This morning I noticed someone had stuck a USB Drive in the back of the server. (I'll have to end that practice later). For some reason CentOS 7 mounted that disk /dev/sdc1 under /run/media//. I also noticed when I ran df that my /dev/mapper/centos-root was full. Ok, ran through / with du and noticed that the flash drive had mounted itself here and took up the remaining space somehow.

So I ran umount /run/media/<user>/<uuid> and it went away. Great. Except df still states that I'm full and the server is experiencing issues common with low disk space. Worse now, a reboot of the server left me only able to get into maintenance mode. How do I free up this space or even know what is bound to the fs?

EDIT: This is essentially another df -H doesn't agree with du problem. According to df I have only 800MB/54GB free on root and I can't boot because of it. But by using du -h --max-depth 1 /mnt/sysimage/ (where /mnt/sysimage/ is just / because I'm in recovery mode right now). It outputs something like this (can't directly copy):

264M    /mnt/sysimage/boot
0       /mnt/sysimage/dev
26G     /mnt/sysimage/home
0       /mnt/sysimage/proc
... small stuff - MB scale
1.7G    /mnt/sysimage/var

So how do those add up to fill a 54GB disk? As stated above before it crashed, this mystery USB drive had been mounted in /run/media/<user>/<uuid> and was thus under / Could this still be bound to the file system?

  • If you are describing commands or semi-hipotetical questions, please do provide the output. The question is too broad as it it. – Rui F Ribeiro Aug 3 '18 at 20:11
  • @RuiFRibeiro I updated my question with outputs, skipped them originally because I thought I new the cause and didn't want this flagged as another df doesn't match du question. – Aaron Chamberlain Aug 3 '18 at 20:29

Mounting does not directly consume space on the mounted on filesystem.

Ensure you have the root of filesystem mounted on /mnt/sysimage (and not, for example, a Btrfs sub-volume) and nothing on top (which could hide large files). You can check this by inspecting /proc/self/mountinfo.

If that doesn't reveal the cause, run the appropriate fsck tool on the filesystem.

For Btrfs, try mounting with the clear_cache option.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.