On 29/7/19 10:04 PM, Swâmi Petaramesh wrote:
On 7/29/19 3:58 PM, Lionel Bouton wrote:
For example I suspected that your SSD is a SATA one and I remember
data corruption bugs where the root cause was wrong assumptions made
between the filesystem layer and the io scheduler. As NVMe devices
triggered major io scheduler rework it seemed worthwhile to mention
that my system might differ from yours on this.
So, I've had the issue of 23 FSes so far :
- BTRFS FS on LVM over LUKS on a SATA SSD.
- BTRFS FS directly over LUKS on an USB-3 mechanical HD.
(All this having been perfectly stable until upgrade to 5.2 kernel...)
What kind of btrfs chunk profiles were used here (I guess its either
single or dup)?
