I had an issue with the root filesystem of my backuppc server (system A) which
required a restore. So I moved the pool disk to another computer (system B),
installed backuppc on that computer, booted system A off of a live CD, and
prepared for a restore.
However, system B assigned a different UID to the backuppc user than system A
did, so all of the files in the backuppc pool (now attached to system B) did
not have the proper ownership. As a result, backuppc would not start. I got
the "test hardlink failed" message.
I considered changing the UID of the backuppc user (using the usermod command)
but the UID I needed to change it to was already in use. So I did
'chown -R backuppc.backuppc /var/lib/backuppc'. This took several hours, as
there were 59 million files (mostly hard links). But it worked and I got my
restore done. (Note: I also chown'd my /etc/BackupPC files).
So I'm trying to figure out the best way to avoid this in the future. Here
are some ideas I had. I'm looking for critiques and advice.
In case of emergency, you can mount the pool with bindfs and the appropriate uid.
See the folowing example.
best regard.
C. Gouinaud
root@sauveur-desktop:~# df
Filesystem 1K-blocks Used Available Use% Mounted on
[…]
/dev/sda1 976251148 39961232 936289916 5% /var/lib/backuppc
root@sauveur-desktop:~# bindfs -u 300 -g 301 /var/lib/backuppc /toto
root@sauveur-desktop:~# ls -al /toto
total 139636
drwxr-x--- 8 300 301 124 abr 25 01:00 .
drwxr-xr-x 23 root root 4096 abr 25 17:02 ..
-rw------- 1 300 301 80 mar 10 14:51 .bash_history
-rw-r--r-- 1 300 301 85314 mar 10 07:39 config.pl
drwxr-x--- 18 300 301 134 abr 25 17:00 cpool
[…]
drwxr-x--- 2 300 301 6 abr 25 13:04 trash