blah blah blah
Occassionally after a reboot/crash of one of the servers, quotas go
a bit screwing and a user is told they are over their quota even though
they shouldn't be. The method of fixing this problem is as follows:
'su -' to the user and run 'quota -v' to verify that they are reported
over their quota and note the disk which is over quota
'df -k | grep <that_disk>' to ascertain where that disk really is
rlogin into that machine as root, cd to the user's directory on that disk
(which may not be their home, but may be scratch space or other space allocated
to them)
run 'du -s `ls -A` | sort -n', directory with the biggest number contains
the file that is causing the quota corruption
cd into that directory and go the previous step until you run into a file
copy the file into /tmp
move the offending file into the mount point's lost+found directory
change the ownership of the file in lost+found to root and change its name
to something like 'dont_delete_cos_this_file_is_on_a_suspect_sector_001'
run 'quotacheck -v <mountpoint>', this should bring the user back
under quota
move the copy out of /tmp back to where it should be and chown it to the
user concerned
go back to the first step to verify things are okay now
$Header: /home/ct/serco/ahuxley/docs/html/serco-faq/new/accounts/quotas-and-user-disks/RCS/index.html,v 1.1 1997/06/13 12:34:51 ahuxley Exp $