I use Jellyfin. I think in your use case, each user would be setup have their own library. You can enable or disable library on a per user basis as will as a per client basis.
Downside is that the default web interface isn’t great as a music player. It does the job but it’s not great.
Other hand, multiple music-first clients exist for a lot of different platforms. Odds are good you can find a client that suits how you listen to music.
Edit: said collection when I meant library.
It’s doable. Stick to the 7b models and it should work for the most part, but don’t expect anything remotely approaching what might be called reasonable performance. It’s going to be slow. But it can work.
To get a somewhat usable experience you kinda need an Nvidia graphics card or an AI accelerator.
@tal has already given a really good answer. To add to it, this thread might help you some: https://lemmy.sdf.org/comment/11963996 I was asked what I thought was “better” than a raspberry pi. Came back with an eBay search and a trio of suggestions in the price range of a Pi 4. TLDR is whatever you have currently will probably work fine but if you need to buy hardware, there are plenty of low cost options. And of course, Pi’s also work fine for anything they are capable of, which is most things.
When I started self hosting, Raspberry Pi’s were the cheapest option available. I learned fairly quickly that the SD card was the weakest part of them but not long after the Pi3 came out we were able to boot off of USB drives which solved that issue. I think I had 8 SSDs hanging off of one pi before I finally decided to plop down the money for a tower. I then added a pair of 6 port SATA cards and added even more storage to that system. Eventually I was hosting so many things that I was running out of RAM, So I bought a second used tower, this one with a much newer processor and a lot more RAM. Now I run both with the old system running as a NAS and the new system hosting my other services. I wouldn’t stress about hardware too much. Hardware can grow with you, to a point.
Mini PCs are too small to house internal drives
Most mini PCs I’ve heard of (and quite a few thin clients) use m.2 drives for internal storage. Not difficult to upgrade. I’ve also heard of a few that had ports and internal space for 2.5 inch SSDs.
I don’t know of ANY reason to go with spinning-platters, nowadays.
Price per terabyte is lower on HDDs. For bulk storage they are currently the best path. SSDs are catching up though, and there are cases where a SSD based NAS does make sense. But most folks at home don’t have the network capability to fully utilize their speed. Network becomes the bottleneck.
Not that I’m aware of.
The only time I heard anyone talking about it was on the podcast Self-Hosted . Supposedly it’s a NUC clone with performance similar to a then current (2023) mid range laptop and draws about the same amount of power. I think they said the N100 processor had Intel QuickSync for hardware transcoding.
Sure, but “better” is massively subjective. For me, when I set up a pi, I’m not usually making use of the GPIO or the camera inputs. I’m generally throwing together a headless server. To do that, in addition to the board itself, I need storage, power, heat sinks, an fan and usually some sort of case.
Using the prices at CanaKit as a rough guide, you can come up with this search on Ebay.
The first entry I saw drew my attention. It’s a 7th gen i5 with 16GB RAM and a 120 GB SSD. Not sure the 500 GB HDD would survive shipping, but it’s $100 shipped. Biggest concern is that the seller only has 65 sales. Possible scam?
On the higher end of that bracket there is this. 6th gen and only 8GB RAM, but the seller does have a history.
With the prices on the Pi5 your potentially getting into the price range where it might make sense to look at the Beelinks mini PCs, based around a 12th gen Intel.
Like I said, prices right now are at a spot where I can’t just say throw a Raspberry Pi at the problem. They are great boards but for someone self-hosting their own services they don’t necessarily always make sense anymore.
Given how old the system is, I’m not sure how long it would survive that type of duty. Power up and downs are a lot rougher on components than if they just stay running.
If you switch the HDD for a pair of SDD (one storage, one swap), it would be somewhat useable. Better to increase the amount of RAM if possible. If I remember correctly, 2-4 GB of RAM was not uncommon at this time period. Although NixOS or a really light Debian install might be able to stay within that amount of RAM. So yea, I think it’s feasible.
Good Idea? Perhaps not so much. That proc has a TDP of 95W. Haven’t found anything on it’s idle power draw, but I’d guess that that system would have a fairly heavy power draw. The slow speed of the processor and low amount of RAM would probably limit the amount of traffic you could put through it. Additionally, the age of the components would probably cause reliability issues.
Generally I like to tell folks to use what they have. Repurposing old hardware is better for the environment and usually the wallet, but this system would probably would not be my first, second or even third choice for any workload. I haven’t found a benchmark comparing the two, but I think a Pi3 would probably run dead even with this system at a far lower power draw. Although the Pi3’s ethernet does run on it’s USB bus (I think), along with it’s storage, so that would slow it down for this workload. If you wanted to run traffic faster, I would probably look into the used micro PC market at the $75-$150 USD price point. This system is old enough to vote. Something merely 10 years old would be considerably faster.
If you mean for ARM based systems (not just SBCs), I would agree, but the software and support ecosystems for amd64 systems far surpasses even the rPi ecosystem because you have backwards compatibility with a lot of the legacy x86-64 and x86 code. And because they support UEFI, distributions don’t need to explicitly support your particular version of your ARM processor so you can run pretty much whatever OS you want.
Not long ago I saw a one of those old small Dell Optiplex workstations with a 4th gen i3, 8GB ram and a 256GB SSD on Amazon for $100 USD. There’s a new BeeLink with an N100, 16 GB RAM, and 500 GB SSD for $200. They’d both be great for any home lab project that doesn’t need the GPIO of the rPi. And they are both in the same price range.
Don’t get me wrong, if I needed to kitbash a desktop or small server together in a hurry, I would probably be using a Pi3 or Pi4 because I’ve 6 of them collecting dust from when my self hosted services outgrew their available compute. I replaced them with a keyboard damaged laptop with a 6th gen i5 and my old desktop with a 4th gen i5. But if I needed to buy something today, I’d be doing some price comparisons first.
If you like Pi’s, use them. They are great kit. But if price or (more recently) power consumption are your primary consideration, it’s no longer as simple a choice as it was pre-pandemic. It’s worth looking around now.
Of course, none of this applies if you need the GPIO. But then you’re looking for project boards, not desktop or home server systems. Different set of criteria. And a different set of head aches.
Been able to use rPis as a desktop for a while now. The 2s and 3s weren’t particularly pleasant but it was doable. The Pi 4 8GB with an USB3 jump drive as root partition was a lot more pleasant, at least until you hit thermal throttle.
Right now though, there are more powerful options in the same price point, once you account for power, storage and optionally, a case. At least for desktop and home server use.
The Raspberry Pi’s just aren’t the go to hardware for the home lab anymore. Probably won’t be again unless the price comes back down on the Pi’s or the price on new and used amd64’s goes back up.
I didn’t use early generation smart phones and was completely bewildered when I discovered apps often used swiping left/right to interact. No app I had used before ever indicated that was an option. I suggested we should add indicators to our app to teach people but that was rejected because “everyone knows that”. It’s easy when you know how.
Oh it’s even worse when you did have experience with early smartphones. I’ve used Windows CE phones, Blackberrys, PalmOS phones, early Androids, and since 2015, iOS. None of them did things the same way, but all navigated using clickable objects on the screen. I was shocked when I had accidentally stumbled upon gestures. In 2017.
I’m still discovering new gestures, usually by accident. It’s becoming more intuitive, but only because I now know that it might be an option.
A cursory web search suggests your CPU may have been damaged. Can’t say for sure as I don’t know jack about modern hardware. My indepth knowledge basically ends at x86.
I found this while searching the error codes in the photo. Maybe it will be of some help.
It’s not difficult to self host. Pretty light on resources. Documentation on how to do so could use some work though. I believe I used a docker image to get up and running.
The main reason I personally don’t allow public signups on my instance is that US law is rather chaotic. If section 230 gets cancelled or repealed I don’t want to be held responsible for what some random person chose to write. It may not be a big risk at the moment but I don’t have the mental bandwidth to deal with it.
Depends on what you are trying achieve.
You can send sms via foss no problem as long as you know what carrier the recipient uses. All of the carriers seem to have email bridges to the sms network. Receiving sms is another question entirely. To receive sms, the network needs to know where to send the message. There are commercial platforms that can link you into the sms network.
Another option is KDEConnect which can link your Android device to your computer and you can send and receive sms that way.
Heads up on the copyright thing. Copyright is different nation to nation. @ThePowerOfGeek@lemmy.world seems to be out of the UK or EU. Not sure what the copyright situation is like there but here in the US, anything you write is already protected under US copyright laws from the moment it’s published (such as when I hit “post” here), subject to any applicable agreements you’ve entered into, of course.
You don’t HAVE to register your work for it to be under copyright protection, but to doing so would give you a stronger case if you ever decided to go to court over copyright. To register a work in the US you would do so through the Copyright Office.
In general though, @ThePowerOfGeek@lemmy.world is right though, you should assume anything you put out in the wild will be used in a manner you never intended, and that you may not like.
For examples of how helpful copyright protection is in a practical sense, might want to check out c/piracy.
For something like this, I think I would use Hugo to generate a static site. You write the post using markdown then tell Hugo to regenerate the site.
If you want to publish and receive comments or feedback via the fediverse, you might take a look at WriteFreely. Simpler to use but more complicated to setup.
Wordpress is also an option, but I personally don’t like it much. I spend all my time playing with all the knobs and never actually publish anything.
https://github.com/awesome-selfhosted/awesome-selfhosted#blogging-platforms
As everyone else has said, if your time is limited, your best path is docker. You don’t need to learn all of docker, but understanding how docker compose works at a fairly high level will drastically speed up setup as well as administrative tasks like updating and backups
As for what to run, you mentioned wireguard and a notes app. The notes app could be solved without needing a central server with Obsidian and I’m not seeing the use case here for Wireguard.
I would start with what problem or pain point are you trying to solve for.
In my case, I had a bunch of IOT devices all making excessive DNS queries and I wanted a network level ad blocker so I setup PiHole (2 in fact, they run my network’s DNS).
I had a large music collection and burning mix CDs was no longer practical so I setup Jellyfin (Navidrome might have also worked), and use FinAmp on my phone.
Google started being a pain in my backside so I setup Nextcloud.
Someone got me some smart devices so HomeAssistant was setup.
I needed a way to find these services so I setup Heimdel as a dashboard.
I wanted some of these publicly available so I setup Caddy as a reverse proxy.
I’ve bounced between both over the last 20 years. The main difference between Gnome and KDE is that KDE has always been far more customizable. Gnome has better support for Wayland over KDE 5, however that is not true for KDE 6. I’m not sure which is default currently for the Fedora KDE spin.
My personal take on the current Gnome DE is that it is a very different way of conceptualizing the desktop from what I’m used to, to the point that it puts me off. I had the same issue with the Unity DE on Ubuntu back when. While it’s not for me, a lot of folks do seem to like it. I’ts quite usable, but I wind up spending time fafing around trying to figure out how to get things done rather then just doing what it is I’m trying to do. Muscle memory runs deep and KDE keeps to the traditional Windows desktop feel (Win95 - Win 7) with a few nice upgrades. Gnome ( at least current Gnome) does it their own way.
I’m going to have to take a look at this! Thanks!
Me too! Not much to look at but it’s a great player on iOS. On Linux, I like SonixD.