Lol. :)

Can confirm you have a problem. I mean, you have two services to document your stuff.

• Forgejo - git hosting
• actual budget - spending tracking mostly
• Vaultwarden
• home assistant - still configuring

If it’s behind a VPN, there’s no extra attack surface.

RAID is production ready on btrfs, the only issue is the write hole on RAID 5/6. If you don’t need RAID 5/6, you’re fine. I use RAID 1, which is 100% production ready.

multi-device support

Ah, I’ve never considered that use case. My HDD RAID 1 array is plenty fast for what I need.

But isn’t that basically what a cache drive does? It mostly caches reads, but I think it can cache writes too.

Good to know if that’s your use case, but it sounds pretty niche to me.

Surely it has a USB controller that translates SATA to USB, no? I’ve heard many of these JBOB enclosures have problems with drives falling off the bus or something in 24/7 operation.

Here’s a video from Level1Techs about USB enclosures, and at the 12 min mark or so, he talks about the USB controllers on these enclosures typically being trash. The one he recommends was $130 ($150 currently) and still has that issue with getting locked up if the connection is bad (e.g. cable gets bumped).

He does mention the USB-C controllers are getting better, so maybe those cheap emclosures are fine.

Yeah, I really don’t know what constraints OP is working under. Here are mine:

• >8TB max capacity - lots of Blu-ray rips, which grows every year (currently 3-4TB, grows by 1TB or so per year)
• RAID mirror - my media isn’t backed up, so this reduces my need to re-rip if a drive dies
• no hard requirement on speed, I only need 1-2 concurrent streams, and a single HDD is probably sufficient for that

If I was building today, I’d probably still go HDD because few mobos have >2 NVMe slots, and NVMe gets expensive at higher capacities, especially if RAID is on the table.

If my NAS was 100% backed up, I wouldn’t need RAID and I would probably use NVMe to save on space and complexity.

bcachefs

Why tho? Just use btrfs or zfs, they’re proven in production, and have a lot of good documentation.

How reliable are those though? The ones I’ve looked at have really crappy controllers.

Yeah, I’m guessing they meant Traefik. I found it too complicated and prefer Caddy, but to each their own.

Do you want a rack? I personally just use old desktop parts. It’s currently a massive ATX box (old PC), but I’ll be upgrading to ITX once I upgrade my PC again.

So my recommendation is:

1. Build around whatever you have on hand
2. Whatever is cheap and meets your current needs and can expand to meet future needs

You don’t need high specs for most things, just make sure the CPU can do transcoding if that’s what you’re planning to do eventually. Consider N100 builds or things with laptop CPUs instead of a normal ITX build, that should save you some cash now and power usage later.

Main issue is drives. If your data is modest, multiple NVMe drives could be affordable, but if you have lots of data, you’ll want HDDs, and those won’t fit. Make sure you actually want a miniPC down the line before buying, because expansion is limited.

Yeah, the case and mobo are frequently 2x the price of a mATX build. But pretty much everything else can be done on a budget (e.g. I’m using an ATX PSU in my ITX case).

If you just want read-only access to seafile, you can use the FUSE extension, but it only works in read-only mode.

For OCIS, look at the POSIX driver, which stores files in a normal directory structure.

The main “wasted” resources here is storage space and maybe a bit of RAM, actual runtime overhead is very limited. It turns out, storage and RAM are some of the cheapest resources on a machine, and you probably won’t notice the extra storage or RAM usage.

VMs are heavy, Docker containers are very light. You get most of the benefits of a VM with containers, without paying as high of a resource cost.

We should probably make a poll thread here, it could be pretty interesting to see everyone’s primary and secondary motivations.

There are a few decent options, all with some caveats:

• Seafile - wicked fast, but uses a funky disk format, so you need either a FUSE layer or the web UI/API to access anything
• OCIS/OpenCloud - default install uses a funky file format, but you can change this to POSIX if you want (experimental on OCIS, might be default now on OpenCloud?)
• others - probably work fine, but they get less blog attention

I’m playing with OCIS and I like it so far. There was some funkiness when I had things misconfigured, but now that it’s working, I like it.

It’s the easiest way to get stuff running, so I’m guessing that’s why. But it’s far from the best way.

It usually comes down to privacy and independence from big tech, but there are a ton of other reasons you might want to do it. Here are some more:

• preservation - no longer have to care if Google kills another service
• cost - over time, Jellyfin could be cheaper than a Netflix sub
• speed - copying data on your network is faster than to the internet
• hobby - DIY is fun for a lot of people

For me, it’s a mix of several of reasons.

Yeah, an IP range totally works. Figure out the subnet info and add that to a whitelist. It’s a pain, but it should keep the script kiddies at bay.

I have my smart TV access it over my local network. If you’re using a friend’s instance, you could set up a WiFi SSID that tunnels everything over your VPN.

If that’s onerous, you can make it publicly accessible, but only for whitelisted client IPs.