著者: Rainer Weikusat 日付: To: dng 題目: Re: [DNG] running with separate / and /usr
Steve Litt <slitt@???> writes: > tito via Dng said on Tue, 17 Jan 2023 08:14:58 +0100
>
>
>>Hi,
>>
>>I always used WD RE 500 GB SATA HDs in my Nas boxes and they
>>lasted years, some of them almost decades and are still happily
>>spinning, just by scrubbing the raids daily, using smartd and good
>>right-sized PSUs.
>
> Let me ask this: Has anybody here had an SSD or NVMe go bad or become
> read-only?
I've had two professional-grade and so far essentially unused SSDs in
important customer installations go belly-up out-of-the blue. Literally,
one second everything fine, next second total failure. As opposed to
this, disks fail gradually and someone who knows the signs and pays
attention to them can usually decomission them before actual data loss
happens (the reason I lost parts of the data on my dead disk was that I
hadn't been paying attention).
I've also once read through the programming documentation of a
NAND-flash. The procedure for writing to it was
1. Try writing the data.
2. Try reading it back.
3. If this didn't work, goto 1.
Forever.
Lastly, I've recently read the manufacturer's (HP) disclaimer on the box
of a USB stick. It's specifically disclaimed that data written to it can
ever be read back again and users are remained that they should always
have a backup on another medium.