I was called to maintain my customer's linux server today. Situation is critical, as server acts as samba server for about twenty workstations, and all work files are stored to server.

Server has adaptec raid card and raid5 configured, there is six 2TB drives giving total 10TB. Raid drive is mounted to /srv/data

But problem is, that separate system disk (320GB) is full, and samba and other services don't work correctly. But I can't find where data is lying.

Server has opensuse 11.3 OS, and I ran disk analyzer, what shows data size inside folders. It only shows about 20GB use. But df reports 100% disk usage, and I can't even modify any file, for example fstab, because os reports that disk is totally full.

I removed any files from /var/log and /tmp to free at least some megabytes, but it did not help at all. I just can't find where data is hidden. I unmounted all mounts (especially /srv/data to find, if there is data under mount points, but nothing.

I did ran fsck.ext4, and it did find some orphaned files. I did reboot and then it ran file check in boot, but that did not help either.

Can ext4 get corrupted that way, that it reports disk free size so wrong?
Or what else could be happened?

Note that server is not in any local use, it even has not display connected normally. So this can't be caused by user.

At first I thought it was the deleted open file issue but a reboot would have solved that and you say you did one.

The issue may be that you've run out of inodes rather than space. Try running "df -i" to see what your inode usage is.

Another idea is that you're using only a subset of the 320 GB drive (e.g. 1 partition of only 20 GB rather than the entire drive). It would be helpful to output the following:

df -h

fdisk -l for any disk shown in the df -h output (or vgdisplay -v for any Volume Group then fdisk -l of PVs shown in that).