Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Yaoi Gagarin
Feb 20, 2014

Grumpwagon posted:

It's just a normal desktop computer, I just multitask a lot. I always have 20+ chrome tabs open, play games (WINE and native), steam, run a couple of background servers (nothing intense, just a personal subsonic server), and various other things. I usually have 3-5 gnome3 workspaces doing various things.

I think 8 GB should be fine for you. I don't play games on Linux, but when I'm programming I'll have lots of Chrome tabs, text editors/IDEs, terminals, etc. across three workspaces, plus a music player and some other crap, and this all runs just fine on 4 GB. Lots of people seem to be able to play games and have Chrome tabs open at the same time with 8 GB on Windows, so unless WINE is using a lot more RAM I can't see it being a problem.

Adbot
ADBOT LOVES YOU

Yaoi Gagarin
Feb 20, 2014

Apologies if this is more of a COBOL question, but this is maybe a Fedora-specific problem so I thought it would be best to ask it here.

I'm working on a project using GLFW. We've set it up so that the GLFW cmake script is called from our own CMakeLists.txt, so GLFW is being treated as if it's part of our own project. This works fine on the 'ancient' CentOS machines in our school's computer lab, but now I want to work from my laptop, which has had Fedora on it since yesterday. The problem is that when I run cmake, I see this:

code:
CMake Error at /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:136 (message):
  Could NOT find OpenGL (missing: OPENGL_gl_LIBRARY OPENGL_INCLUDE_DIR)
Call Stack (most recent call first):
  /usr/share/cmake/Modules/FindPackageHandleStandardArgs.cmake:343 (_FPHSA_FAILURE_MESSAGE)
  /usr/share/cmake/Modules/FindOpenGL.cmake:175 (FIND_PACKAGE_HANDLE_STANDARD_ARGS)
  glfw/CMakeLists.txt:88 (find_package)
When it tries to go through the GLFW CMakeLists.txt. I've googled this but the only results are from 2008/9, and for ubuntu - the solutions vary, but they're mostly along the lines of "install package 'mesa-devel'". There doesn't seem to be a package like that in the Fedora pkgdb, though.

I know there's a GLFW package, but my partner and I agreed to build all libraries for this project from source so that we can keep it consistent across multiple systems.

I ran the Intel Graphics Drivers Installer, thinking that maybe this was a problem with the drivers provided by Fedora, but that hasn't solved the problem. Maybe the school's sysadmins used blood magic to make everything work okay on CentOS.

tl;dr wtf do i have to do to compile against opengl on fedora

edit: Nevermind, it turns out I needed mesa-libGL-devel. And I also needed a whole bunch of libraries but I found them all. Compile -> install library mentioned in error -> repeat.

Yaoi Gagarin fucked around with this message at 20:51 on May 16, 2015

Yaoi Gagarin
Feb 20, 2014

Is there a "right" way/best practice to get software that's newer than what's in the distribution's repository? For example, if I'm reading this right, the latest stable version of Fedora only has version 2 of Mono, while the upstream is on version 4. I could get version 4 by moving the system to Rawhide, but then I'm on a nightly update schedule and everything on my system is going to be changing all the time and it's just like Arch all over again. :smith:

The way I see it, my options are:
  1. Deal with old software
  2. Use a binary I manually download, if available
  3. Use a third-party repository, if available
  4. Compile from source

I'm leaning towards #3 as the right answer, but there might be something I'm missing, and in any case I'd like a second opinion. #2 would be probably be my fallback, but then I have to get all the dependencies myself and keep track of updates myself.

Also, even though I'm using Fedora as an example, I'd still like to know what the best practices are in general. I'm not tied to it, either, so if the solution is actually "use distribution X" I'll gladly switch.

Yaoi Gagarin
Feb 20, 2014


Did I write something dumb?

Marinmo posted:

This will help you install the latest mono into /opt, which probably is the best of solutions. Nothing more than a few path additions and you'll be good to go, and avoid breaking anything by upgrading between major versions of the package.

Thanks.

Yaoi Gagarin
Feb 20, 2014

Thanks for the help, everyone. In this case, using the external repository provided by Mono looks like the most convenient option, but I'll look into backporting for some other stuff I might want.

Hollow Talk posted:

Quality poster and linux savant nosl tried to mime to you that the only real solution is to use Arch Linux, for everything, forever.

And somewhat less facetious: Marinmo's link to the OBS works, since they build for Fedora. Another option would be the official mono repository from the mono project itself, which should keep you up-to-date without the need to trust a random person building packages: http://www.mono-project.com/docs/getting-started/install/linux/#centos-fedora-and-derivatives

That's technically option #3.5, since it's an external repository, but at least it is a first-person repository.

ExcessBLarg! posted:

Sometimes you can install packages from a newer version of your distro without pulling in all newer packages. Debian and Ubuntu let you do this with apt pinning. However, if the newer pages have dependencies on core libraries, it might badger you into pulling in a bunch of unrelated new packages anyways.

Some distributions also run backport repos, which would be the same idea (pin newer versions of the packages you want from the backport repo), but are specifically built against core libraries in the latest stable release to avoid pulling in unrelated dependencies.


spankmeister posted:

I'd get the source rpm for mono4 from f23 and rebuild it.


Yaoi Gagarin
Feb 20, 2014

If they don't show up as symlinks when you ls -l they're probably hard links

E: even if they're not, you probably shouldn't delete them. I think Autoconf looks for that type of prefix when deciding on a tool chain to use

Yaoi Gagarin fucked around with this message at 18:29 on Mar 14, 2017

Yaoi Gagarin
Feb 20, 2014

VOTE YES ON 69 posted:

btrfs is the systemd of filesystems: cool poo poo implemented by naive idiots

Systemd actually works tho

Yaoi Gagarin
Feb 20, 2014

Thermopyle posted:

You sure you don't want to investigate the VM route more thoroughly?

I only ask because I worked in a Linux VM on a Windows host for years and once configured it was basically indistinguishable from running on bare metal for most tasks.

I am extremely interested in how you did this. If you could write up a little bit about what hypervisor you used and how it was configured I'd appreciate that a lot. At work I use VMware on a Mac and it's phenomenally smooth but all my attempts at doing this on my windows machine at home with vbox, VMware player, and hyper-v have been pretty bad.

E: to clarify, the things I really want in order to make a Linux VM usable for me:
- seamless copy-paste between host and guest
- no graphical hiccups
- I/o performance should be good enough to compile things, run an IDE, etc.

Yaoi Gagarin fucked around with this message at 01:04 on Mar 5, 2018

Yaoi Gagarin
Feb 20, 2014

Thermopyle posted:

I haven't used it in a couple years, but I basically just installed VMware player, installed Ubuntu, installed my IDEs.

Granted I always use a pretty decent pc for my work machine so I could afford to dedicate a couple cores and 8+GB RAM to the VM.

It easily met your listed requirements.

drat, I must have done something really wrong last time then. Or maybe the particular distro I used didn't play well with virtualization in general (might have been Debian or fedora, I forget).

The RAM isn't a problem, I have 16 GB so I can definitely give 8 to the one VM I'll ever run, but I've only got 4c/8t so it would probably be a bad idea to give the vm exclusive use of anything, if that's even possible on windows. Maybe I'll try it again tonight and see what happens.

Yaoi Gagarin
Feb 20, 2014

VostokProgram posted:

drat, I must have done something really wrong last time then. Or maybe the particular distro I used didn't play well with virtualization in general (might have been Debian or fedora, I forget).

The RAM isn't a problem, I have 16 GB so I can definitely give 8 to the one VM I'll ever run, but I've only got 4c/8t so it would probably be a bad idea to give the vm exclusive use of anything, if that's even possible on windows. Maybe I'll try it again tonight and see what happens.

Trip report: this works pretty well with very little fiddling, 10/10 would VM again.

I installed fedora 27 and it worked fine out of the box. All I did was give it 8 GB of memory and 4 virtual CPUs. I'll probably increase that to 8 to match my logical core count, seems like it wouldn't hurt.

Now I just have to decide on a project to use this VM for...

Yaoi Gagarin
Feb 20, 2014

Out of curiosity, can one of you real Linux admin tell me if anyone actually uses kvm/qemu filesystem sharing? It seems incredibly janky

Yaoi Gagarin
Feb 20, 2014

RFC2324 posted:

I have in my home lab, but it seems fine once you get it working. I don't remember exactly what I had to do but there was some trick to getting the guest to see and mount the shared filesystems.

Then I just used an NFS share on an internal only network, and never looked back

The nfs thing is a problem for me because my VM has to use some lovely VPN software to access another company's network and it does something weird that prevents the machine from accessing our internal network (which is why I put this in a VM)

Yaoi Gagarin
Feb 20, 2014

RFC2324 posted:

i mean the internal only network, running on a separate virtual nic.

Tell me more about this, it sounds like it might solve all my problems

Yaoi Gagarin
Feb 20, 2014

SoftNum posted:

Which lovely VPN software is this? Cause there's patches for most of them to overwrite this behavior.

Cisco AnyConnect I believe

Wrt to the internal network stuff, I'll see if I can figure out how to do that with kvm. Thanks!

Yaoi Gagarin
Feb 20, 2014


Well the Cisco thing isn't something controlled by our IT department, it's something our very bullshit customer uses, and we have to use it to access their internal website. Our own VPN is very nice and well-behaved.

Yaoi Gagarin
Feb 20, 2014

I wish the fedora sysadmin guide were as exhaustive as the freebsd handbook. That thing covers pretty much anything you'd ever want to do with the system

Yaoi Gagarin
Feb 20, 2014

RFC2324 posted:

no i mean any internal only network with no access to the outside network at all. the only issue i can see is that i don't know how tat would handle being clustered, if that's happening.

essential you can set up a network that only communicates with the vms or the host, no external communication at all. it works functionally just like mounting a drive locally but you issue the commands to mount it as nfs

I've spent some time trying to look up how to do this but couldn't really get anywhere, do you mind providing more details?

Yaoi Gagarin
Feb 20, 2014

I have been using kde on my work desktop for a couple months now and it absolutely grinds my gears that all konsole windows apparently run in the same process. If you kill one they all die. I might go back to xfce at this rate.

Yaoi Gagarin
Feb 20, 2014

ToxicFrog posted:

Settings -> Configure Konsole -> General -> [ ] Run all Konsole windows as a single process

That defaults to off for me. :shrug:

I don't have that option

Yaoi Gagarin
Feb 20, 2014

CaptainSarcastic posted:

Weird, I just checked and I have the same as ToxicFrog describes.

Maybe it didn't exist in whatever old rear end version Ubuntu 16.04 has? Or maybe the maintainers of ubuntu's kde packages decided that feature should be on by default with no option? Who knows

If it were me managing my own machine I'd just use Fedora but our IT department uses Ubuntu for everything and I didn't feel like making life hard for them when I got my workstation replaced. Oh well :shrug:

Yaoi Gagarin
Feb 20, 2014

I liked KDE but it behaves erratically on my work computer (kubuntu 16.04 lts I think). Sometimes pressing alt+f1 to open the applications menu doesn't work and you have to hit it one or two more times. Most recently it's started treating some keyboard inputs as shortcuts, kind of like windows with sticky keys on. It completely refuses to let me type the 's' key, and 'q' causes the activities pane to open up. :shrug:

I don't care too much because I won't be using that computer for much longer but the general jankiness of the experience has kind of soured me on KDE.


E: I use fedora with gnome on my Linux VM at home and it's very solid. All I had to do was use gnome tweak tool to change a few things. Moderately interested in trying Solus but I worry about package availability

Yaoi Gagarin fucked around with this message at 18:40 on Dec 30, 2018

Yaoi Gagarin
Feb 20, 2014

feedmegin posted:

Umm. Are you sure there isn't something physically wrong with your keyboard? 'KDE won't let me type the letter S' would be the sort of bug a lot of people would notice, you know.

100% sure, it's the first thing I thought of. Tried the keyboard on Windows and it was fine, tried another keyboard on the KDE machine, same results. Shift+s works though, so you can type lowercase if you caps lock and then shift+s. Absolutely no idea how this even happens. Pretty sure the keys also work fine under tty1, it's just the X/KDE environment that's borked somehow

Yaoi Gagarin
Feb 20, 2014

taqueso posted:

Do you accidentally have a keybind for S with no modifiers?

It's possible, since Q with no modifiers opens the Activities panel now, I'll have to check it on Monday.

Yaoi Gagarin
Feb 20, 2014

What if I use emacs vim and sublime and sometimes even that lovely notepad clone that ships with every linux de depending on circumstance

Yaoi Gagarin
Feb 20, 2014

SoftNum posted:

Do you use rsub w/ sublime? Cause you should use rsub with sublime.


also if you don't use emacs but use evil mode to make it more like VI, idk who you are.

What is rsub?

I tried evil mode briefly when I was trying out spacemacs but I decided to stick with just normal emacs since the learning curve on vi-style controls is just a bit too steep for me. Although if there were a shorter way to use M-x replace-string that would be handy....

Yaoi Gagarin
Feb 20, 2014

Computer viking posted:

Cron does not; it's started outside and perhaps even before X is up.

Also, Cron has a deliberately small set of environment variables. You can check what they are by inserting a Cron job that does "env > somefile" and check the contents of that file after the job runs.

Yaoi Gagarin
Feb 20, 2014

Absolutely everyone who ever goes within 50 feet of a unix-like machine should learn hjkl, i, escape, :wq, and :q!

This is so when said machine invariably gets completely hosed you can still edit config files on it.

But besides that use whatever text editor you want


E: actually / is also extremely useful and you can use it in less and when reading manpages too

Yaoi Gagarin
Feb 20, 2014

hifi posted:

If I accidently start vim I have to reboot my computer, and if I have to edit a text file without a gui as some sort of a Real Computer User Test I use emacs from another computer and edit it over ssh with tramp.

whatcha gonna do when you dont have a network connection and your only shell is over a serial port

also gently caress Android, that poo poo doesn't even have vi so you gotta edit files with loving sed or cat

Yaoi Gagarin
Feb 20, 2014

Buttcoin purse posted:

Why not? I'm not looking at images and I can just use the terminal's mouse support so who needs X? I spend a lot of my time using Emacs under screen (the TTY multiplexer). If it's okay to use vi in non-X mode, why not Emacs?


Or learn Emacs, and then use the info command instead of man, or just read man/info documentation and all text files from within Emacs :smuggo: (do not try for really large text files, and to be honest I still run man frequently too)

I actually do use for editing text but typing "man whatever" is way easier for me than opening that file in emacs. less I use a lot because you can do cool things like pipe the output of a commnd to it, or read a file that's continuously being appended to

Does anyone besides GNU actually use info?

mystes posted:

There are various ways to get vi on Android.

We probably could have cross compiled it if we tried hard enough but it did not seem easy. Actually now that I remember better, sed was only used in extreme cases [1], more often we'd use adb to get the file off and push it back on

[1] when we couldn't use adb because we needed to test the actual normal behavior of the USB stack on this device. :suicide: God that project still makes me salty and I don't even work there anymore

Yaoi Gagarin
Feb 20, 2014

Bit of a strange question but, if I built a computer without a GPU what could I do to actually install and configure Linux on it? Obviously once the network is up you can ssh in and do whatever but how do you get it to that point? On an embedded board with no screen I would connect to a serial port to use the shell, can I do something similar on a PC?

Context being I want to set up a server at home and it seems silly to spend money on a GPU just to run the installer. I know some motherboards have IPMI that solves this problem completely but I don't know if I'll actually end up getting one of those.

Yaoi Gagarin
Feb 20, 2014

xzzy posted:

It's actually pretty hard these days to find any computer that doesn't have some kind of video output.. even all those SBC hobby systems are likely to have an hdmi out. An embedded graphics chip is pretty much standard equipment.

But if you do manage to find one, yes, you can install using a serial port as your primary interface, we still do it regularly at my place of work and life is good. You might have issues tweaking anything in the bios but the actual running an OS installer will work fine.

I was looking at using AMD Ryzen CPUs, those don't have an integrated GPU and the motherboards don't provide one either.

Yaoi Gagarin
Feb 20, 2014

apropos man posted:

Last week I bought my first Ryzen rig (well, mobo and CPU). I wanted more cores, as I've recently been getting more interested in running a variety of OSes in ESXi and remotely accessing the desktop. I already have a home-made Linux server which is used for keeping important/personal stuff, which is purely a command line environment.

Anyway, I decided to get a B450 motherboard so that I can upgrade to Zen2 when it comes out and I figured I'd start off with a Ryzen 1700. I was gonna get the motherboard brand-new and go second-hand for the 1700 but when I saw that I could get an 8core/16thread CPU for £150 brand-new with Wraith Spire cooler I just bought the lot brand new.

I already had 32G of cheap 2400MHz Corsair DDR4 which I may replace further down the line with something faster but it's fine for now.

First impressions are very positive. I still need to tinker with the clocks and get the 1700 running at close to 4Ghz on all cores this weekend but I'm really pleased that I have 16 threads to provision to VM's. The Wraith Spire cooler is really nice: barely audible with the case open.

I wish I'd got a Ryzen box earlier and the 1700 is such a nice price at the moment for what you get.

What sort of thing are you thinking of building?

Going to build a combination nas/application server, and also use it for occasional video encoding. It'll run Plex, backup my Windows PC, and maybe host that whole suite of programs for getting stuff off Usenet automatically. And I'm sure over time I will find more uses for it. Running a local dns server and doing the steam caching thing maybe?

Basic plan is:
- 8 core Ryzen (probably 1700 like you)
- some sweet spot of ram capacity vs price, if I can get 64 GB that would be perfect
- probably a used SAS HBA of some kind?
- a whole lot of hard drives in either zfs or md raid
- little nvme boot ssd

I haven't decided what distro to use, but it'll probably be either fedora or Debian. I haven't had a Linux computer at home since college so it should be a fun project.

Yaoi Gagarin
Feb 20, 2014

apropos man posted:

1700 is such a good price at the moment the only downside I can think of is that the price may slip when Zen2 comes out. You may want to try out Proxmox (https://en.wikipedia.org/wiki/Proxmox_Virtual_Environment). It's based on Debian and is similar to a free version of ESXi. I tried it the other week and, while I find ESXi a bit easer to use, Proxmox has ZFS built into it. There's a decent amount of documentation and forum advice for running it, too. I wish ESXi had native ZFS.

I've got an LSI HBA in my other machine (this one: https://www.servethehome.com/lsi-9207-9217-hbas-sas2308-6gbs-sas-sata-pcie-30/). It's running a ZFS mirror on CentOS. They run quite hot so I bolted a small fan to the heatsink on mine. I only have two WD reds on it at the moment but when 2TB SSD's are cheap enough I plan to gradually add more ZFS mirrors.

I'd say it's a pretty sound plan that you have. I went with a B450 motherboard over one of the X series ones because it's a virtualisation host and I'm not looking to get bleeding edge speeds out of it. I've got 3.7Ghz on all cores and I'm about to tweak it to 4GHz. If it's stable at 4 that's pretty awesome considering the core count.

One thing about the B450 that I bought is that it's got a Realtek network interface and initially didn't play nicely with ESXi but it was absolutely fine on Linux (I tried Fedora first to make sure everything was working, then tinkered with Proxmox for a while and that was fine, too.

I didn't try setting up a ZFS array in Proxmox but I did have a quick look though the ZFS menus and ZFS config looks really nicely set up.

Ryzen 1700 + Proxmox would be a really useful multi-purpose setup. And the possibility of upgrading the CPU to 7nm in a few months could ne killer. If AMD get it right then the performance/watt ratio at 7nm could be amazing. There's always the possibility of Zen2 not quite delivering but the first-gen 1700 is still a fine CPU.

What's a good setup for using proxmox as both VM host and NAS?

I'm imagining something like:

Proxmox has separate datasets for storing media, the Plex metadata/config databases, and my other computers' backups, plus 1+ datasets for random crap I want to store on it

Plex + any related programs can run in either a VM of lxc container with the relevant datasets mounted in using bind mounts or something

The backup program lives in it's own VM or container

Another VM or container has access to the media dataset and the random crap datasets and exposes them to the network over samba

I could even combine the Plex and samba VMs/containers and use open media vault or something

Yaoi Gagarin
Feb 20, 2014

Kreeblah posted:

Yeah, it's a terrible loving idea for a feature for so many reasons, especially since there doesn't seem to be a way to require authentication for it, but it does what I need, so hopefully I won't come home one day to a ream of paper in the output tray full of solid black pages.

I think you'll be fine as long as you don't expose that through your routers' firewall.

Yaoi Gagarin
Feb 20, 2014

Someone should write an init system where pid 1 is make. The entire process of bringing up the os is just rules in makefiles all the way down

Yaoi Gagarin
Feb 20, 2014

The Milkman posted:

Docker is a PITA to set up (especially if you're learning as you go) but once it's all working it's much simpler to maintain. FreeNAS plugins work fine but they're often pretty far out of date. Then, things get iffy if you update through other means. You can roll your own jails but you're still at the mercy of the Ports upstream. If you're looking to learn linux some Docker/container knowledge will be far more useful, as neat as jails are.

I've been running freenas for a few years now and I'm on my 4th migration (FreeNAS 9 plugins > FreeNAS Corral docker [RIP] > FreeNAS 11 plugins > FreeNAS RancherOS VM). Rancher is pretty slick, way overkill for the home, but it's fun to poke at. Doing all this docker stuff has got me thinking about just moving the whole system to something like CoreOS + ZoL sometime in the future. Not sure about how mature ZoL is right now though.

I run two stacks: Sonarr, qBittorrent, Emby; and NextCloud, Postgres, Collabora. And then a Caddy container as a reverse proxy + built-in automatic LetsEncrypt. I used to have Pi-hole and Unifi on there too but they got promoted to real hardware when I got second hand cloudkey and RPi 3B's to play with.

I've looked into docker and ZoL a bit and from what I can tell, you can't use the overlay2 storage driver on top of a zfs dataset, and the zfs storage driver is only supported on Ubuntu. It's probably fine for home use though

Yaoi Gagarin
Feb 20, 2014

I want to write a program/script to look at some process trees, what interfaces can I use to get information like "ps -t sometty --forest"? I assume I have to muck around in /proc to figure out the parent/child relationships, but how do I constrain to just processes under a specific tty?

Yaoi Gagarin
Feb 20, 2014

RFC2324 posted:

y'all use alot of letters to write 'vi'

:bsdsnype:

that's not how you spell ed, the standard text editor

Yaoi Gagarin
Feb 20, 2014

SoftNum posted:

code:
[utena@ohtohori Downloads]$ ed
bash: ed: command not found
[utena@ohtohori Downloads]$ vi
[utena@ohtohori Downloads]$
:smugdog:

what the gently caress

Adbot
ADBOT LOVES YOU

Yaoi Gagarin
Feb 20, 2014

Why use any of that poo poo when you can just C-x M-c
M-butterfly

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply