News for hackers. Stuff that matters.
Re:Testing two 18 TB white label SATA hard drives from datab... by beala on Saturday October 11, 2025 @02:48PM (#45551592)
The official TrueNAS docs recommend against using USB drives [1]. My understanding is that between the USB controller, flaky connectors and cables, and usb-to-sata bridges of varying quality, there are just too many unknowns to guarantee a reliable experience. For example, I’ve heard that some usb-to-sata controllers will drop commands and not report SMART data. That said, there are of course many people on the internet who have thrown caution to the wind and report that it’s working fine for them.

Personally I’m in the process of building a NAS with an old 9th gen Intel i5. Many mobos support 6 SATA ports and three mirrored 20 TB pairs is enough storage for me. I’m guessing it’ll be a bit more power hungry than a ugreen/synology/etc appliance but there will also be plenty of headroom for running other services.

[1] https://www.truenas.com/docs/core/13.0/gettingstarted/coreha...

Re:Testing two 18 TB white label SATA hard drives from datab... by mannyv on Saturday October 11, 2025 @03:31PM (#45551974)
Been using like 7 external usb drives with 40-50tb total for a few years with no issues. Not raid, just backing up drive to drive. No controller or drive issues. Mix of seagate and wd 8/12/16gb.

I hate blanket recommendations like this by docs. To me, it just sounds like some guy had a problem a few times and now it's canon. It's like saying "avoid Seagate because their 3tb drives sucked." Well they did, but now they seem to be fine.

Re:Testing two 18 TB white label SATA hard drives from datab... by zettabomb on Saturday October 11, 2025 @03:54PM (#45552153)
RAID is much different. You can try it over USB, you won't have a good time. TrueNAS is primarily talking about RAID users.
Re:Testing two 18 TB white label SATA hard drives from datab... by beala on Saturday October 11, 2025 @04:56PM (#45552635)
Yes I should have specified that this advice is specific to RAID configurations in NAS applications.

If you're occasionally copying data to an external USB drive, that's totally fine. That's what they were designed for.

The issue is that they were not designed for continuous use, or much more demanding applications like rebuilding/resilvering a drive. It's during these applications that issues occur, which is a double whammy, because it can cause permanent data loss if your USB drive fails during a recovery operation. I did a little more research after posting my last comment and came across this helpful post on the TrueNAS forums going into more depth: https://forums.truenas.com/t/why-you-should-avoid-usb-attach...

Re:Testing two 18 TB white label SATA hard drives from datab... by jcalvinowens on Saturday October 11, 2025 @05:47PM (#45552986)
YMMV. I have a 4-drive 20TB mdraid10 across two different $50 USB3.0 2-drive enclosures, I've read petabytes off this array with years of uptime and absolutely zero problems. And it runs on one of those $300 off brand NUCs. The 2.5G NIC is the bottleneck on reads.
Re:Testing two 18 TB white label SATA hard drives from datab... by Yokolos on Saturday October 11, 2025 @04:56PM (#45552639)
What may work anecdotally can't necessarily be used for official recommendations for a large range of users across an unknown range of hardware configurations. If it works for you, that's fine. That isn't sufficient to make a general statement that everybody will be fine using external USB drives, particularly for RAID, especially when people will then make you responsible if something goes wrong for not making sufficiently safe recommendations. You understand that, right?
Re:Testing two 18 TB white label SATA hard drives from datab... by cerved on Saturday October 11, 2025 @06:08PM (#45553097)
Is that with ZFS or something else?

Mainly I wouldn't do it because of there's space and SATA ports it seems stupid. Hotter. Worse HW.

Can't really see much good reason to do it tbh except it's in a small hot case which is relatively easy to move around. Maybe if you do occasionally backups and you don't care about scrubbing and redundancy? Otherwise why not shuck them and throw them in a case?

Re:Testing two 18 TB white label SATA hard drives from datab... by faust201 on Sunday October 12, 2025 @12:12AM (#45555150)
you say

> 40-50tb total

> 8/12/16gb

How many drives are those?

You are kidding.

Re:Testing two 18 TB white label SATA hard drives from datab... by bluedino on Saturday October 11, 2025 @05:09PM (#45552732)
I've had the same thing from random disconnects etc from various USB hard drives and SSD's over the years.
Re:Testing two 18 TB white label SATA hard drives from datab... by bakugo on Sunday October 12, 2025 @02:24AM (#45555774)
These shucked USB adapters from WD Elements external drives are pretty reliable, from my experience. They kinda have to be, since otherwise it would affect the reputation of WD's external drives as a whole.

Obviously, direct SATA is still better if possible, but if not, these are probably the next best thing.

Re:Testing two 18 TB white label SATA hard drives from datab... by riobard on Sunday October 12, 2025 @10:59AM (#45558672)
Those pesky WD bridges usually support USB Bulk Storage only but not UASP, resulting in worse performance and higher CPU usage.

Also HDD power management is often complicated by the bridge chip sometimes intervening.

Not recommended for long-term use.