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 $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«779 »
  • Post
  • Reply
taqueso
Mar 8, 2004









Fun Shoe

netbsd will do it

Adbot
ADBOT LOVES YOU

Mr Shiny Pants
Nov 12, 2012


mystes posted:

Is anyone using GPU passthrough with Looking Glass? I'm thinking of trying consolidating from a linux computer and a windows computer with multiple monitors and barrier for keyboard/mouse sharing to one 4k monitor attached to a linux computer with windows running in a vm.

What do you want to know?

Nostalgia4Butts
Jun 1, 2006

WHERE MY HOSE DRINKERS AT

CaptainSarcastic posted:

It's been years since I played with a PPC-based system. About 10 years ago I had good luck with the PPC spin of openSUSE (I think that might have actually been the first time I actually ran openSUSE at all).

At this point I'm not sure what distros have PPC versions available, be they official or unofficial.

If memory serves the biggest thing it was missing was Flash, but that's so deprecated it doesn't seem like a big issue compared to how much of a problem it was at the time.

You could always scrounge up an old image of Yellow Dog. I never had a good experience with Yellow Dog even when it was current.

Fake edit: I did a quick search and it looks Ubuntu doesn't have a recent PPC build, but it looks like openSUSE does.

https://en.opensuse.org/Portal:PowerPC

I'm not sure, but it might be possible to find a BSD to install on PPC, too, but I didn't look extensively.

taqueso posted:

netbsd will do it

thanks so much for the advice, I'll poke around a bit more. Another friend recommended Puppy Linux- any thoughts?

CaptainSarcastic
Jul 6, 2013

HAIL SATAN


Nostalgia4Butts posted:

thanks so much for the advice, I'll poke around a bit more. Another friend recommended Puppy Linux- any thoughts?

I haven't used Puppy in years, so don't feel like I can really offer much current input, especially about PPC.

I will say that it was Puppy that I used to get Linux running on a Pentium II Toshiba laptop someone had given me, and I was able to get it up and running and get online wirelessly with a PCMCIA wireless card. It was kind of painful on an 800x600 display, and was slow as hell, but I accomplished the task I'd set myself.

D. Ebdrup
Mar 13, 2009



taqueso posted:

Pulseaudio is really nice to work with in my experience, and complex setups are kinda it's thing

And audio has been fine in Linux for years, the alsa pulse switch had some issues but that was like a decade ago
If you think Pulseaudio is really nice to work with, that's neat. It's exactly the same reaction Lennart had with his infamous and often-mocked "but I tested it on my laptop, so it works for me" that he applied to both PulseAudio and systemd. Luckily he appears to have grown out of it. Won't you? Just because it works for you doesn't mean it doesn't give other people problems.

Pulseaudio absolutely sucks for complicated setups compared to PCM in FreeBSD, but admittedly the biggest issue - which was that the lack of documentation - appears to have been fixed. When I was really struggling with it, there was no documentation aside from a list of the command-line switches and an undocumented configuration sample file.

Varkk posted:

I thought that pulseaudio was created to overcome issues with oss and alsa in regards to more complex setups etc. A couple of distros pushed it before it was ready so it didn’t have a great introduction. From memory Fedora used it as default before the developers said it was ready to be used.
PulseAudio was created for the same reason it's now being replaced with TripWire by RedHat, who Lennart still works for: Not Invented Here-syndrome is strong on Linux.

If you compare sound on Linux with sound on FreeBSD, both of them started out as the the exact same codebase (that would at some point come to be named VoxWare) since Jordan K Hubbard imported the sound stack from Linux 1.0c, as Hannu had licensed it as GPLv2. Improvements were made in Linux and regularly imported into FreeBSD.
Then 4Front bought the rights to it and started paying Hannu to develop it for them, and it got renamed to OSS.
At some point after that, OSS became sort-of-proprietary, and instead of simply forking the last GPL version as was already very common then, someone decided to reinvent everything with the argument that it was magically better. Over on the FreeBSD side of things, Luigi Rizzo was maintaining local patches which were essentially a complete rewrite (only file not rewritten was soundcard.h), but maintained compatibility with OSS.
On the Linux side again, ALSA was doing rather well, except of course that everything has to be rewritten to support it - but why worry about a little thing like userspace compatibility in the 90s, or today?
Around 2000, Lennart Pottering decides that ALSA doesn't suit him, so more NIH is needed and PulseAudio begins taking shape for an initial release a few years later. Meanwhile, on FreeBSDs side, Cameron Grant (RIP) rewrote the audio stack again, still maintaining compatibility with OSS but this time supporting the newbus paradigm (which meant drivers didn't care about which bus they attached to, be it PCI, USB, or anything else), as well as added a bunch of new drivers.
By 2005, the initial version of PulseAudio had come out and was magical and everyone loved it because it could do everything everyone ever wanted. On the other side of the pond, Ariff Abdullah picked up the code and added multi-channel support, virtual channels, per-(virtual)channel volume adjustments, low-latency and bit-perfect I/O, removal of Giant locks, and a high-quality resampler (one that regularly beats commercial solutions in quality), plus OSSv4 compatibility and device driver restructuring.

CaptainSarcastic posted:

It's been years since I played with a PPC-based system. About 10 years ago I had good luck with the PPC spin of openSUSE (I think that might have actually been the first time I actually ran openSUSE at all).

At this point I'm not sure what distros have PPC versions available, be they official or unofficial.

If memory serves the biggest thing it was missing was Flash, but that's so deprecated it doesn't seem like a big issue compared to how much of a problem it was at the time.

You could always scrounge up an old image of Yellow Dog. I never had a good experience with Yellow Dog even when it was current.

Fake edit: I did a quick search and it looks Ubuntu doesn't have a recent PPC build, but it looks like openSUSE does.

https://en.opensuse.org/Portal:PowerPC

I'm not sure, but it might be possible to find a BSD to install on PPC, too, but I didn't look extensively.
NetBSD supports PowerPC and a whole bunch of other architectures, to the point that it's almost silly; it might even support more than Linux given that it also supports stuff like the SHARP X68000 (and having NetBSD for troubleshooting instead of using the Human64k OS is wonderful, lemme tell you).
FreeBSD has also had a bunch of work put into PPC and PPC64 in recent years with a few developers using it at their desktop OS, so maybe that's worth checking out?

D. Ebdrup fucked around with this message at 08:20 on May 26, 2020

VostokProgram
Feb 20, 2014



D. Ebdrup, do you run BSD on your main computer? Curious how the desktop experience is.

D. Ebdrup
Mar 13, 2009



VostokProgram posted:

D. Ebdrup, do you run BSD on your main computer? Curious how the desktop experience is.
I run FreeBSD as a desktop OS on my workstation (since 2000) and laptop (since 2015) - but my "desktop experience" consists of a a tiling window manager (ratpoison-wm until 2015 when I switched to i3) with alacritty on one workspace and firefox running on the other, so it's hardly anything to talk about.
There are FreeBSD developers and porters who use gnome, kde, mate, cinnamon, and basically every UI I know of - and there's even a couple of FreeBSD teams attached to the major UI projects.
So far as I've heard, the vast majority of stuff works and is pretty up-to-date.

There are some rough edges, which is probably to be expected?
It certainly seems true of everything except Mac OSX which I abandoned in 2015 because of all the UI changes they made since 10.5, which was the last good version as far as I'm concerned.

mystes
May 31, 2006



Mr Shiny Pants posted:

What do you want to know?
Just how is the experience in general (e.g. is it reliable?) so I can make sure it doesn't suck before I do this.

Mr Shiny Pants
Nov 12, 2012


mystes posted:

Just how is the experience in general (e.g. is it reliable?) so I can make sure it doesn't suck before I do this.

Some objectively cool things:

It works. Like really works, it is just a complete second machine without needing to dual boot.
It is fast. I can play Doom and other games at almost native speeds.

Some not so cool things:
The keyboard & mouse stuff sucks. It is better to have a complete second set. The keyboard works but keeping a button pressed doesn't work. Which is really annoying. I have not tested Synergy.
Weird screen behaviour, like not seeing your mouse cursor in Visual Studio within the editor window.
When the looking glass server quits, your screen is stuck. It is probably best to have a second monitor input which you can switch, or some sort of 4K "fake" monitor dongle.
UAC does not work.

Why do i keep it around?
I like Linux better for coding to be honest and I like to wean myself off Windows.
RDP using Remmina fixes like 100% of the problems with code editing inside Windows and other glitches, it works really smooth. It probably works better than Looking Glass when not gaming or doing high end 3D.
I game on my TV and have a long HDMI cable running to my receiver which I use to start Steam on the VM so I don't really need/use Looking Glass for gaming.

If I do any Unity/Blender/Visual Studio stuff I'll dual boot into Windows because it is more fluid to use than Linux. For other stuff I try to use Linux and have my Windows VM running in the background. Blender works fine on Linux but just switching applications all running on the same OS is just smoother.

ewe2
Jul 1, 2009

TRUMP TRUMP TRUMP

put clinton in prison imo


Lipstick Apathy

D. Ebdrup posted:

PulseAudio was created for the same reason it's now being replaced with TripWire by RedHat, who Lennart still works for: Not Invented Here-syndrome is strong on Linux.

I think you mean PipeWire. Either way, it just confirms for me that my decision to move to Windows for my music production is the sane option, I don't want to wait another half-decade for that piece of poo poo to level out in some future app while the rest of the distro flounders.

mystes
May 31, 2006



Mr Shiny Pants posted:

Some objectively cool things:

It works. Like really works, it is just a complete second machine without needing to dual boot.
It is fast. I can play Doom and other games at almost native speeds.

Some not so cool things:
The keyboard & mouse stuff sucks. It is better to have a complete second set. The keyboard works but keeping a button pressed doesn't work. Which is really annoying. I have not tested Synergy.
Weird screen behaviour, like not seeing your mouse cursor in Visual Studio within the editor window.
When the looking glass server quits, your screen is stuck. It is probably best to have a second monitor input which you can switch, or some sort of 4K "fake" monitor dongle.
UAC does not work.

Why do i keep it around?
I like Linux better for coding to be honest and I like to wean myself off Windows.
RDP using Remmina fixes like 100% of the problems with code editing inside Windows and other glitches, it works really smooth. It probably works better than Looking Glass when not gaming or doing high end 3D.
I game on my TV and have a long HDMI cable running to my receiver which I use to start Steam on the VM so I don't really need/use Looking Glass for gaming.

If I do any Unity/Blender/Visual Studio stuff I'll dual boot into Windows because it is more fluid to use than Linux. For other stuff I try to use Linux and have my Windows VM running in the background. Blender works fine on Linux but just switching applications all running on the same OS is just smoother.
Thanks for the detailed information. I'm already using Barrier for the keybord/mouse with my current two computer setup and I already actually ordered a fake monitor dongle just in case so I'll see if I can get it to work decently with that. I guess if all else fails I could switch between RDP and changing the monitor to the VM for stuff that requires graphics acceleration but there might be a point where it isn't worth the trouble compared to using two computers.

It would really be nice to clean up my desk so I just have one bigger monitor and one computer without a rats nest of cables, though.

Edit: I guess once WSL2 has the new gui support I could theoretically just use windows but that seems like sacrilege.

mystes fucked around with this message at 14:07 on May 26, 2020

taqueso
Mar 8, 2004









Fun Shoe

D. Ebdrup posted:

If you think Pulseaudio is really nice to work with, that's neat. It's exactly the same reaction Lennart had with his infamous and often-mocked "but I tested it on my laptop, so it works for me" that he applied to both PulseAudio and systemd. Luckily he appears to have grown out of it. Won't you? Just because it works for you doesn't mean it doesn't give other people problems.

You keep saying pulse doesn't work for you but it works for everyone else, maybe you should grow out of that. I'm kidding, but you haven't said what makes it bad as far as I can see. (beyond docs)

I haven't had a chance to use PCM beyond whatever is the default in freebsd or netbsd, but I do have an application coming up where I could. Would I be able to do live DSP/digital crossover stuff? I'd seriously give it a try.

My pulse is good experiences involve manual configuration, maybe the UI is better for PCM for general use?

Mr Shiny Pants
Nov 12, 2012


mystes posted:

Thanks for the detailed information. I'm already using Barrier for the keybord/mouse with my current two computer setup and I already actually ordered a fake monitor dongle just in case so I'll see if I can get it to work decently with that. I guess if all else fails I could switch between RDP and changing the monitor to the VM for stuff that requires graphics acceleration but there might be a point where it isn't worth the trouble compared to using two computers.

It would really be nice to clean up my desk so I just have one bigger monitor and one computer without a rats nest of cables, though.

Edit: I guess once WSL2 has the new gui support I could theoretically just use windows but that seems like sacrilege.

It is reliable, I have not had any hard locks or anything with it. So if you are comfortable running 2 computers the way you are using them now, I think it will work out pretty well.

Just running Linux as a main for the way updates work was a big one for me. No reboots most of the time and you can just choose when to change stuff.

The funky stuff for me is the way I have it setup when running through my receiver, sometimes it will just hang the Looking glass host because of Display Port doing some funky things when switching inputs. If you have a dedicated dongle I am pretty sure it is rocksolid.
One thing though, kernel 5.4 has changed some stuff with VFIO which is giving me problems isolating my second GPU. I have to run 5.3 to have it work.

mystes
May 31, 2006



Mr Shiny Pants posted:

It is reliable, I have not had any hard locks or anything with it. So if you are comfortable running 2 computers the way you are using them now, I think it will work out pretty well.

Just running Linux as a main for the way updates work was a big one for me. No reboots most of the time and you can just choose when to change stuff.

The funky stuff for me is the way I have it setup when running through my receiver, sometimes it will just hang the Looking glass host because of Display Port doing some funky things when switching inputs. If you have a dedicated dongle I am pretty sure it is rocksolid.
One thing though, kernel 5.4 has changed some stuff with VFIO which is giving me problems isolating my second GPU. I have to run 5.3 to have it work.
Cool thanks. I really want to try this soon but I probably also need to upgrade my PSU in order to move the video card from my windows computer and they're out of stock and/or stupidly expensive everywhere right now so it might end up being a month or two until I can actually do it.

SoftNum
Mar 31, 2011



ewe2 posted:

This is stupid and unhelpful, grow up.


D. Ebdrup posted:

If you think Pulseaudio is really nice to work with, that's neat. It's exactly the same reaction Lennart had with his infamous and often-mocked "but I tested it on my laptop, so it works for me" that he applied to both PulseAudio and systemd. Luckily he appears to have grown out of it. Won't you? Just because it works for you doesn't mean it doesn't give other people problems.

Pulseaudio absolutely sucks for complicated setups compared to PCM in FreeBSD, but admittedly the biggest issue - which was that the lack of documentation - appears to have been fixed. When I was really struggling with it, there was no documentation aside from a list of the command-line switches and an undocumented configuration sample file.

PulseAudio was created for the same reason it's now being replaced with TripWire by RedHat, who Lennart still works for: Not Invented Here-syndrome is strong on Linux.

If you compare sound on Linux with sound on FreeBSD, both of them started out as the the exact same codebase (that would at some point come to be named VoxWare) since Jordan K Hubbard imported the sound stack from Linux 1.0c, as Hannu had licensed it as GPLv2. Improvements were made in Linux and regularly imported into FreeBSD.
Then 4Front bought the rights to it and started paying Hannu to develop it for them, and it got renamed to OSS.
At some point after that, OSS became sort-of-proprietary, and instead of simply forking the last GPL version as was already very common then, someone decided to reinvent everything with the argument that it was magically better. Over on the FreeBSD side of things, Luigi Rizzo was maintaining local patches which were essentially a complete rewrite (only file not rewritten was soundcard.h), but maintained compatibility with OSS.
On the Linux side again, ALSA was doing rather well, except of course that everything has to be rewritten to support it - but why worry about a little thing like userspace compatibility in the 90s, or today?
Around 2000, Lennart Pottering decides that ALSA doesn't suit him, so more NIH is needed and PulseAudio begins taking shape for an initial release a few years later. Meanwhile, on FreeBSDs side, Cameron Grant (RIP) rewrote the audio stack again, still maintaining compatibility with OSS but this time supporting the newbus paradigm (which meant drivers didn't care about which bus they attached to, be it PCI, USB, or anything else), as well as added a bunch of new drivers.
By 2005, the initial version of PulseAudio had come out and was magical and everyone loved it because it could do everything everyone ever wanted. On the other side of the pond, Ariff Abdullah picked up the code and added multi-channel support, virtual channels, per-(virtual)channel volume adjustments, low-latency and bit-perfect I/O, removal of Giant locks, and a high-quality resampler (one that regularly beats commercial solutions in quality), plus OSSv4 compatibility and device driver restructuring.


This is clearly an accurate representation of what happens. IDK why everyone else has to grow up.

Also slightly more content based, ALSA was never really "fine" it was just kind of there. ALSA / Pulse stack used to suck as they fought for things but it's fine now. Sure you gotta get it configured right, but that's linuxexperence.txt

EDIT: GPU Passthrough / EMU chat: It's worth noting that some games STILL don't work through passthrough on kqemu. Valorant is a big one that just refuses to run, and other games like FF14 are way more sensitive to DX/Heat issues than is normally experienced. I've heard of problems with Destiny and other games with super-invasive anti-cheat.

Also worth noting that with Vulkan becoming way more a thing Lutris / Proton / wine are now much better than they used to be. All Blizzard games work native without an VM, and most non cutting edge steam stuff "just works" in proton. THe only things I need to do on my VM (and this is more launcher problems than game problems) are: Rainbow 6, FF14, Apex, and some bad rts I forget off hand.

tl;dr: do some research in your intended use case to make sure you get what you want.

SoftNum fucked around with this message at 16:33 on May 26, 2020

Mr. Crow
May 22, 2008

Snap City mayor for life


In my experience with pulse audio on Linux is that it works out of box for most situations but as soon as you start doing anything outside of the most basic use cases flip a coin if it will work for you and expect a complete nightmare if it doesn't.

The newer your hardware the more likely you are to have a bad time.

I'm not really sure that can be said about much else in Linux these days, driver support et al is pretty top notch out the gate.

So yes the Linux audio stack still deserves ridicule and shame despite that it probably works ok for most users.

The Milkman
Jun 22, 2003

No one here is alone,
satellites in every home


Lipstick Apathy

Granted I don't do Serious audio production and mostly stick to Fedora, the playground for the dastardly Lennart's designs, but I've never, ever had a problem with PulseAudio, even on a brand new X570 board with supposedly fancy audio. Whereas in the old days with ALSA/OSS/whatever just basic audio playback was busted most of the time and people were right to write off desktop linux

SoftNum posted:

Also worth noting that with Vulkan becoming way more a thing Lutris / Proton / wine are now much better than they used to be. All Blizzard games work native without an VM, and most non cutting edge steam stuff "just works" in proton. THe only things I need to do on my VM (and this is more launcher problems than game problems) are: Rainbow 6, FF14, Apex, and some bad rts I forget off hand.

Proton is really great now and it's worth giving it a shot before you try dealing with GPU passthrough. It's really wild just being able to mash the Play button on Steam like I normally would on Windows and having it all Just Work, excepting certain invasive anti-cheat stuff like what's in Destiny 2. Lutris works too, but most of the games I play are through Steam anyway and I've never really had to reach for anything besides the built in stuff.

D. Ebdrup
Mar 13, 2009



"You call that single DSP+OpAmp which fits on a motherboard a knife? This is a knife"

D. Ebdrup fucked around with this message at 22:16 on May 26, 2020

Mr. Crow
May 22, 2008

Snap City mayor for life


The Milkman posted:

Granted I don't do Serious audio production and mostly stick to Fedora, the playground for the dastardly Lennart's designs, but I've never, ever had a problem with PulseAudio, even on a brand new X570 board with supposedly fancy audio. Whereas in the old days with ALSA/OSS/whatever just basic audio playback was busted most of the time and people were right to write off desktop linux

Funny I'm running the exact same setup and have had some annoying issues, what brand? Running the aorus master.


And yeah native Linux gaming is pretty much here, if you're browsing this thread you're the target audience. Don't think it's ready for Joe Blow but it's largely low barrier to entry with proton and protondb.

I have a dual boot setup I use rarely for some specific games (the mentioned ones with anti cheat) but most multiplayer games I play work fine cross platform.

CaptainSarcastic
Jul 6, 2013

HAIL SATAN


Mr. Crow posted:

Funny I'm running the exact same setup and have had some annoying issues, what brand? Running the aorus master.


And yeah native Linux gaming is pretty much here, if you're browsing this thread you're the target audience. Don't think it's ready for Joe Blow but it's largely low barrier to entry with proton and protondb.

I have a dual boot setup I use rarely for some specific games (the mentioned ones with anti cheat) but most multiplayer games I play work fine cross platform.

I'm running an X570 Aorus Elite and the sound has been working just fine. No significant problems here. I do seem to remember reading about people having problems with the sound output from my board's rear ports, but that was independent of OS and wasn't a problem for me.

The Milkman
Jun 22, 2003

No one here is alone,
satellites in every home


Lipstick Apathy

Mr. Crow posted:

Funny I'm running the exact same setup and have had some annoying issues, what brand? Running the aorus master.

I've got the ASUS ROG STRIX X570-I.

Warbird
May 23, 2012

Burn the 'dawgs
Kill the Yellowjackets
Purge the Tiger
It is better to die for Bama than to live for yourself


Fun Shoe

No one told me tab completion was a thing until like a year ago and I think I have a complex now.

CaptainSarcastic
Jul 6, 2013

HAIL SATAN


I just checked and the Aorus Elite and the Aorus Master use different audio chipsets.

unimportantguy
Dec 25, 2012

Maybe she has no parents and was raised by dogs?

Now I'm nervous about my audio setup. I have my monitor speakers, a headset, and a separate recording microphone, and how well it all works with Pulse has been a real eye-opener for me after the mess that was trying to get this all working with ALSA back when I was running Ubuntu in '06 or so. Now I feel like I should be waiting for the other shoe to drop.

Antigravitas
Dec 8, 2019

Outside Context Problem


Tbh., my Linux audio experience has been vastly better than on Windows.

Just the ability to quickly drag&drop applications onto the output device I want them to use is an amazing usability feature that I have not seen anywhere else. Similarly, being able to control each application's audio volume from a single panel is incredibly useful.

Pulseaudio also doesn't forget my microphone settings each update, while every single Windows update pretends the Samson GoMic that is connected now is completely different from the Samson GoMic that was connected before. The worst part is that my screams of rage can't be heard until I go into the settings again…

D. Ebdrup
Mar 13, 2009



Warbird posted:

No one told me tab completion was a thing until like a year ago and I think I have a complex now.
This might slightly interest you then: TENEX, which TOPS 20 (the OS for the 36-bit DEC mainframe that was launched in 1969) was based on, is the OS that invented command completion, and TENEX is also where the derivative of the C shell, called tcsh, gets the t from - because it was the first shell on BSD (and any Unix-like) that had command completion.
Incidentally, it provides command completion via regular expressions, so making command completion for your favorite commands is really easy. I happen to have some examples from my .tcshrc (down near the bottom of the file).

Something else that might interest someone is that the -rc suffix on many files, the rc shell in Plan 9 from Bell Labs, and the rc(8) service daemon on FreeBSD all derive the meaning of rc from the RUNCOM command that existed in MITs Compatible Time Sharing System from 1965.

Antigravitas
Dec 8, 2019

Outside Context Problem


I want tab-completion for forum posts.

mystes
May 31, 2006



Antigravitas posted:

Tbh., my Linux audio experience has been vastly better than on Windows.

Just the ability to quickly drag&drop applications onto the output device I want them to use is an amazing usability feature that I have not seen anywhere else. Similarly, being able to control each application's audio volume from a single panel is incredibly useful.

Pulseaudio also doesn't forget my microphone settings each update, while every single Windows update pretends the Samson GoMic that is connected now is completely different from the Samson GoMic that was connected before. The worst part is that my screams of rage can't be heard until I go into the settings again…
It's sort of off topic, but for windows you might want to try this volume control which makes things bit more convenient than having to go into the settings to change what audio is playing on which device: https://github.com/File-New-Project/EarTrumpet

Personally I haven't had problems with PulseAudio in years. Maybe if there hadn't been the OSS licensing issue it would have been better to continue with that rather than developing ALSA/PulseAudio but that's not how things happened.

xtal
Jan 9, 2011



So, I'm setting up a NAS again, using a raspberry PI and two external hard drives (wd blues in powered, ventilated enclosures.)

What's the best way to set this up? I previously used btrfs to have one filesystem copied across the two drives, but it didn't behave like I expected (wouldn't read one drive if the other died.)

Should I use btrfs raid, zfs raid, lvm raid, or just manually sync the files? My current plan is to have the two drives be mirrored but still run independently, storing snapshots of the the drives on the drives themselves (either filesystem snapshot or cp -a), and periodically storing a full copy of the drives on Glacier.

xtal
Jan 9, 2011



Also could be off topic at this point but I would like to hear about UPS for graceful shutdown to prevent drive corruption.

Matt Zerella
Oct 7, 2002


personally I would not use any hashing or COW filesystems on a pi.

LVM mirror will probably be the lowest overhead but you've still got USB overhead there.

xtal
Jan 9, 2011



Matt Zerella posted:

personally I would not use any hashing or COW filesystems on a pi.

LVM mirror will probably be the lowest overhead but you've still got USB overhead there.

Not super concerned about the throughput, and if I was, I would just switch to a different SOC. I also like having transparent compression.

My main priorities are that the drives be mirrored, but separable, and usable by other machines. (Assuming that separable means one can boot without the other, not some kind of merge conflict resolution.)

D. Ebdrup
Mar 13, 2009



So long as the OS supports atomic operations for the architecture used, most COW filesystems should run just fine.
Problem is, only FreeBSD supports atomics for any 32bit platform, and it's only ia32.
On the other hand, zfs on aarch64 runs on as little as 512MB - although admittedly it's no speed-daemon.

xtal
Jan 9, 2011



D. Ebdrup posted:

So long as the OS supports atomic operations for the architecture used, most COW filesystems should run just fine.
Problem is, only FreeBSD supports atomics for any 32bit platform, and it's only ia32.
On the other hand, zfs on aarch64 runs on as little as 512MB - although admittedly it's no speed-daemon.

The PI I'm using uses aarch64, so is it time for me to learn about zfs?

D. Ebdrup
Mar 13, 2009



xtal posted:

The PI I'm using uses aarch64, so is it time for me to learn about zfs?
Go hog wild, friend

Truga
May 4, 2014


Lipstick Apathy

Antigravitas posted:

Tbh., my Linux audio experience has been vastly better than on Windows.



alsa was really powerful and has been my go-to for doing custom sound stuff since like 2003 or whenever they got dmix working, but requires manual tampering in a terminal for any custom scenarios so pulse audio was created.

pulse audio is only really different in that it has a gui option for some stuff. still, it provides everything i need from a sound system so i can't really complain. i guess someone could have just provided a decent gui interface for alsa and we wouldn't have to go through the process of uninstalling pulse in 2008-9 when it didn't really work but still got pushed in every distro, but that's probably too simple.

best thing about pulse tho, when you break it, the fix is always as easy as killall -9 pulseaudio

pipewire is probably gonna be pulse 2.0: audio boogaloo. i'm cautiously optimistic, but also prepared to hate it for the first 3 years.

windows instead just requires you to buy an audio card to do realtime sound stuff. this is not a big problem for strictly professional sound stuff because everything will support it. however, if you're doing stupid bullshit like me you need buggy/laggy 3rd party apps that will convert your normal wsapi application sound to asio streams, which introduces some delay and sometimes causes weird artifacts/pops to happen in sound because windows scheduling is apparently a barbage fire and realtime priority means nothing lmao

Mr. Crow
May 22, 2008

Snap City mayor for life


Antigravitas posted:

Tbh., my Linux audio experience has been vastly better than on Windows.

Just the ability to quickly drag&drop applications onto the output device I want them to use is an amazing usability feature

Which desktop/settings manager has this? I'm running xfce and it's pretty painless drop down but that sounds even niftier.

Adbot
ADBOT LOVES YOU

D. Ebdrup
Mar 13, 2009



It's called cating from a file and piping to another file, isn't it?
You know, the thing so smart even CP/M (and therefore Windows) stole the idea.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«779 »