What explains the discrepancy in usage (82 GB
vs 13 GB
) that I see below?
Using
df
:$ df -h / Filesystem Size Used Avail Use% Mounted on /dev/sda2 96G 82G 9.9G 90% /
Using
du
:$ sudo du -cshx / 13G / 13G total
What explains the discrepancy in usage (82 GB
vs 13 GB
) that I see below?
Using df
:
$ df -h /
Filesystem Size Used Avail Use% Mounted on
/dev/sda2 96G 82G 9.9G 90% /
Using du
:
$ sudo du -cshx /
13G /
13G total
-x
option is a false friend as its purpose is to skip things. That option never gives you the complete picture.
To get a complete listing, use bind mounts and then du
, ncdu
, xdiskusage
, baobab
or whatever you wish on the bound directory without skip options:
mkdir /mnt/root
mount --bind / /mnt/root
ncdu /mnt/root
Then you might discover you have lots of stuff in /mnt/backup
(because it wasn't mounted when the backup task ran), or a giant file in /dev
(result of a dd if=/dev/zero of=/dev/sdx
when no /dev/sdx
existed and no tmpfs was mounted in /dev
).
It could also be a deleted file still used by a process, but people don't usually ask about it as it's gone after reboot. It could also be a filesystem inconsistency, but that too would be gone after reboot (if it forces fsck
in the process).
du
"loses" track of space until I restarted the machine
– Amelio Vazquez-Reina
Apr 11 '17 at 14:38
du
can't count it because it just traverses the directory.
– Barmar
May 22 '17 at 19:56
du
can't see it? – Shadur-don't-feed-the-AI Apr 11 '17 at 13:48lsof +L1
. – user2233709 Apr 11 '17 at 13:48