Hi,
if your nfs-mount are unmounted, the backups write to the root - that's right. If you after that, mount the nfs again, the nfs-storage map the local path.
But if you simply umount the nfs-mount, you see the local directory-structure - delete the files and mount the nfs-mount again.
To prevent: (i'm not realy sure, but it should work) If you mount your nfs-storage to /backup and create an directory on that, like /backup/procmoxbackup you can use this directory for backup.
If the nfs-share isn't mountet, the directory /backup/procmoxbackup don't exist and the backup should fail?!
Udo
Thank you for your help. The check for the correct mount should be executed before the backup is started. This is something i will have to create in my backup-script, but it shouldn't be too difficult.
I unmounted the NFS share, could not find any deleted files or 'broken' backups (*.tmp) and re-mounted it again.
This is our server:
proxmox005:/var/backups# df -h
Filesystem Size Used Avail Use% Mounted on
/dev/mapper/pve-root 40G 38G 0 100% /
tmpfs 16G 0 16G 0% /lib/init/rw
udev 10M 776K 9.3M 8% /dev
tmpfs 16G 0 16G 0% /dev/shm
/dev/mapper/pve-data 493G 374G 119G 76% /var/lib/vz
/dev/cciss/c0d0p1 504M 31M 448M 7% /boot
172.16.47.105:/var/lib/vz/Backup/
493G 374G 119G 76% /mnt/pve/NFSBackup
As you can see, this server is it's own NFS server.
/dev/mapper/pve-root uses a huge amount of diskspace.
Any ideas?