On 7/28/21 8:07 AM, John Mellor wrote:
Hi Chris,
I can only describe my experiences. On this Lenovo P300 machine, I
have installed btrfs on a consumer drive, 2 enterprise drives, a
matching pair of enterprise drives in btrfs RAID-1, and an ssd. All
are single-ended SATA. I have also replaced the locking short sata
cables each time, and run the half day of extended BIOS tests to
confirm that there is nothing wrong with the hardware. I have even
replaced the power supply twice, in case there was a problem getting
flakey voltage. I literally do not have any more spare hardware to
swap in. All of them have experienced corruption after some weeks of
usage. I highly doubt that the problem is the firmware.
I also have a Lenovo T500 laptop that has experienced the same problem
once, so I also doubt that it is a motherboard issue.
I have not tried to recover using that mount option. Is it documented
somewhere?
I agree that btrfs should be an advantage over ext4 in a fault
scenario. That just has not been my experience. I've found ext4 to
be an order of magnitude more reliable, albeit with a lot more seeking
happening with the same workloads. I'm back running on btrfs again at
this point, and this week it is running as expected. I'm waiting and
watching in trepidation, after about 9 or 10 total reinstalls on my
daily driver machine.
Is it possible that the uncorrected i915 faults are causing a kernel
fault to prevent btrfs from keeping things sane? To try to mitigate
that problem, I've now installed an old AMD cedar card, so that the
i915 graphics are not used.
IMHO, coming from a lot of unix and bsd experiences, flipping the root
filesystem to read-only is a very bad thing. Even if it is corrupt,
you need it to recover. It would probably be better to work like ZFS
and remount it using your nifty "usebackuproot" option, and get the
machine up so that you can figure out what was lost instead of keeping
it unusable.
Regarding btrfs corruption issues...
I recently upgraded to a new desktop PC, my first with NVMe drives (no
spinning disks). FC33 installed OK, then I'd copy files from an old PC
to the new one. Before too long, I'd start to see messages about I/O
errors and the NVMe-based file system would be remounted as read-only.
dmesg always showed I/O errors. The PC was under warranty so I shipped
it back to the manufacturer. They discovered that the NVMe drive was OK
_but_ one of the memory DIMMs was faulty. They replaced the DIMM and
shipped it back to me. I've not had a single disk I/O error since then.
Therefore, if you're experiencing file system corruption, you might want
to run a memory test to see if perhaps your system has faulty RAM.
Dave
_______________________________________________
users mailing list -- users@xxxxxxxxxxxxxxxxxxxxxxx
To unsubscribe send an email to users-leave@xxxxxxxxxxxxxxxxxxxxxxx
Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: https://lists.fedoraproject.org/archives/list/users@xxxxxxxxxxxxxxxxxxxxxxx
Do not reply to spam on the list, report it: https://pagure.io/fedora-infrastructure
