: configure 470 to use nouveau (just for display, after all) and configure proprietary nvidia drivers for 1050
[20:07] <sneakyimp> brb
[20:07] <sarnold> sneakyimp: OH!
[20:07] <sarnold> sneakyimp: excellent idea :D
[20:30] <irgendwer4711> hi, when will Thunderbird 78.8.x or 78.9. appear for Ubuntu 20.10?
[20:30] <[VMGuy23]> When will 21.04 come out
[20:31] <sarnold> [VMGuy23]: probably tomorrow
[20:31] <troozers> tomorrow I believe, due for release 22.April last I heard
[20:31] <[VMGuy23]> Nice, I've been waiting for it
[20:34] <sneakyimp> sarnold: no dice on nouveau driver
[20:34] <sarnold> sneakyimp: daaaang :( I had hopes, apparently unfounded :( thanks for reporting back
[20:36] <leftyfb> irgendwer4711: thunderbird 78.10.0 is available as a snap
[20:39] <sneakyimp> i removed 1050 card. 470 card was in some weird state, only supporting 1024x768. Software & Updates->Additional Drivers will not let me change the driver in that control panel, stuck on 'continue using proprietary'. So, I did `sudo apt install nvidia-driver-390` and rebooted. Computer back to ground zero from last week, 1920x1080, looks great, I can also change the driver now so I change it to nouveau in the contr
[20:39] <sneakyimp> t 470 card in machine, running nouveau driver, looks pretty good. Some command line details:  https://paste.ubuntu.com/p/gBvK9m3p7N/
[20:42] <sneakyimp> sarnold: i shut the machine down, put the 1050 card back in, reboot. Both cards in machine, monitor plugged into 470 card. Machine boots, i can see screen via 470 card. VERY SLUGGISH. Both cards running nouveau driver and machine is just really slow. I go to softare & updates -> additional drivers and am able to select nvidia-driver-460 for the 1050 card without changing anything for the 470 card. Reboot.
[20:43] <irgendwer4711> leftyfb: I dont wanna use snap
[20:44] <sneakyimp> sarnold: monitor still connected to 470, machine boots. NO SCREEN. 470 cannot show video for some reason. unplug monitor from 470 and connect to 1050 and VOILA screen comes up, but once again the 470 card is UNCLAIMED. Back to where I was when I showed up here today, except my machine complains nvcc not found, must install cuda toolkit. HOWEVER the nvidia-smi output says the card speaks CUDA 11.2. https://paste.ub
[20:44] <sneakyimp> total mess
[20:44] <leftyfb> irgendwer4711: then you'll have to wait till the maintainers update the package
[20:45] <irgendwer4711> sneakyimp: there are new packets for weeks, but nobody puts them into update channel
[20:45] <sarnold> sneakyimp: heh, your client is *trying* to wrap long messages across multiple messages, but it's failing :( the last bit from the most recent long message ends at "https://paste.ub" -- is this message length configurable in your client/
[20:45] <ogra> irgendwer4711, because all developers are busy doing the 21.04 release
[20:46] <ogra> use the snp or wait ...
[20:46] <ogra> *snap
[20:46] <irgendwer4711> I removed all snap shit...
[20:46] <ogra> poor you ...
[20:46] <irgendwer4711> no. poor Ubuntu
[20:47] <leftyfb> snaps work fine for me. I'm running the thunderbird snap among others just fine
[20:48]  * ogra wouldnt want to use a non snapped browser anymore ... i value my data and privacy 😉
[20:49] <irgendwer4711> ogra privacy???
[20:49] <ogra> yep
[20:50] <irgendwer4711> explain
[20:50] <ogra> snap confinement massively restricts what a binary can do, see or access
[20:51] <ogra> with the snapped firefox there is no way it ever accesses ~/.ssh or ~/.config to steal data .. even if it has some security bug that allows some javascript to access this
[20:51] <leftyfb> with the more frequent updates, security and privacy are usually kept up to date better as well
[20:51] <sneakyimp> sarnold: hmmm...using pidgin. other client recs welcome. link here: https://paste.ubuntu.com/p/yf844p5bDn/
[20:52] <ogra> if you are paranoid you can actually also "snap disconnect firefox:home" and it wont even be abe to access and docs
[20:52] <ogra> *any
[20:52] <irgendwer4711> orga: haha funny, apparmor do this already
[20:52] <sarnold> sneakyimp: irssi here, but it's pretty different from pidgin :) heh
[20:52] <ogra> not OOTB
[20:53] <ogra> you'd have to do a lot f manual tinkering to get the same behaviour in the non-snapped FF
[20:53] <irgendwer4711> I have a profile for FF, shouldnt be a problem for TB.
[20:53] <ogra> and snapd dont use apparmor alone, confinement uses sccomp (no binary executions of any stuff that isnt explicitly allowed), namespaces etc etc
[20:53] <ogra> *snaps
[20:54] <irgendwer4711> I have other protection
[20:54] <ogra> there is a lot more involved than apparmor, thats only a very minor part
[20:54] <ogra> sure
[20:54] <ogra> but does your mom (without you setting it up) ?
[20:55] <irgendwer4711> my mom is dead. moron!
[20:55] <ogra> snaps are a way to get better security to everyone ... even my and your mom
[21:06]  * summonner grabs the popcorn
[21:06] <ogra> summonner, to late (s)he went ...
[21:07] <el> oh
[21:07] <el> +j wasn't set i was fixing it
[21:10] <grimpo> When connecting via ssh to a host for the first time (i.e. not in the known_hosts file), the client in 20.04 now gives you the options of "yes / no / [fingerprint]". What is the difference between choosing Yes and choosing Fingerprint?
[21:10] <dax> grimpo: [fingerprint] indicates that if you paste a fingerprint into the prompt it will check to see if it matches what it's seeing
[21:11] <dax> e.g. if you were given the fingerprint out-of-band somehow
[21:11] <sarnold> oh that's awesome
[21:11] <sarnold> no more pretending that checking the first and last N chars is good enough :)
[21:11] <dax> indeed :)
[21:33] <ponch> Anyone familiar with Ghostscript and how to either downgrade to 9.27 from 9.50, or reinstall with -dNOSAFER?
[21:33] <catphish> is there a standard way to load iptables rules at boot?
[21:33] <ponch> but after adding argument -dNOSAFER, I would need git stasch, not sure how. :)
[21:35] <catphish> i think iptables-persistent might be what i nee
[21:35] <catphish> *need
[21:39] <oerheks> ponch,  nope, https://launchpad.net/ubuntu/+source/ghostscript does not give that version, install by building manually perhaps? good luck there https://launchpad.net/ubuntu/+source/ghostscript/9.27~dfsg+0-0ubuntu3.1
[21:41] <sarnold> ponch: many imagemagick options can be configured by editing your local site policy
[21:41] <sarnold> ponch: I don't know what specifically can and can't be changed that way; it's possible you might be in 'compile from source' territory
[21:41] <sneakyimp> these nvidia CUDA installations say i should run a process on boot...can anyone suggest the best way to make sure this happens for ubuntu 20.04?
[21:41] <sneakyimp> https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#install-persistenced
[21:42] <sarnold> ponch: I'd like to suggestyou use apparmor or another mandatory access control system to confine what your ghostscript can do; it's had hundreds of vulnerabilities over the years, and if you're feeding potentially untrusted inputs into it, you need to take steps to protect your system from your users
[21:43] <oerheks> sarnold, not sure what the issue is..
[21:44] <sarnold> sneakyimp: heh, I went looking for a systemd service file for it, and found some complaints that it's hard to use https://github.com/NVIDIA/nvidia-persistenced/issues/2 ..
[21:44] <sarnold> sneakyimp: maybe you can find a .service file that just works, but if not, then adding a @reboot line to a crontab wouldn't be so horrible.
[21:49] <sneakyimp> sarnold: thanks for looking. that post is 2 yrs old. I'm now running CUDA 11.3. Hopefully fixed? process just ran for me from CLI without complaint...still working thru post install
[22:02] <today911> Hey! I'm just checking the proprties menu of a usb drive in nautilus and I can see file system type: FUSE (I did google a bit, but have never seen anything like such... it should be NTFS as far as I know); anyone could give me two words of explanation, please? :)
[22:03] <today911> and, why are some usb drives automatically mounted under /mnt and others under /media ?
[22:06] <leftyfb> today911: fuse is what ubuntu uses to mount ntfs filesystems. It's not the filesystem of the drive
[22:07] <leftyfb> today911: I've never seen any usb storage device mounted under /mnt automatically unless purposely specified in fstab or a udev rule
[22:08] <today911> leftyfb: thank you!
[22:08] <today911> well, I believe nautilus somehow gets the wrong info displayed then as it was confusing for a few seconds (until I opened up disks which showed NTFS correctly)
[22:10] <today911> and regarding the /mnt you are very right, I also thought everything gets mounted under /media ... but somehow... this usb device I am using right now, gets mounted under /mnt automatically... could it be anything related to some drivers that are mapped to one or the other mount-point?
[22:22] <kill-animals> hey I running kubuntu on a thinkpad x290 yoga, and when I try to use my pen, it causes a system crash, and I have to force restart
[22:36] <kill-animals> how can I find the logs for this?
[22:37] <sarnold> kill-animals: journalctl   will show most
[22:37] <kill-animals> sarnold: thanks.
[22:37] <sarnold> check also for ~/.xsession-errors, maybe your system is set up to send stderr from gui programs there
[22:42] <kill-animals> sarnold: I am going through journalctl right now.
[22:45] <kill-animals> tsk. Can't really find anything obvious.
[23:00] <kill-animals> how do I update my kernel and nothing else?
[23:02] <matsaman> kill-animals: https://askubuntu.com/questions/44122/how-to-upgrade-a-single-package-using-apt-get
[23:27] <woodonia> So, I'm stupid, and pretty sure I know the answer to this, but:  I was playing around with 21.04 daily images, and went with encrypted ZFS.  Unfortunately, my witty attempt at a good passphrase (in XKCD style) has bitten my behind, as I can't remember all of the passphrase.  For some reason I did remember to record the recovery key during
[23:27] <woodonia> installation... but didn't write down my passphrase.  Is there any way to recover the data?  (the data is all backed up and all readily replaceable, but I've been using this as a learning exercise, and have hit nothing but dead ends)
[23:29] <matsaman> that's the purpose of the recovery key yes?
[23:29] <sarnold> oo encrypted zfs is in the installer now?
[23:30] <woodonia> Yes... but I'm not sure how to use the recovery key
[23:30] <shibboleth> well, that depends on zfsutils version
[23:31] <shibboleth> which, i'm certain sarnold knows for sure, more surprised at how it wasn't part of 2004
[23:32] <sarnold> heh, I haven't actually used zfs encryption yet myself :(
[23:32] <shibboleth> well, it's "interesting" and i can see use-cases where it'd be neat
[23:32] <woodonia> Next time I need to write it down... witty XKCD style passphrases at 2am are much harder to remember days later, that are only used on boot...
[23:33] <shibboleth> still, one-ring-to-rule-them-all like systemd isn't necessarily the best choice
[23:33] <sarnold> yeah, it'd be cool to be able to eg unlock a dataset for taxes once a year at tax time, but otherwise keep them unmounted and locked up..
[23:33] <shibboleth> you can do luks on the zpool and luks on say.. child zvols
[23:34] <shibboleth> one drawback of zfs is how royally screwed you are when it breaks
[23:35] <woodonia> Snapshots and zfs send/receive are pretty nice features, though...
[23:36] <shibboleth> btrfs does snapshots better
[23:36] <shibboleth> also offers send/receive
[23:38] <shibboleth> still, i'm not saying recovering broken btrfs is less painful
[23:38] <woodonia> I played with it for a bit on Arch, but ended up getting bitten by one of the kernel bugs.  I still mostly use ext4, but figured i'd play with ZFS
[23:38] <shibboleth> luks operates transparently though, far easier to find tools to recovery zpools behind luks than tools supporting zfsenc
[23:39] <shibboleth> i totally get zfs for "deep/big" storage, but for system drive/volume? hell noes
[23:40] <woodonia> Yeah - I use it on my NAS, and figured I'd try playing with send/receive of /home data sets
[23:41] <woodonia> appears to be native encryption - blkid shows the partition type = "zfs_member" instead of LUKS
[23:47] <cbreak> native zfs encryption doesn't support recovery keys
[23:47] <cbreak> you either have a key file, or a passphrase, or similar.
[23:48] <cbreak> I think ubuntu works around this by having an encrypted partition (non-zfs) that contains the zfs key file
[23:48] <cbreak> you can try to find and unlock that thing
[23:48] <woodonia> hmm
[23:49] <va7lnx> oops
[23:50] <woodonia> I remember there being a keystore dataset... let me see
[23:51] <woodonia> Yup - rpool/keystore
[23:54] <woodonia> Urgh, but it won't let me set the mountpoint of the keystore dataset
[23:55] <cbreak> what for?
[23:56] <cbreak> and who is "it"?
[23:56] <woodonia> Oh
[23:57] <woodonia> I managed to forget my passphrase, but saved my recovery key - I'm trying to figure out how to use the recovery key to recover my data
[23:57] <cbreak> how does that require changing the mountpoint?
[23:58] <woodonia> the rpool/keystore doesn't have a mountpoint when I import the pool
[23:58] <woodonia> I assumed I'd have to mount it...
[23:59] <cbreak> zfs get mountpoint rpool/keystore