There’s also the option of just leaving an offline disk at someone’s and visiting them regularly to update the backup.

Having an entirely offline copy also protects you/mitigates against a few additional hazards.

If you don’t process any user data beyond what is technologically required to make the website work, you don’t need to inform the user about it.

Should have just been a reply.

I doubt most user have any need for great nc performance.

I also doubt those “super performant nextcloud flakes” are actually any faster than a plain old default nc deployment; especially for our use-cases.

Using NixOS is a good recommendation though. Just don’t do flakes unless you actually understand what problem they intend to solve and how catastrophically bad they are at it.

I’d suspect the bots would just try again with a masked user agent when they receive a 403.

I think the best strategy would be to feed the bots shit that looks like real content.

The web version works without an account? That’d be news to me.

I wouldn’t go ARM unless you really like tinkering with stuff.

I bought a used Celeron J4105-based system years ago for <100€ and it’s doing just fine. The N100 is its successor that should be better in every way.

Don’t be afraid to buy cheap used hardware. Especially things like RAM or cases that don’t really ever break in normal usage.

Two 4TB HDDs for 120€ each is a rip-off. That’s twice what you pay per GB in high capacity drives. Even in the lower capacity segment you can do much better such as 6TB for 100€.

If you have proper (tested!) backups and don’t have any specific uptime requirements, you don’t need RAID. I’d recommend getting one 16TB-20TB drive then. That would only cost you as much as those two overpriced 4TB drives.

Sure but that won’t do anything about software issues :p

Specifically this section:

Why is Magic Earth free? What is the business model?

Magic Earth is free for all our end-users but we also have a paid Magic Earth SDK for business partners. For instance Selectric.de (a supplier for navigation solutions for ambulances and fire trucks), Smarter AI (developing ADAS systems) or Absolute Cycling (using the platform on bicycles). For more info on the SDK, you can check magiclane.com.

after a week of runtime it told me 2.5kwh average. could be average per hour

If it gives you kWh as a measure for power, you should toss it because it’s obviously made by someone who had no idea what they were doing.

If this was over an hour, yes. Though you’d typically state it as 100W ;)

I was drawing an average of 2.5kWh after a week of monitoring my whole rack

That doesn’t seem right; that’s only ~18W. Each one of those systems alone will exceed that at idle running 24/7. I’d expect 1-2 orders of magnitude more.

Have you heard of peertube?

It’s slightly overkill for your purposes but it is basically a self-hosted Youtube with a similarly nice UI and everything.

Without knowing what you’ll use it for, neither.

Both don’t sound ideal though w.r.t. power consumption.

Oh absolutely; I would never advocate against verifying your data’s integrity.

I think I’d split that into two machines; a low power 24/7 server and a on-demand gaming machine. Performance and power savings don’t go well together; high performance machines usually have quite high idle power consumption.

It’d also be more resilient; if you mess up your server, it won’t take your gaming machine with it and vice versa.

putting all the components together to be a step up in complexity too, when compared to going pre-built. For someone who is comfortable with building their own PC I would definitely recommend doing that

I’d recommend that to someone who doesn’t know how to build a PC because everyone should learn how to do it and doing it for the first time with low-cost and/or used hardware won’t cause a great financial loss should you mess up.

Interesting. I suspect you must either have had really bad luck or be using faulty hardware.

In my broad summarising estimate, I only accounted for relatively modern disks like something made in the past 5 years or so. Drives from the 2000s or early 2010s could be significantly worse and I wouldn’t be surprised. It sounds like to me your experience was with drives that are well over a decade old at this point.

JBOD is not the same as RAID0

As far as data security is concerned, JBOD/linear combination and RAID0 are the same

With RAID0, you always need the disks in sync because reads need to alternate. With JBOD, as long as your reads are distributed, only one disk at a time needs to be active for a given read and you can benefit from simultaneous reads on different disks

RAID0 will always have the performance characteristics of the slowest disk times the stripe width.

JBOD will have performance depending on the disk currently used. With sufficient load, it could theoretically max out all disks at once but that’s extremely unlikely and, with that kind of load, you’d necessarily have a queue so deep that latency shoots to the moon; resulting in an unusable system.
Most importantly of all however is that you cannot control which device is used. This means you cannot rely on getting better perf than the slowest device because, with any IO operation, you might just hit the slowest device instead of the more performant drives and there’s no way to predict which you’ll get.
It goes further too because any given application is unlikely to have a workload that even distributes over all disks. In a classical JBOD, you’d need a working set of data that is greater than the size of the individual disks (which is highly unlikely) or lots of fragmentation (you really don’t want that). This means the perf that you can actually rely on getting in a JBOD is the perf of the slowest disk, regardless of how many disks there are.

Perf of slowest disk * number of disks > Perf of slowest disk.

QED.

You also assume that disk speeds are somehow vastly different whereas in reality, most modern hard drives perform very similarly.
Also nobody in their right mind would design a system that groups together disks with vastly different performance characteristics when performance is of any importance.

Personally I went with an ITX build where I run everything in a Debian KVM/qemu host, including my fedora workstation as a vm with vfio passthrough of a usb controller and the dgpu. It was a lot of fun setting it up, but nothing I’d recommend for someone needing advice for their first homelab.

I feel like that has more to do with the complexity of solving your use-case in software rather than anything to do with the hardware. It’d be just as hard on a pre-built NAS as on a DIY build; though perhaps even worse on the pre-built due to shitty OS software.

Your currently stated requirements would be fulfilled by anything with a general-purpose CPU made in the last decade and 2-4GB RAM. You could use almost literally anything that looks like a computer and isn’t ancient.

You’re going to need to go into more detail to get any advice worth following here.

What home servers differ most in is storage capacity, compute power and of course cost.

• Do you plan on running any services that require significant compute power?
• How much storage do you need?
• How much do you want it to cost to purchase?
• How much do you want it to cost to running?

Most home server services aren’t very heavy. I have like 8 of them running on my home server and it idles with next to no CPU utilisation.

For me, I can only see myself needing ~dozens of TiB and don’t forsee needing any services that require significant compute.

My home server is an 4 core 2.2GHz Intel J4105 single-board computer (mATX) in a super cheap small PC tower case that has space for a handful of hard drives. I’d estimate something on this order is more than enough for 90% of people’s home server needs. Unless you have specific needs where you know it’ll need significant compute power, it’s likely enough for you too.

It needs about 10-20W at idle which is about 30-60€ per year in energy costs.

I’ve already seen pre-built NAS with fancy hot-swap bays recommended here (without even asking what you even need of it, great). I think those are generally a waste of money because you easily can build a low-power PC for super cheap yourself and you don’t need to swap drives all that often in practice. The 1-2 times per decade where you actually need to do anything to your hard drives, you can open a panel, unplug two cables and unscrew 4 screws; it’s not that hard.

Someone will likely also recommend buying some old server but those are loud and draw so much power that you could buy multiple low power PCs every year for the electricity cost alone. Oh and did I mention they’re loud?