[Wiltshire] Suspect hard drive infant mortality

[David Corking]

Hi Dave,

I had a similar (but not identical) experience with a 6-month old laptop
system drive, spinning disk, starting roughly when I upgraded to Ubuntu
20.04.

I was so frustrated with frequent crashes, manual fscks, and data loss,
that I converted the ext4 partition to BTRFS. I suspected the FS driver
because I have an NTFS partition on the same disk that was running mostly
smoothly on both Windows and Linux, and because Dell's hardware endurance
tests, and the on-board SMART statistics reported zero issues.

I did *not* notice any strange noises.

According to `sudo hdparm -I /dev/sda` the disk is a TOSHIBA MQ04ABF100
Firmware Revision:  JU000D Transport:          Serial, ATA8-AST, SATA 1.0a,
SATA II Extensions, SATA Rev 2.5, SATA Rev 2.6, SATA Rev 3.0

Ever since the conversion it has been plain sailing. (Almost plain. I have
had no filesystem issues but I started to find the 8 GB RAM was too small
to run docker, few browser tabs and Slack. After one too many out-of-memory
crashes I swapped the RAM to 32 GB which fixed the crashes. This might not
be a coincidence.)

This seems particularly odd because I have used ext4 without issue for over
10 years on other machines, it is hugely popular in industry as well as
with hobbyists like me, it is based on ext3/ext2 which are also battle
tested, and if there was any recent bug I would expect to see that widely
reported. I suspect, with only weak circumstantial evidence, that there
might be some kind of _extremely rare_ incompatibility between the ext4
driver in recent kernels and my laptop's hard drive interface, or the drive
itself. (I never swapped the hardware over to investigate.)

One other possible cause for my experience is that btrfs and ntfs might be
able to survive a poor hard drive better. I haven't looked into whether
that is a realistic possibility. Also it is worth noting that the ntfs
partition wasn't a regular home or tmp directory: writes were pretty rare
compared to the ext4 partition which was a main system partition.

Therefore, I would suggest that you consider trying another filesystem on
the disk for a while, and see if your issues continue. But don't overrun
your warranty period: I think it is quite short on drives these days.

Just my point of view. I think my experience is rare, and my 'fix' entirely
misguided. I only spotted one or two similar reports on forums. Hope it
helps, David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.lug.org.uk/pipermail/wiltshire/attachments/20201121/243c4605/attachment.html>