Random problems with filesystem corruption.

[Colin Watson]

On Sun, Dec 15, 2019 at 06:44:29PM +0000, Paul Groves wrote:
> So I have my own personal web server running Ubuntu 18.04.1 LTS, it has two
> partitions on a 160GB HDD, sda1 is the grub partition and sda2 is the root
> partition.
> 
> A few months ago I had a bad superblock error and it wouldn't boot.
> 
> I booted to the installer USB and run fsck -y /dev/sda2 and this fixed the
> problem.
> 
> Now today, I noticed mysql wouldn't start and that the root partition was
> read only....
> 
> So I run fsck as before and bam! the server is working again.
> 
> I have tested the hard disk using SMART with the long and short tests and
> cannot find any errors.

SMART isn't necessarily perfect, and it's not unheard of for some
failures not to show up there.  (One might have thought that the long
test would scan the whole disk surface and catch that, but I'm not sure
how guaranteed it is that it will scan the whole disk surface, and even
then if there's an error that only happens on some percentage of
reads/writes to a particular sector then it could still be missed.)

Unless you routinely have power failures or other events that cause the
system not to shut down cleanly, I'd work on the assumption that it's
time to replace the drive; it's small enough that doing so hopefully
won't be prohibitively expensive.

-- 
Colin Watson                                       [cjwatson at ubuntu.com]