Power-consumption.

Also, the vibration produced by the 2.5" drives is less, but they’re more-sensitive to it, to begin with.

I’d not even consider spinning-platter drives, nowadays, though:

SATA SSD’s for a NAS strike me as being the sanest choice.

Samsung what are those called, Evo drives?

excellently-high MTBF, ultra-short ( compared with rotating-platters ) seek-time ( literally orders-of-magnitude quicker ), etc.

I don’t know of ANY reason to go with spinning-platters, nowadays.

( & I’m saying that as a guy stupid-enough to have not realized this in time, & who spent money on such a thing, when SSD’s really were the answer )

My experience is that USB storage sometimes breaks-connection for no discernable reason.

That if one REALLY wants to do USB storage, then put it inside the housing, and don’t use one of the external-connectors, use something you can permanently-fix, so nothing can even sneeze in its direction.

This mayn’t help you with your puzzle, but it’s bedrock and unchangeable, in my experience.

USB-storage is an unreliable joke.

ANY revision of it, that I’ve tried.

hth…

USB-storage isn’t reliable.

Period.

ANY fscking thing that bumps any connection, can break the dam link.

Then your kernel can re-label the device when it re-connects,

and you’ve got to reassemble your RAID.

just my experience.

use ANY other method you can, other than USB.

stick a SATA adaptor on there somewhere, if you can.

Get a different motherboard.

ANYthing, but not USB.

Reliability’s kinda high on my priority-list.

Try Samsung.

Nowadays I can’t imagine using SATA for anything but archival storage ( get the fastest NVMe you can for your operating-system, and be stunned by how much quicker your machine is ).

Last time I was digging into stats, the reliability-rate for Samsung devices was much higher than that of Western Digital,

and the off-brands … often are a bit of a bad-joke, for reliability ( Adata & Kingston, I’m looking at you, and will never trust such scum again ).

just my experience/opinion, is all.

SanDisk usb-keys work.

You really want to use the thing for read-only, though, if you can:

the writes it takes to kill some portion of a filesystem, vs the writes you get before corrupting things, on a USB driver, don’t line-up.

Use NVMe as your 1st-choice for storage ( future purchases, obviously ), the fastest you can get, and be stunned by how much faster the same motherboard is, with superfast OS storage…

I’d stick /home, not /usr, on the USB.

IF JBOD, && Linux, THEN yes you can know, through SMARTTOOLS, or something like that…

However, I can’t imagine how you’d get 2 separate PCIe

( presuming NVMe devices …

… no, this thing must be presuming SATA, NOT NVMe …

even in SATA, there’s no bifurcator for SATA, I don’t think:

SAS has expanders, which can take a single SAS channel & attach something like 128 SAS devices onto it,

PCIe has some kind of equivalent, and there is a PCIe card which crams loads of NVMe’s into it, out in the last year, but SATA??

Hmm… )

shrug

There was a youtube vid, testing multimeters, & there was a specific condition that produced wrong results in all the meters except Fluke, who had engineered to prevent that wrongness.

That was what decided me on trusting Fluke, in the future.

been years, no idea what channel it was on, sorry, but it should be findable for someone with patience, knowing that only the Fluke got it right, of the ones tested.

Do pay attention to the calibration-certificates, though:

Anybody paying for Fluke who ignores that their handhelds have no more than 2.5-digits of actual-accuracy, is foolish/incompetent.

( the cheap ones are sooo much worse… )

I am trying to lear basic HTML/CSS/JavaScript ( again, last learned HTML back in the 1990’s, am using “JavaScript: The Good Parts” & other books ),

& have discovered that you can have, on the same phone/tablet, Termux/Nginx running,

you have to feed /data/data/com.termux/files/usr/etc/nginx/nginx.conf the root-dir you want it to use

( which is actually in a proot-distro install, down below

/data/data/llcom.termux/files/usr/var/lib/proot-distro/installed-rootfs/ … )

… and then you can have your browser hit

http://localhost:8080/

and it’ll grab index.html.

Notice that that is http, NOT httpS.

None of the browsers I’ve tried can get the default connection to localhost, because they all default to https, & nginx isn’t serving https.

That wasted an entire fscking day, to discover.

Now learning can begin!

Thanks to this post, and the other comments in here, I’ve discovered that the ultimate ui for ai-models may well be

https://github.com/ParisNeo/lollms-webui

and on HuggingFace ( that name is aweful: to me it is the creepy-horrible FaceHugger, from the movie Alien, that I saw so many decades ago ) TheBloke has some models which are smaller

https://huggingface.co/TheBloke/

so you can choose a model that will actually-work on your hardware.

I think Llama-2 for brainstorming & CodeLlama-instruct for learning programming examples seems to be the cleanest pair, from what I’ve read, and he’s got GGUF versions with different quantizations, so you can choose what will actually-fit on your hardware.

There are other models on huggingface which seem very useful, like

• whisper-large-v3 for speech-to-text,
• whisperspeech for text-to-speech,
• sdxl-turbo for image-making ( for some copyright-free subjects to practice drawing with ), and so-on…

Some models require GPU, not all.

Damn things moved fast!

I read somewhere, years ago, that RAID6 takes about 2 cores, on a working server.

That may have been a decade ago, and hardware’s improved significantly since then.

Bet on 1 core being saturated, min, with heavy use of a RAID6 or Z2 array, I suspect…

I’d go with software raid, not hardware: with hardware RAID, a dead array, due to a dead controller-card, means you need EXACTLY the same card, possibly the same firmware-revision, to be able to recover the RAID.

With mdadm, that simply isn’t a problem: mdadm can always understand mdadm RAID’s.

_ /\ _

I’m in BTRFS, and wish I wasn’t.

Booting into a failed mdadm RAID1 is normal,

whereas booting into a failed BTRFS RAID1 requires competent manual intervention, and special parameters given to the boot-kernel.

mdadm & lvm, with a fixed version of ZFS would be my preference.

ZFS recently had a bug discovered that was silently corrupting data, and I HOPE a fix has been got in.

Lemme see if I can find something on both of these points…

https://linuxnatives.net/2015/using-raid-btrfs-recovering-broken-disks

https://www.theregister.com/2023/11/27/openzfs_2_2_0_data_corruption/

_ /\ _

EasyDNS.ca or if they also do EasyDNS.com

GoDaddy was a bunch of sleazebags, back in the day…

Go search http://slashdot.org/ for them, and see…

not only hosting lots of sleazebags, but also having tons of compromised mail machines, so their machines were, according to what I’d read there, the source of much of the world’s spam, and they wouldn’t fix things.

EasyDNS was recommended by one of the SysAdmin reporters on The Register, a few years ago.

He also recommended Linode & Vultr, back then, too.

This stuff in this comment is just my opinion, and my memory of what trustworthy people were reporting a few years ago.

_ /\ _

Replace them with SSD’s?

Your seek-time drops to near-zero,

your throughput increases to SATA ( or your motherboard’s SATA controller ) speed,

and you become absolutely-immune to such vibrations.

I cannot understand using rotating-disk drives in any notebook machine, under any circumstances, ever, for the last decade or so.

_ /\ _