Lots of hardware problems on display, especially suspend and resume which is notoriously buggy (broken ACPI tables that happen to work in Windows so the hardware manufacture never noticed they were busted, etc). I recommend spending extra to get ECC ram, and running ZFS filesystems. Both can catch a number of types of errors before they corrupt your data. With those precautions I haven’t lost any data in many years.
Though one time at work we had a few thousand hard drives from a particular vendor that had an interesting firmware bug. Very, very occasionally they would write a sector with incorrect data. No individual drive did it very often, but after a few thousand full drive writes we noticed it half a dozen times. We also discovered that the garbage data was always the same across all of the drives. Crazy. Sadly we weren’t running ZFS on those systems, which would have caught the problem and corrected it from redundancy. Thankfully we were able to get a refund. Never put your trust a hard drive.
To get back on topic, I’ve always assumed that journald was reasonably robust against minor corruption, but honestly I’ve never had a reason to test it. At the end of the day no one component of the system is solely responsible for data integrity; every level of the hardware and software must cooperate to prevent corruption else there will be cracks for the data to slip through.
