Yet none of this will reduce Fuckerberg’s cash pile. We need to take that back.

This sounds like some RFK “facts” bullshit. I assume the number is not actually zero, but will wait for the math.

Clop Slop

Also, possibly an STI? “They got the Clop Slop!”

Edit: Actually, I think just “Clop” works

MacOS, phones, off-brand handhelds that run SteamOS…that’s the goal.

It’s not going to be effortless. That’s not really the goal. Anything that uses Google Play Services is going to be a problem still as the underlying service layer just mocks out those API calls a la Waydroid. They’re working more on the FEX stuff from what I’ve seen in the repos.

Having the disks connected externally is the same as having them connected internally

No, it 1000% is not, especially in the case of USB that I used. Even in the way Linux handles everything as a file and target, it is vastly different.

No RAID solution I know of would lose the array on a power outage

Hardware RAID enclosures have batteries on the disk controllers for this very reason. We aren’t talking about those though, we’re talking about software RAID on JBOD, which wouldn’t have those sanity protections. Here’s some random blog explaining deeper.

Honestly I don’t see how interrupt handling would be any different between internally or externally connected devives, except for different buses/protocols handling it differently intrinsicly

See above

Maybe I’m too spolied by using ZFS, but again I don’t think this would actually be a problem

That’s a filesystem solution to a hardware problem, so yes, probably a bit spoiled there, or at least it’s skewing your understanding of what RAID is and how it works. One of the reasons ZFS exists, actually. It’s nice to have nice things though.

Well…don’t overspeak. There WILL eventually be something better, and then people will complain about that as well 🤣

Also software. Literally in the description and options.

There are very few use cases for hardware controllers anymore, and they are on SAN controllers at a massive scale. Every single device you point me to at under $50k is going to be software.

Just wanted to clarify so you understand.

Software. They’re all software. They run Linux, actually.

Link me to a single hardware controlled disk array that you’re considering.

The main issue is statefullness of the host.

Say you’re on a laptop, and you get an external JBOD box without any hosted controller. You use that laptop to setup a RAID1 array on 2 disks, and go about your business. Few weeks in you’re in the middle of some editing of video or whatever, and you have a power outage.

That RAID array is assuredly damaged or dead. Your host machine being the controller in the middle of a write when the entire array dissapears is going to give up quickly, and the cached data in flux to write is gone. You miiight be able to recover the array if you’re lucky, but whatever you’re working on is gone.

A number of diff5scenariis where this may happen exist without a power outage, but the problem is the target not being able to manage its own interrupt, and you have two different states in two different devices that won’t match. It’s toast.

There’s a few things at work here:

1. Not much “hardware” RAID anymore because offloading works just fine and doesn’t draw excessive resources.
2. It sounds like you want to just take your existing disks and pop them into something else, which won’t work.
3. You shouldn’t be running RAID over any external connections for a number of reasons if the coordinator (your machine) is hosting it. I can go deeper into that if you want.

You want a self-contained NAS that manages its own RAID and disks. I would honestly just get a diskless unit and start clean. You’ll be better off in the long run.

Why would I care about swaying your opinion? Nobody here responding to you is invested in YOUR opinion on the matter, or cares what you think about it. They are simply correcting your misinformed attitude about some things from what I can see.

If anything they’re concerned you’re running around in the world with misguided opinion, and potentially misinforming others.

You sound new to the ecosystem at large, and I don’t mean that to be condescending, just that you may not have all the context needed to understand why it exists. Any distro that exists right now can flip back to SysV if they want to. They just don’t want to. It may be more flexible to the neckbeards, but it’s massively more comprehensive in scaling and integrating than a set of Init scripts. It has huge benefits to system integrators, OEMs, and especially the people who manage the largest concentration of Linux deployments: Datacenter Ops teams.

The fact that you, a Desktop user takes issue with that is meaningless to the ecosystem at large. I manage thousands of deployed bare metal machines, and I’d never switch back, because it SysV was fucking painful. Sure it was easier to debug in some cases, but was it as useful or reliable? Not even close.

Just go use something else and stop letting it bother you. You’ll feel better in the long run.

It’s Open Source. Nobody needs to use it, and it’s especially not all-inclusive. That being said, it’s also not new at all as it’s been around in most distros for well over a decade. It has its pros and cons like anything.

Your assumption that “freedom” has something to do with Linux writ large is misguided though. You have distros that have communal decision making, and if they find a benefit to systemd, then they’ll use systemd. Don’t use that distro if you don’t like it. There’s your freedom of choice.

Not really unless you need a specific optimization or module that isn’t available otherwise. Most distros make distribution of external modules available via package manager, and most of the optimizations you would want to enabke can be turns on or off elsewhere as feature flags.

The dependency chain for building the entire kernel depends on what you included in the feature set. The reason it probably worked on Ubuntu is because the build-essential package covers the most common deps needed to build the kernel.

Need output to be able to tell you anything. Post the errors.

Regardless, if you’re building something without a purpose, assign it a dedicated purpose instead of just making it some other running machine.

You literally say in your post you’re building another machine.

Make it a single purpose machine that does the thing you need it for.