Le 02/09/2019 à 16:44, Miroslav Skoric a écrit :
On 9/2/19 10:28 AM, Reco wrote:
fsck.ext4 -f /dev/localhost/tmp
mount -t ext4 /dev/localhost/tmp /tmp
umount /tmp
fsck.ext4 -f /dev/localhost/tmp
If the mounting succeeds, change filesystem type to ext4 for /tmp in
/etc/fstab, and do the same for /home.
New pictures ... looked to me as the second command (mount) failed.
Btw, somewhere it suggested 'dmesg | tail' so I inserted it too.
It seems that you are still running the old kernel which does not
support the inline_data feature. You won't be able to mount a filesystem
with this feature enabled until you run a kernel 3.8 or newer, or
disable the feature (but tune2fs does not seem to support disabling it).
I dit a bit of testing and it appears that
- mke2fs requires an inode size of 256 to enable inline_data
- tune2fs does not support turning inline_data on or off
So I really wonder how this feature could be turned on on your ext3
filesystem with an inode size of 128. It looks like metadata corruption.
Also I am a bit pessimistic about how a newer kernel will react.
