Sonnics external HD fault

I have had a Sonnics 320GB external hard-drive for the last couple of years attached to my RISC OS Rpi3. Earlier this year, I now realize, it developed a fault, which explains why I got problems with some software. Attempts to verify it ended early with report of a hardware fault. I managed to delete a copy of !NewsBase in which the fault appeared and I have been using webmail ever since. Nevertheless it still functions as a hard disk, and I am backing up about 11Gb-worth of files onto a rescue memory stick. Is there any way of repairing the fault, or should I just buy a replacement?

You could look at the SMART information by attaching it to a Pi running Linux (or something running another common OS), to find the exact nature of the failure. But once a disc of any type gives you the slightest indication it is failing, it’s best to replace straight away.

once a disc of any type gives you the slightest indication it is failing, it’s best to replace straight away.

Wise words. It will never under any circumstances get better.

and I am backing up about 11Gb-worth of files onto a rescue memory stick.

You don’t have a habitual backup system?

320GB external hard-drive for the last couple of years attached to my RISC OS Rpi3.

I’m afraid I simply don’t trust FileCore enough. Back in the olden days, fsck was a must due to the propensity for “Broken directory”.
Whatever it was must have been fixed along the way as I’ve not seen this yet in the 32 bit world (touches head), but I feel we’re only swapping one problem for a different one, in my case the supposed fragility of flash memory, but at the same time 320GB is a massive amount, especially for a filing system originally aimed at floppies that’s been patched and tweaked along the way (it’s our “FAT”, isn’t it?).

So my golden rule is quite simple. If I can’t attach it to a PC and take an image of it, it’s not suitable. That’s why I stick with an 8GB µSD and a 16GB USB key as overspill (thank you Manga ;) ).

You could look at the SMART information by attaching it to a Pi running Linux

That’ll tell you what you already know, something’s wrong.

Best suggestion is to get a replacement, copy over everything, and then destroy the faulty drive so nobody can recover anything from it.
https://www.youtube.com/watch?v=Phe3qZ_q9Fk

then destroy the faulty drive so nobody can recover anything from it.

Er, stick it in a cupboard (marked beware of the leopard) more like. The more media backups are on, and the more places they are, the better chance you have.

Back in ye olden days at Acron, I wrote a little utility, CRCTree, so that you could store the CRCs of backed-up files along with them, which you could check at restore time. Restoring a bit-rotted backup is almost as bad as losing data in the first place. (Stems from an episode with really shonky floppies we were given to use as backup media)