Will definitely check to see if I can work OpenSuperClone into my workflows. Haven’t had failing drives drop out like that before so I can’t speak to that scenario. I imagine if it drops out why would that software have a harder time to recover under SATA?

Mark my words. Don’t ever use SATA to USB for anything other than (temporary) access to non critical preexisting data. I swear to god if I had a dollar for every time USB has screwed me over trying to simplify working with customers’ (and my own) drives. Whenever it comes to anything more advanced than data level access USB just doesn’t seem to offer the necessary utilities. Whether this is rooted in software, hardware or both I don’t know.

All I know is that you cannot realistically use USB to for example carbon copy one drive to another. It may end up working, it may throw errors letting you know that it failed, it may only seem to have worked in the end. It’s hard for me to imagine that with all the individual devices I’ve gone through that this is somehow down to the parts and that somewhere out there would be something better that actually makes this work. It really does feel like whoever came up with the controlling circuits used for USB to SATA conversion industry-wide just didn’t do a good enough job to implement everything in a way that makes it wholly transparent from the view of the operating system.

TL;DR If you want to use SATA as intended you need SATA all the way to the motherboard.

tbh I often ask myself why eSATA fell by the wayside. USB just isn’t up to these tasks in my experience.

2500 miles sheesh. That shit’s nuclear war proof then.

Once you face the (seemingly) inevitable necessity of further hardware purchases it does become sort of tedious I must say. I used to treat my raid parity as a “backup” for way longer than I’d like to admit because I didn’t want my costs to double. With unraid I at least don’t have the same management workload that I have on my main box where I have a rolling release Arch with manually installed ZFS where the build always has to line up with the kernel version and all that jazz. Unraid is my deploy and forget box. Rsync every 24h. God bless.

Proxmox has been recommended to me before I switched my main server to Arch but once I realised that it has no direct docker support I thought I’d rather just do things myself. It really is a matter of preference. It’s kind of hard to believe that all the functionality in Proxmox can be had for absolutely free.

It’s understandable that you want to take your virtualization-capabilities to the next level but I also don’t see the appeal of containerizing unraid like many others here. I started using unraid last autumn and to me it really is about being able to mix drive sizes. It’s a backup to my main server’s ZFS pool so (fingers crossed) I don’t even really worry about drive failures on unraid. (I have double parity on ZFS and single parity on unraid.)

Anyways my point is I started out with 8 SATA slots plus an old USB-based enclosure with i set to JBOD mode and that was a pretty stupid idea. unraid couldn’t read SMART data from those USB drives. Every once in a while one of the drives would suddenly show up as having an unsupported partition layout. Couple weeks ago all 5 drives in the enclosure started showing up as unusable. So as you can imagine I dropped that enclosure and now am working solely off the 8 internal slots. I’d imagine that virtualizing unraid’s disk access might potentially yield similar issues. At least the comments of people here remind me of my own janky setup.

I recently “upgraded” one of my raspberrys SD cards to an industrial grade one. Seems to me like those are a lot slower but for that particular use case it doesnt matter to me. What matters is that the card doesn’t die. It runs noticeably cooler when lots of data is being written to it so I feel like I must be onto something there.

I used to (over a span of about 4 years now) just rely on a RaidZ2 (ZFS) pool (faulted drive replacements never gave any issues) but I recently did an expansion of the array plus OS reinstall and only now am I starting to incorporate Docker containers into my workflows. The live data is in ~ and nightly rsynced onto the new larger RaidZ2 pool but there is also data on that pool which I’ve thus far never stored anywhere else.

So my answer to the question would be an off-site unraid install which is still in the works. This really will only be that. A catastophe insurance. I probably won’t even rely on parity drives there in order to maximize space since I already have double parity on ZFS.

As far as reinstallation goes, I don’t feel like restoring ~ and running docker compose for all the services again would be too much of a hassle.