| pragmaticenigma | sybariten: would think that .bash_history might be able to help with that | 00:03 |
|---|---|---|
| sybariten | pragmaticenigma: that's the command history? | 00:04 |
| pragmaticenigma | yes | 00:06 |
| pragmaticenigma | sybariten: so with something like `grep ^z .bash_history` I'd think you'd be able to get a listing of what was executed. And there's probably a way to remove all but the most recent instance of any given line | 00:07 |
| pragmaticenigma | probably `grep '^z ' .bash_history` would be more precise | 00:08 |
| kubast2 | Yeah after an upgrade to 23.10 there's no issues with plasma-discover crashing with an active flapak repo | 00:09 |
| pragmaticenigma | I thought zoxide looked pretty cool at first, but I go to too many random locations, and didn't think the way it presented options when given ambiguous paths | 00:10 |
| sybariten | pragmaticenigma: bash command history != bash pwd history | 00:13 |
| pragmaticenigma | sybariten: I'm aware... but all the commands issued to zoxide can be greped out from there. Unless you're looking for the full expanded paths that zoxide emitted. | 00:15 |
| sybariten | pragmaticenigma: i'm not sure i follow... i'm looking for the visited directory locations (paths). Like a history. Exactly like a shell history, but with locations instead | 00:18 |
| pragmaticenigma | I think we're thinking the same things... I'm just not articulating the idea well | 00:19 |
| pragmaticenigma | zoxide doesn't have an export feature of what it's executed. I was suggesting that maybe bash history could help in that, except you would only see the command `z foo` that you entered, not the fully expanded/qualified path to the destination. | 00:20 |
| tomreyn | a bash history record such as "cd .." or "cd $HOME" would not return what was the absolute path at the time, but that's (I'm guessing) what sybariten is looking for. | 00:21 |
| pragmaticenigma | ships crossing in the night tomreyn :) | 00:21 |
| pragmaticenigma | sybariten: I think you have a very good idea for a feature request | 00:24 |
| tomreyn | i'm guessing that auditd could record and provide such information | 00:25 |
| sarnold | oof that's a heavyweight tool for that :) | 00:27 |
| tomreyn | yes. but... is there something smaller which provides absolute paths and isn't easily falsified? | 00:28 |
| sarnold | I wonder if you could get notified of /proc/pid/cwd updates? | 00:32 |
| tomreyn | inotify :) | 00:33 |
| sarnold | yeah, does it work for these magiclinks? | 00:33 |
| tomreyn | i never tried | 00:34 |
| sybariten | pragmaticenigma: well it does have a query command and i'm suspecting that this could be helpful | 00:43 |
| === antonispgs8 is now known as antonispgs | ||
| === JanC is now known as Guest6621 | ||
| jadeia | Quick question on upgrading - can I go from 23.10 > 24.01 LTS without a re-install? | 03:04 |
| jadeia | or do I have to stay on LTS? | 03:05 |
| xangua | Yes you can | 03:10 |
| xangua | 24.04.1 | 03:10 |
| jadeia | Perfect - I thought so, wanted to confirm. Sorry yep messed up that Ver number. | 03:12 |
| jadeia | Thank you xangua | 03:12 |
| jadeia | Prob going to hop to the LTS so I can check out Gnome46 and then go back to the quicker release version when it comes out. Very excited for this release of Ubuntu. | 03:13 |
| andymandias | you can download a deb from https://www.google.com/chrome/ | 04:29 |
| bind | http://explorer.net.ru/share/20%20years.png منذ 20 عاما... توصلنا إلى رمز "الرقم السري" ، فهذا يعني النزاهة | 05:48 |
| === bleb_ is now known as bleb | ||
| === JanC is now known as Guest8427 | ||
| === JanC_ is now known as JanC | ||
| |N3on21| | Hello world - I am having some problems when streaming video (youtube) music (spotify) my audio drops out for two sec. and then gets back.... it is starting to get on my nerves now. My setup is a labtop with a dockingstation (usbc) and a external monitor from where i would like to get my audio output from. Any ideas where to look ? | 08:04 |
| eleve_ | salut | 08:24 |
| ice9 | how to paste text in Vinagre remote desktop viewer? or do you recommend another VNC client what can do that? | 09:34 |
| brian__ | yes | 09:39 |
| stenno | vnc itself doesn't really allow for that iirc | 09:39 |
| stenno | it's basically an interactive picture of the screen | 09:39 |
| brian__ | Yeaah I recommend that | 09:39 |
| === antonispgs5 is now known as antonispgs | ||
| brian__ | I have a question | 09:40 |
| stenno | the price for platform independence, would be my guess | 09:40 |
| stenno | no pm please | 09:40 |
| stenno | ... | 09:42 |
| Guest71 | Hi, I am running ubuntu 23.10. When executing `apt upgrade`, I see the results `The following packages have been kept back: gir1.2-mutter-13 libmutter-13-0 mutter-common mutter-common-bin`. Why? | 09:54 |
| jeremy31 | Guest71: Likely phased updates | 09:55 |
| Guest71 | jeremy31: thanks a lot. I just googled. So updates to a subset of users? Everyone will eventually updated those packages when it releases to everyone. is that right? | 09:59 |
| lotuspsychje | !phasedupdates | 10:01 |
| ubottu | Since Ubuntu 21.04, APT may hold back some updates on some systems while they are being phased in. This is called "phased updates". See https://ubottu.com/y/phased and https://help.ubuntu.com/community/PhasedUpdates for more info. | 10:01 |
| Guest71 | ok, got it. thanks. | 10:04 |
| Guest71 | Another issue is I have met three times that ubuntu windows manager or UI? broke. There is nothing I can do. The only thing I can think of is to reboot. Black screen. I am not sure what the cause is. Maybe because I suspended my laptop for multiple times? or maybe because I ran virtualbox at the same time when I suspend? The system (or the UI | 10:06 |
| Guest71 | system) breaks randomly. I never shutdown my laptop before. But now, ubuntu 23.10 forces me to reboot it in several days. | 10:06 |
| Guest71 | I don't know what broke. I used the word windows manager or UI. maybe wayland? | 10:09 |
| Guest71 | ctrl+alt+f1 and then relogin, does not bring the UI back. | 10:09 |
| Guest71 | normally, I do ctrl+alt+f3 and then login in text mode and then sudo reboot. | 10:09 |
| lotuspsychje | if you mean gnome shell Guest71 check your /var/crash dir see if crashes exist there | 10:10 |
| Guest71 | lotuspsychje: there are two files in that dir. | 10:12 |
| lotuspsychje | what are the called | 10:13 |
| Guest71 | _usr_bin_file-roller.1000.crash | 10:13 |
| Guest71 | _usr_lib_virtualbox_VirtualBoxVM.0.crash | 10:13 |
| lotuspsychje | no gnome shell crash then | 10:13 |
| Guest71 | two files are created today. 20 minutes differences between them. | 10:13 |
| Guest71 | But the system only broke once today. | 10:14 |
| lotuspsychje | but you mentioned virtualbox issue | 10:14 |
| Guest71 | two more times in other days. | 10:14 |
| Guest71 | I guess might because of virtualbox. I don't know. just a wild guess. | 10:14 |
| Guest71 | no glue. broke randomly. | 10:15 |
| Guest71 | mayeb caused by chrome. Because I am opening youtube at that time. | 10:15 |
| lotuspsychje | Guest71: maybe checkout your full journal logs to see what happened | 10:16 |
| Guest71 | lotuspsychje: thanks for the help. Gotta go now. I will check the log later today. | 10:17 |
| lotuspsychje | kk | 10:17 |
| lotuspsychje | good luck | 10:17 |
| loswedseded | will nmcli connection set wlp2s0 down disable automatic wifi activation after booting permanently? | 10:18 |
| === antonispgs0 is now known as antonispgs | ||
| nteodosi | loswedseded, that is not a valid command for nmcli. | 10:42 |
| === pushkarnk1 is now known as pushkarnk | ||
| xx | is this official? https://github.com/jspw/Ubuntu-Launcher | 12:19 |
| xx | it has ubuntu in the name | 12:19 |
| loswedseded | what command can I use to disable automatic wifi activation after booting permanently? | 12:28 |
| CosmicDJ | xx: anything official is under https://github.com/ubuntu/ or the main web site | 12:29 |
| xx | can some app even use 'ubuntu' in the name, legally, if it's not made by ubuntu? | 12:29 |
| loswedseded | would this work with ubuntu 23.10? https://askubuntu.com/questions/1445221/permanently-disable-network-interface-in-ubuntu-22-04 | 12:36 |
| loswedseded | Im about to copy that code and delete the default config | 12:36 |
| loswedseded | talking about: network: version: 2 ethernets: interface: activation-mode: off | 12:36 |
| BluesKaj | Hi all | 12:37 |
| loswedseded | join #ubuntu-offtopic | 12:38 |
| CosmicDJ | xx: please talk to a lawyer if you have legal questions | 12:44 |
| xx | CosmicDJ: I mean, my lawyer said I can even create a linux distro called Ubuntu, using all the official branding, because I'm not in a jurisdiction where canonical could do anything. So that doesn't really help. | 12:48 |
| pragmaticenigma | xx: If you're talking with a lawyer, you should continue to ask them these kinds of questions. This is a community channel run by volunteers. Very few, if any, here are able to speak on behalf of Canonical. If you really want the authoritative answers to your questions, you should probably direct your questions to legal@canonical.com | 13:10 |
| sia_ | xx: Why you may wanna want to use the name ubuntu | 13:22 |
| ogra_ | loswedseded, you might want to use rfkill to turn off the wifi chip altogether ... | 13:29 |
| xx | sia_: because 'Ubuntu App' sounds better than 'App for Ubuntu' | 13:33 |
| sia_ | Sure but with a different meaning. As the first one can mean ubuntu itself which is an app. | 13:34 |
| loswedseded | ogra_, but then I can restart it with sudo ip link set wlp2s0 up and sudo systemctl start NetworkManager.service, right? | 13:34 |
| ogra_ | loswedseded, you might need to call "rfkill unblock all" before that | 13:37 |
| loswedseded | ogra_, does this command survive rebooting? | 13:38 |
| ogra_ | not sure, it should (it is the backend of what you know as "airplan mode") | 13:39 |
| ogra_ | but better try it first 🙂 | 13:39 |
| === EoflaOE- is now known as EoflaOE | ||
| === PinkMusen is now known as Kristine | ||
| pycurious | What is a good solution to backup a running ubuntu machine to a state (say ./backup now "mystate"). Then I can say (./restore from "mystate") - and reboot the machine. And I want the ubuntu to roll back to the backup state? Is there an easy to use and reliable solution to this problem? | 15:06 |
| leftyfb | run it in a VM or container | 15:07 |
| pycurious | I need a VM - which VM ? Virtualbox doesnt pass through | 15:07 |
| leftyfb | "doesnt pass though" what do you mean? | 15:08 |
| pycurious | I cant run VM or container, because I cant find an easy to use solution for multiple GPUs to pass through them. | 15:09 |
| === JanC is now known as Guest3663 | ||
| === JanC_ is now known as JanC | ||
| pycurious | that is why I'm looking for a bare-metal solution for the problem | 15:09 |
| leftyfb | pycurious: why do you need to "roll back"? | 15:09 |
| pycurious | leftyfb: Why do you have to reinstall a machine from scratch? To reset the state? | 15:10 |
| leftyfb | pycurious: I do that maybe every 2 years or so to get onto the latest LTS release. I reinstall Ubuntu from scratch and then run an ansible playbook that brings my machine back to the state I prefer it in and restores my files from backup | 15:11 |
| pycurious | @leftyfb I'm testing a software on this machine - which needs a lot of installs. I currently do it in the cloud by creating a new machine from scratch every hour. I want to reduce the cost and do it in my own machine. | 15:12 |
| leftyfb | you pay for GPU's in the cloud? | 15:12 |
| pycurious | https://github.com/bryansteiner/gpu-passthrough-tutorial/ - this is what i've to go through - if i want to do it in a VM | 15:12 |
| pycurious | @leftyfb can we go back to the problem - is there a backup/restore cli that is easy to use on ubuntu, that can reset state? | 15:13 |
| leftyfb | not in the way you think, that I'm aware of or suggest | 15:14 |
| leftyfb | is your software graphical or can it be run on a server? | 15:14 |
| pycurious | @leftyfb its running on an ubuntu server- and it needs testing. I need ubuntu to be in the original state to test it. | 15:15 |
| leftyfb | so run your tests in containers | 15:15 |
| leftyfb | lxc launch ubuntu:23.10 mynewcontainer | 15:16 |
| leftyfb | you can spin those up in under a minute (depending on internet speed) or create snapshots of them | 15:16 |
| pycurious | @leftyfb The software spins 10 of those - so testing needs more than a container. Docker inside docker is painful, especially when it has to deal with multiple GPUs. Can we go back to my original problem? | 15:17 |
| leftyfb | lxc/lxd != docker and no, it's not difficult to run docker containers in lxd | 15:18 |
| pycurious | Does lxc pass through multiple GPUs easily? | 15:19 |
| leftyfb | you mentioned doing all this in the cloud but haven't confirmed whether or not you're paying for GPU time for your application testing | 15:19 |
| leftyfb | https://ubuntu.com/tutorials/gpu-data-processing-inside-lxd#1-overview | 15:20 |
| preach | nothing passes through GPUs "easily" to my knowledge | 15:20 |
| leftyfb | specifically https://ubuntu.com/tutorials/gpu-data-processing-inside-lxd#6-add-your-gpu-to-the-container | 15:20 |
| leftyfb | it's 1 command to pass through GPU to a container | 15:20 |
| preach | I've always had to change kernel parameters and initramfs config to even attempt it | 15:21 |
| leftyfb | not needed here | 15:21 |
| preach | why not | 15:21 |
| pycurious | Can an lxc container run 10 docker containers? some of which need gpu, and some of which don't? | 15:21 |
| preach | lol what kind of questions are these | 15:21 |
| leftyfb | 10 docker containers, sure. GPU within docker, if you can accomplish that on bare metal, then yes | 15:21 |
| pycurious | @preach iommu/vi-flags/… - there is a bunch of things you've to setup before you can do gpu's pass through. That 1 command assumes you've done all that - which is not that bad! The problem is the containers don't play nice with containers inside them. | 15:22 |
| leftyfb | sure they do | 15:22 |
| leftyfb | lxc config set mynewcontainer security.nesting=true | 15:23 |
| preach | pycurious: I figured they were trivializing it somehow | 15:23 |
| leftyfb | dobne | 15:23 |
| leftyfb | done* | 15:23 |
| leftyfb | I run docker in multiple containers | 15:23 |
| pycurious | @leftyfb Thanks! I did not know that lxc containers can run docker containers inside them easily - that might work - will have to investigate that route. | 15:23 |
| leftyfb | https://ubuntu.com/tutorials/how-to-run-docker-inside-lxd-containers#1-overview | 15:24 |
| leftyfb | I usually set security.nesting=true and security.privileged=true | 15:25 |
| pycurious | @leftyfb how do i enable 3 gpus inside a lxc container? | 15:25 |
| leftyfb | pycurious: I don't know, but fiddling around with https://documentation.ubuntu.com/lxd/en/latest/reference/devices_gpu/ you might be able to sort it out | 15:27 |
| pycurious | @leftyfb can you help me setup a gpu baremetal (ubuntu host) that can do lxc container with docker containers inside it? I do need to kill that container and repawn it everyday :) | 15:27 |
| leftyfb | do you? You asked about "rolling back". Can you spin it up, create a snapshot and restore the snapshot every day? | 15:28 |
| leftyfb | pycurious: I've literally given you the exact 4 commands you need to spin up a container, configure it to allow docker and add a GPU. Since I've never added multiple GPU's to a container, I don't know how to do that. I have given you the documentation for it though. You can always ask for help in #lxd as well. | 15:29 |
| pycurious | @leftyfb I was thinking of lxc can do this - then my problem is solved ? I can just create and destroy that container - unless i am missing anything? | 15:30 |
| leftyfb | you can, but it seems overkill when you have snapshots built in already | 15:30 |
| leftyfb | up to you if you want to do more work | 15:30 |
| pycurious | @leftyfb I think docker does not see the gpus that lxc passes - there are issues there for nvidia - that need to be sorted out | 15:32 |
| leftyfb | I see no reason why docker running in an lxd container server with gpu passed through to the container wouldn't see the gpu if docker running on bare metal can see the gpu | 15:34 |
| ogra_ | you dont "enable" anything in lxd containers ... lxd is a hypervisor, it shares the HW with the host | 15:34 |
| ogra_ | if your host has 3 GPUs the lxd containers should have access to them | 15:35 |
| leftyfb | ogra_: what's the link I posted above for then? | 15:36 |
| leftyfb | (I've never cared about the gpu with containers) | 15:36 |
| ogra_ | i know people using it with multiple GPUS and to my knowledge it just worked for them ... (but i have no first hand knowledge, that much is true) | 15:37 |
| leftyfb | yeah, sure enough | 15:38 |
| leftyfb | root@mantic:~# lspci|grep -i vga | 15:38 |
| leftyfb | 00:02.0 VGA compatible controller: Intel Corporation Raptor Lake-P [Iris Xe Graphics] (rev 04) | 15:38 |
| leftyfb | that's a container I just spun up on my laptop | 15:38 |
| ogra_ | yeah | 15:39 |
| pycurious | I just downloaded my first lxc ubuntu image - still havent spun the container - soon :) | 15:40 |
| leftyfb | uh | 15:40 |
| leftyfb | pycurious: https://canonical.com/lxd/install | 15:41 |
| leftyfb | I would follow that | 15:41 |
| ogra_ | sudo lxd init --auto; lxc launch ubuntu:22.04 jammy; lxc shell jammy ... | 15:41 |
| ogra_ | or that 🙂 | 15:41 |
| leftyfb | right | 15:41 |
| leftyfb | I don't follow this "I jsut downloaded my first lxc ubuntu image" | 15:42 |
| leftyfb | they just use cloud images | 15:42 |
| ogra_ | (oh, the last two might need a re-login to pick up the new group membership) | 15:42 |
| leftyfb | nah | 15:42 |
| leftyfb | worked fine for me | 15:42 |
| ogra_ | ah, good then | 15:42 |
| leftyfb | I rebuilt this laptop about a month ago and this is the first time I've run the lxc command on it | 15:43 |
| ogra_ | (i mainly use lxd on UbuntuCore nowadays ... there a little tinkering is needed for the groups stuff) | 15:43 |
| leftyfb | yeah, I have some custom networking I do with them sometimes | 15:43 |
| pycurious | @leftyfb https://dpaste.org/FvAsb - can you tell me how i get nvidia-smi inside this lxc container? | 15:46 |
| pycurious | nvidia-smi works on the host OS | 15:47 |
| leftyfb | pycurious: you undo everything you did up till this point and follow the suggestions and instructions I gave you instead | 15:47 |
| leftyfb | pycurious: I suggested to use lxd, not lxc. It's a bit confusing because the "lxc" command is named as such | 15:48 |
| leftyfb | pycurious: if you want help with lxc, you'll need to /join #lxc. I haven't used that in many years | 15:48 |
| pycurious | wow! Sorry for bombing this with one character! | 15:49 |
| leftyfb | pycurious: otherwise, you can use lxd and follow the documentation I've provided you already which takes you step by step on how to accomplish most, if not everything you're asking for | 15:49 |
| pycurious | Trying to remove what i did | 15:49 |
| leftyfb | the documentation I provided is for lxd, not lxc | 15:49 |
| === john is now known as Guest2794 | ||
| pycurious | @leftyfb This time I'm following this one https://canonical.com/lxd/install - and apt update doesnt go through - networking issue - how do i resolve that? | 16:00 |
| pycurious | the last command doesnt work on that page | 16:01 |
| pycurious | lxc exec ubuntu-container -- apt-get update | 16:01 |
| leftyfb | pycurious: login to the container "lxc shell ubuntu-container" | 16:02 |
| pycurious | lxc-ls --fancy ==> Nothing running. lxc shell ubutu-container ==> Instance not found! | 16:04 |
| pycurious | lxc exec ubuntu-container -- apt-get update ==> it did run this and gave me a bunch of errors | 16:04 |
| leftyfb | pycurious: which instructions tell you to run lxc-ls ? | 16:04 |
| pycurious | When it gave me instance not found - I thought I would run lxc-ls to see if anything was running. I guess that does not list ubuntu-container? There is another command for that? | 16:05 |
| leftyfb | pycurious: which instructions tell you to run lxc-ls ? | 16:06 |
| pycurious | I ran it on my own - did that kill what i was running? | 16:06 |
| leftyfb | please close and disregard any instructions that mention "lxc-ls" as they are for lxc and not lxd | 16:07 |
| leftyfb | pycurious: lxc list | 16:07 |
| pycurious | that one shows ubuntu-container running | 16:07 |
| leftyfb | lxc shell ubuntu-container | 16:07 |
| pycurious | This time when i ran lxc shell ubuntu-container -> it worked! | 16:07 |
| leftyfb | because you spelled it correctly this time | 16:08 |
| pycurious | apt update inside does not work - connection time out issues | 16:08 |
| leftyfb | does the container have an ip address? | 16:08 |
| pycurious | @leftyfb Thanks! | 16:08 |
| pycurious | I selected "auto" in lxd init - for both ip4/6 - not sure about that | 16:08 |
| preach | why use lxc over docker? | 16:08 |
| leftyfb | preach: lets not | 16:09 |
| preach | ? | 16:09 |
| leftyfb | pycurious: does your container have an ip address? | 16:09 |
| pycurious | @preach its "lxd" | 16:09 |
| pycurious | @leftyfb lxc list does show an ip - both 4 and 6 | 16:10 |
| preach | I must be missing something | 16:10 |
| preach | it was just a question | 16:10 |
| leftyfb | pycurious: do you know how to check the ip address in linux from the command line? | 16:10 |
| pycurious | # hostname -I —> 10.157.55.81 fd42:1c12:43ac:b675:216:3eff:fe5f:a46f | 16:12 |
| leftyfb | ip a # would be a better way, but sure | 16:12 |
| leftyfb | how about the ip of the lxbr0 interface on your host? | 16:13 |
| pycurious | @leftyfb LXD containers can be spun using vagrant using a plugin - have you ever used that? Can I try that? | 16:13 |
| pycurious | inet 10.157.55.1 | 16:13 |
| leftyfb | I haven't used vagrant. I don't see how that would help anything here | 16:13 |
| leftyfb | pycurious: can you ping .1 from the container and vice versa? | 16:13 |
| leftyfb | pycurious: I prefer to get things working and more importantly, understood before I introduce additional tools which layer on top which are meant to "simplify" the original goal which was already pretty simple to begin with | 16:18 |
| pycurious | ping 10.157.55.1 - works from the container - no issues there | 16:18 |
| leftyfb | pycurious: can you ping .1 from the container and vice versa? | 16:18 |
| pycurious | @leftyfb from both host and guest - ping 10.157.55.1 - works | 16:19 |
| leftyfb | vice versa means can you ping the guest ip (which isn't .1) from the host? | 16:20 |
| pycurious | both pings work : From Host: ping 10.157.55.81 - works | 16:21 |
| leftyfb | pycurious: on the guest: resolvectl status|grep "DNS Servers" | 16:22 |
| pycurious | DNS Servers: 10.157.55.1 | 16:24 |
| leftyfb | pycurious: and running "apt update" on the guest is timing out? | 16:25 |
| ogra_ | do you have docker installed on that machine alongside ? IIRC docker does mangle a lot of firewall rules unasked and unannounced that could break stuff | 16:26 |
| pycurious | yes | 16:26 |
| pycurious | to both questions - docker is installed and running. and apt update doesnt work | 16:26 |
| ogra_ | https://github.com/docker/for-linux/issues/103 | 16:27 |
| -ubottu:#ubuntu- Issue 103 in docker/for-linux "Docker blocking network of existing LXC containers" [Open] | 16:27 | |
| ogra_ | see this | 16:27 |
| leftyfb | pycurious: on the host run this: sudo iptables -P FORWARD ACCEPT | 16:27 |
| leftyfb | then try running apt update on the guest | 16:27 |
| ogra_ | right | 16:28 |
| ogra_ | (docker is such a mess ... i never understood how it got so popular) | 16:28 |
| pycurious | ogra_: are lxd containers better than docker? | 16:29 |
| ogra_ | (... but so got VHS videotapes ... so i should not be surprised 🙂 ) | 16:29 |
| leftyfb | pycurious: they each have their uses for different purposes | 16:29 |
| ogra_ | pycurious, well, depends, i mean docker is probably a tad more lightweight than lxd since it does not even run an init system inside the containers, they have a different focus | 16:30 |
| pycurious | @leftyfb - isnt that a security issue -> sudo iptables -P FORWARD ACCEPT - maybe we should be more targetted towards the host lx interface? | 16:31 |
| leftyfb | no | 16:31 |
| arkanoid | hello! how can I configure which DNS my ubuntu should use? I have a single interface eth0, which is configured by dhcp to use 3 dns servers, first is local, second is 8.8.8.8, third is 1.1.1.1. When I ping local name, ubuntu resolves it using 8.8.8.8 instead of the first DNS server. The server works if I do "dig localname @localserver" | 16:31 |
| leftyfb | arkanoid: if you need local resolution, don't assign public nameservers via DHCP | 16:32 |
| arkanoid | leftyfb: this is not how dns works | 16:32 |
| leftyfb | arkanoid: configure DHCP to only assign your local DNS server which should then forward external requests as needed | 16:32 |
| arkanoid | there's a primary, a secondary, and so on | 16:32 |
| leftyfb | arkanoid: this is 100% how DNS works | 16:32 |
| leftyfb | arkanoid: false | 16:33 |
| leftyfb | arkanoid: primary and secondary is for failover of multiple DNS servers which all resolve that same way in case one or more of them is completely down. | 16:33 |
| arkanoid | leftyfb: then why ubuntu picks secondary if primary is up and running? | 16:34 |
| leftyfb | DNS queries do not query multiple nameservers until it finds a record. That is not how DNS works | 16:34 |
| leftyfb | arkanoid: that's not how DNS works | 16:34 |
| leftyfb | arkanoid: configure DHCP to only assign your local DNS server which should then forward external requests as needed | 16:34 |
| arkanoid | this does not answer question why ubuntu queries servers not in the order specified | 16:35 |
| pragmaticenigma | arkanoid: it's not an ordering, that's what being conveyed | 16:35 |
| leftyfb | multiple nameservers are not an order, they are queried randomly and only 1 is ever hit | 16:35 |
| arkanoid | really? I'm pretty sure I've read the opposite on a manual. I might be wrong, let me search for it. | 16:36 |
| arkanoid | btw thanks, if that is not an ordered list, this explains everything | 16:36 |
| leftyfb | it's a big misconception that there's idea of "primary and secondary" nameservers configured on clients. It's wrong | 16:36 |
| leftyfb | there's another misconception that a DNS query will hit multiple nameservers until it gets a hit. That's also wrong | 16:37 |
| leftyfb | a query will ask 1 nameserver, if it responds with anything, that's the answer you get, transaction is now complete | 16:38 |
| leftyfb | sorry, let me rephrase | 16:38 |
| leftyfb | there's another misconception that a DNS query will hit multiple nameservers until it gets a valid resolution. That's also wrong | 16:39 |
| arkanoid | then what is the point of configuring multiple DNS server into a DNS client? | 16:39 |
| leftyfb | in case 1 is down | 16:39 |
| plastikman | the only way a resolver "fails over" is if the first in the list fails to respond | 16:39 |
| leftyfb | it will hit one of the 2 randomly, if the one it hits is completely inaccessible, then it will hit the next one | 16:39 |
| plastikman | The way I solved this in the past was to put a reverse proxy in front of my DNS servers so there was multiple behind a VIP | 16:40 |
| leftyfb | if one responds and says "I have no idea what that hostname is", then that is the end of the transaction | 16:40 |
| arkanoid | got it, so there's the implicit assumption that the two dns server respond with same anwers | 16:40 |
| plastikman | and just so yuo know, its always DNS | 16:40 |
| arkanoid | thanks for the clarification | 16:40 |
| leftyfb | arkanoid: exactly | 16:40 |
| leftyfb | if they respond with different answers, you get intermittent failures and is usually because you configured DNS on the client incorrectly | 16:41 |
| leftyfb | plastikman: it's almost never DNS. It's usually someone's misconception or misconfiguration of DNS :) | 16:43 |
| plastikman | lies! | 16:44 |
| plastikman | Im kidding of course, but many times people blame DNS when its a resolver quirk or assumptions of resolvers. anyway... | 18:00 |
| whoami | Hi - not sure where best to ask about ufw, so I'm doing it here. Please let me know if there's a better spot. I'm writing docs for a product, and trying to give instructions for ufw. With iptables, we install the following two rules to allow forwarded traffic, and return traffic: iptables -I FORWARD 1 -i enp5s0 -o defined1 -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT / iptables -I FORWARD 2 | 19:25 |
| whoami | -i defined1 -o enp5s0 -s 192.168.100.0/24 -d 192.168.86.0/24 -j ACCEPT | 19:25 |
| whoami | With ufw, this seems to work: ufw route allow in on defined1 out on enp5s0 from 192.168.100.0/24 to 192.168.86.0/24 | 19:25 |
| whoami | However, I'm somewhat surprised when I run iptables -vnL, I don't see a conntrack rule for return traffic. Is this expected? | 19:26 |
| whoami | Hm, I see `ufw-before-forward` has a generic rule from `/etc/ufw/before.rules` which seems to cover this case: -A ufw-before-forward -m conntrack --ctstate RELATED,ESTABLISHED -j ACCEPT | 19:28 |
| pragmaticenigma | whoami: if you're still looking for some help, #networking may be a good community to reach out to | 20:27 |
| whoami | thanks :) my question is mostly about ufw behavior (whether ufw route allow creates a rule for reverse traffic / whether that is the intention of the default rule), but perhaps there are some people familiar with ufw over there. | 20:28 |
| whoami | i'll give that a go tomorrow morning perhaps. appreciate the suggestion | 20:29 |
| pragmaticenigma | whoami: ufw is uncomplicated firewall ... all I could say is that by design it's supposed to be approachable more for a consumer level, rather than enterprise. | 20:29 |
| pragmaticenigma | I personally don't know what your iptables rule does, to me it's a magic encantation that I'd hope is accurate | 20:30 |
| pragmaticenigma | I use ufw with my main focus being opening and closing ports | 20:30 |
| whoami | yeah, i understand. it does have support for forwarding though, so i don't think i'm doing anything it's incapable of. | 20:31 |
| whoami | as it happens, there are customers who would like to use it. | 20:31 |
| pragmaticenigma | there always is :) | 20:31 |
| whoami | and fwiw, it is working in practice, i just want to make sure my understanding is correct :) | 20:31 |
| plastikman | I feel like both UFW and other tools like it start to get murky after you stop being simple | 20:32 |
| whoami | as all networking seems to... heh. | 20:33 |
| pragmaticenigma | whoami: totally encourage you to verify your theory. it makes me happy to know there's someone such as yourself that is going that extra mile :) | 20:34 |
| === kostkon_ is now known as kostkon | ||
| arraybolt3 | If UFW is Uncomplicated FireWall, I don't even want to see what CFW (Complicated FireWall) looks like... | 20:48 |
| plastikman | that is called iptables | 20:49 |
| arraybolt3 | lol, right... yeah that really is a mess | 20:49 |
| === JanC is now known as Guest2005 | ||
| === JanC_ is now known as JanC | ||
| JanC | UFW is for doing uncomplicated things, not for making complicated things uncomplicated or something like that :) | 21:28 |
| JanC | if you want to do things that UFW wasn't designed to do, you should look for an alternative | 21:29 |
| SnoopJ | The copy of `perf` provided by `linux-tools-6.5.0-18-generic` is apparently built without `libtraceevent`. I'm relatively comfortable with building my own, but wondering what I need to do to get the appropriate kernel sources to do that. The sources provided `linux-source-6.5.0` do not seem to include `tools/perf` | 21:44 |
| SnoopJ | (alternatively, if there's a way to ask Ubuntu to give me a more useful copy of `perf`, I'd welcome that advice too) | 21:44 |
| elhoir | any 6.5.x kernel from upstream should be fine (the higher version the better) | 21:45 |
| elhoir | you just have to get config from the installed one and you`re done | 21:46 |
| SnoopJ | does that mean that Ubuntu itself cannot provide me the non-mangled source that goes with my installed kernel? :( | 22:16 |
| JanC | SnoopJ: with "apt source" you can get the source package of any binary package | 22:17 |
| JanC | (no need for sudo!) | 22:17 |
| oerheks | https://cgit.freedesktop.org/xorg/driver/xf86-video-radeonhd/plain/README | 22:18 |
| oerheks | get it from git? | 22:18 |
| oerheks | git://anongit.freedesktop.org/git/xorg/driver/xf86-video-radeonhd | 22:18 |
| SnoopJ | JanC, is that supposed to work for `linux-tools-6.5.0-18-generic`? It doesn't work for me, it picks `linux-hwe-6.5` instead and then fails. | 22:19 |
| JanC | because that's the source package | 22:19 |
| oerheks | oops wrong window | 22:19 |
| JanC | one source package often builds several binary packages | 22:19 |
| SnoopJ | I'm afraid I don't speak Debianese well enough to unpack that into a next step | 22:20 |
| SnoopJ | Maybe Debian/Ubuntu are sufficiently opinionated about breaking this workflow that going upstream is just the Best Way™ | 22:25 |
| oerheks | https://bugs.launchpad.net/ubuntu/+source/libtraceevent/+bug/2051916 | 22:27 |
| -ubottu:#ubuntu- Launchpad bug 2051916 in libtraceevent (Ubuntu) "[MIR] promote libtraceevent as a trace-cmd dependency" [Undecided, Incomplete] | 22:27 | |
| oerheks | add yourself? | 22:27 |
| SnoopJ | I would like to solve this problem on a timeline that does not depend on sitting on my hands until a maintainer fixes it | 22:30 |
| oerheks | old one https://bugs.launchpad.net/ubuntu/+source/linux/+bug/2019247 | 22:30 |
| -ubottu:#ubuntu- Launchpad bug 2019247 in linux (Ubuntu) "perf should be compiled with libtraceevent" [Undecided, Confirmed] | 22:30 | |
| SnoopJ | right, I saw that one when trying to search up a solution | 22:30 |
| crobi | hello! I'm having some strange issues with pulling from archive.ubuntu.com the /Ubuntu directory keeps recurring over and over about 20+ directories deep. is that normal? I'm noticing it on the eu mirror as well. | 22:38 |
| oerheks | if you keep clicking ubuntu ... http://archive.ubuntu.com/ubuntu/ubuntu/ubuntu/ubuntu/ubuntu/ | 22:39 |
| oerheks | have fun | 22:39 |
| leftyfb | crobi: that's not how you're supposed to "pull" from the repo | 22:39 |
| leftyfb | crobi: don't do that | 22:40 |
| crobi | yeah the problem is I'm trying to install 20.04 and the installer is not able to see the mirror | 22:40 |
| leftyfb | why are you installing 20.04 over 22.04? | 22:40 |
| crobi | sorry poor choice of words, I'm not pulling anything | 22:40 |
| crobi | dependencies for ROS noetic is why 20.04 | 22:41 |
| leftyfb | crobi: please pastebin the error you are getting | 22:42 |
| crobi | I'm on my phone unfortunately this system is on a different network. | 22:43 |
| leftyfb | ok, feel free to ask for help when you are in a position to troubleshoot the machine in question | 22:43 |
| crobi | gimme a sec transcribing the error | 22:44 |
| crobi | "bad archive mirror. additional details in /var/log/syslog" I checked the logs and it's not able to resolve the mirror. the weirdness is that the installer was able to resolve it earlier during the install. | 22:45 |
| leftyfb | please pastebin /etc/apt/sources.list | 22:46 |
| crobi | I don't have net-tools available so I can't check much beyond pinging the mirror | 22:47 |
| crobi | which it cannot resolve. | 22:47 |
| crobi | oh sorry network connection is bad so messages are coming in sporadically | 22:47 |
| crobi | I don't seem to have a sources.list in /etc/apt | 22:48 |
| leftyfb | then you aren't running ubuntu or have modified it in an way that is not recommended and probably causing your issues | 22:49 |
| preach | lol | 22:49 |
| leftyfb | crobi: do you have any files in /etc/apt/sources.list.d/ ? | 22:50 |
| crobi91 | anyway I appreciate the help, my cell signal is cutting out too much I'll try again next week | 22:50 |
| sarnold | heh, is the bad network connection why you couldn't resolve the name? | 22:50 |
| leftyfb | "bad archive mirror" is not a failure to resolve | 22:51 |
| crobi91 | probably but the system in question is on a different network in my lab, I'm on LTE with my cell | 22:51 |
| leftyfb | and a lack of a sources.list shows it's either not ubuntu(not official anyway) or modified in a way that is likely causing the issue | 22:51 |
| leftyfb | crobi91: this should be your /etc/apt/sources.list https://paste.ubuntu.com/p/wQ7g5yV3Yn/ | 22:52 |
| crobi91 | mmk good to know I'll bug the vendor some more and see if they will be able to offer suggestions | 22:52 |
| leftyfb | you can probably leave out the partner repo | 22:52 |
| oerheks | original can be found in /usr/share/doc/apt/examples/sources.list | 22:54 |
| leftyfb | or that | 22:54 |
| leftyfb | mine is much cleaner though :) | 22:54 |
| oerheks | indeed, original has no universe enabled | 22:54 |
| leftyfb | which causes a lot of people grief | 22:54 |
| preach | where is the code that copies that file to /etc/apt | 22:55 |
| leftyfb | the apt package does it | 22:57 |
| preach | where? | 22:57 |
| leftyfb | preach: somewhere in the source code for the apt package | 22:57 |
| leftyfb | you are welcome to go through it: apt source apt | 22:57 |
| preach | I'm looking here: https://github.com/Debian/apt/tree/main/debian | 22:59 |
| preach | but I don't see it anywhere | 22:59 |
| sarnold | that's not the package source, that's some old snapshot of the upstream git tree | 23:00 |
| preach | last commit last week is old? | 23:01 |
| preach | and why wouldn't it be in there either way? | 23:01 |
| leftyfb | preach: why do you need to know? | 23:02 |
| preach | I've always wanted to know and it has come up in conversations several times recently with different people | 23:02 |
| preach | wanting to know how sources.list is generated and how it gets put there | 23:03 |
| leftyfb | preach: I pulled down the apt source and used grep and it looks to me like it might be in apt-2.4.5/dselect/setup | 23:03 |
| preach | but the ask anyone has come up with is that the /usr/share file _looks_ like the stock one, but I haven't seen any proof that it is the exact same | 23:03 |
| sarnold | preach: (a) I was just going by the banner at the top " This repository has been archived by the owner on Dec 12, 2022. It is now read-only. " (b) it's still just some github clone and it isn't the package source :) apt-get source apt | 23:03 |
| preach | I'm not on an ubuntu machine | 23:05 |
| preach | and why would the github source be different than the package source? | 23:05 |
| leftyfb | :/ | 23:05 |
| preach | and even if it was, why would the same code in question not be there in one and only in the other? | 23:05 |
| preach | makes no sense | 23:05 |
| leftyfb | preach: feel free to ask in #debian or the maintainer of that git tree | 23:06 |
| oerheks | what package provides /usr/share/doc/apt/examples/sources.list and /usr/share/keyrings/* | 23:07 |
| oerheks | apt i bet | 23:07 |
| oerheks | i never installed apt though | 23:08 |
| sarnold | preach: ah, you should have said that first :) | 23:09 |
| sarnold | preach: there's a dozen supported versions of apt https://launchpad.net/ubuntu/+source/apt and you can click through to grab the tarballs of upstream sources and ubuntu patches from them all. if you've got dget on your system, dget https://something/path/to/apt.dsc will download the other files for you, and it can even unpack the sources if you give it the right command line arguments | 23:11 |
| szstr | mega good luck to all ayn this channel withaynout 2 minutes withayn/within 20 minutes | 23:13 |
| sarnold | hey szstr, do keep the sillyness to #ubuntu-offtopic please :) | 23:13 |
| szstr | yes no no no yes, sir | 23:14 |
| sarnold | thankya :) | 23:14 |
| szstr | i don't like ubuntu, but i do like that kubuntu 23.10 has: yt-dl3, it needs kwrite, could use abiword, needs C/C-- runners/konguilerz, we need a web sergher gui-auto-aynstaller aynstaller, mantis ayz/is a good nickname, | 23:58 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
