I have already run `bitcoind -reindex-chainstate` about 3 times this week. The node then runs fine, until this happens again.
Since that command can consistently fix it, it's your "
UTXO set" that's always been being corrupted.
Could be during the time when the cached chainstate is being written to disk.
Or after stopping Core when the SSD is moving data from its/system cache to the actual drive.
For the former, that is usually caused by a too high dbcache with high dbbatchsize, but it's mostly an issue if the drive is suddenly unplugged.
For the latter, it could happen if you unmount the drive right after stopping Core, but AFAIK, it will not finish unmounting until it's finished.
This shouldn't be an issue you didn't pay attention to the message when you unmount when sometimes it doesn't actually unmount, then proceeded to unplug the External SSD.
Of course, it's better to ask hardware-related stuffs on computer hardware forums.
Why is the data corrupted so often? Is it normal?
Obviously not normal, all nodes would get their chainstate corrupted often just like yours if it's the norm.
Start by checking the drive for "
bad sectors" because it's the easiest to test. Samsung should have its official SSD tool to do that.
CPU and Memory are quite troublesome to test, those appear to work fine when not stressed enough.
Those are more of a computer technical issue than Bitcoin-related though so search the internet for related tutorials.