I have a very old 2.5" disk from Toshiba, on which there is not a single error in SMART! But there is broken data on it, which was not there just a few years ago, but they were recorded even earlier (and this was definitely checked by me)! How is this possible now without errors reading such files at the firmware level?

I have already found out many times that manufacturers defame with disk firmware even 10 years ago in terms of the reliability of reading disks and displaying errors in SMART.

I have a drive that does not have any read errors (including data relocation errors) and it reads files from the partition, but the data inside them is broken. The funny thing is that when writing, I forcibly covered these archives with a checksum and immediately re-read the archives and I know for sure that they were written whole, but now with normal full reading of these files, the checksum (which matches the copies on other disks) shows that the data is broken, which lay there for 10-12 years. These are WD drives. The Green series (which in 2010-2011 was even sold with a 5-year warranty and I can prove it, although some forum participants did not believe me =)).

This is clearly a deliberate forgery on the part of disc manufacturers.

I also recently backed up system volumes to an 8TB disk (Toshiba MG06 series) and the volume made literally 5 minutes before the next system partition turned out to be broken during the integrity check (about 30GB in size in the backup). How is this possible? Despite the fact that the memory and processor of the laptop are not overclocked and pass all tests. And as the events described above with the Green series show, even writing with the correct checksum after the subsequent reading immediately DOES NOT guarantee that in a few years the firmware will meanly not read corrupted files and will not signal you with a file reading error! That's the meanest thing.

If the sector checksum does not match the one originally written, the disk firmware MUST issue a read error on that file. But for some reason the firmware gives out such errors! And this is a fact that has already been proven by me personally, because I had the checksums of all the files written there years ago and they were checked in advance and previously read normally, especially with regard to photo albums, where multiple errors in files are immediately visible even visually.

I don't know what led to the errors, whether intentional actions or erroneous algorithms in the firmware - but you should always check the integrity of the data IMMEDIATELY, after copying the data, otherwise unpleasant surprises await you later. Use only copy algorithms with on-the-fly integrity control, such as TeraCopy programs - in no case use Windows tools or ordinary file managers to copy large amounts of data without checking the integrity.

For example, take the same popular Total Commander - it still cannot copy all three timestamps of a file, like the free FAR, and when copying in integrity control mode, for some reason it does not immediately create a file with a hash sum on the disk where we copy. The author does not seem to understand the meaning of all these operations ...

You have a very solid point there. I have always noticed a huge read/write error figure in the SMART parameters of most external hard drives that I connected to my computer. I don't even want to get started about corrupted files that were supposed to be in partitions for storage and safe from damage...

In our retail chains, without any discounts, 16TB models (alas, with helium, there are no models without it anymore) cost about $350 with a 5-year warranty.

The main problem of modern HDDs is that all manufacturers indicate an extremely low read/write resource per year - from 180 to 550Tb. Those. the mechanics of the disks have become so delicate that reliability is out of the question during normal intensive operations, which we are used as normal practics to on old HDDs 10 years ago.

Even worse, many naive buyers do not pay attention to the fact that, for example WD models, (which are most often more expensive than others, even with a 3 year warranty), have an increased level of read/write errors compared to server models - one bit to the power of 10^14 (or even 10^13), not 1 bit to the power of 10^15 for enterprise versions of HDD. Those. when copying data to such models, people need to pay increased attention to controlling the integrity of such data, which many do not worry about, even without using redundant RAID arrays.