CIT 470 Advanced Network and System Administration: Lab #18: Snapshot Backups

Lab #18: Backups
Date: April 21, 2008

Backing up with Tar

Create a backup of the /home partition.

mkdir /var/backups
cd !$
tar cjf home_backup_`date +%F`.tar.bz2 /home

View the contents of the backup

tar tjf home_backup_`date +%F`.tar.bz2 
tar tvjf !$

Restore the backup to /tmp.

cd /tmp
tar xvjf /var/backups/home_backup_`date +%F`.tar.bz2 
ls -lR home

Remove the restore directory once you've verified it works.
```
rm -rf /tmp/home
```

Network Backups

Start your virtual machine. I'll refer to the IP address of the CentOS VM as IPADDR.

Copy your backup from the previous section to the VM.

scp -p /var/backups/home_backup_`date +%F`.tar.bz2 IPADDR:/var

Performing backups as a two step process using tar first, then copying the resulting file across the network, uses a lot of disk space on the machine being backed up. Many machines may not have the disk space for this network backup mechanism to work. Therefore, we need to perform the backup as a single step, without using a temporary file on the local disk.
```
tar cvjf - /home | ssh IPADDR "cat >/var/wbel-home.tar.bz2"
```
If we had a tape drive on the remote machine, we could use this command to write directly to tape:
```
tar cjvf - /home | ssh tape_server "cat >/dev/tape"
```
Restore the backup on the VM. Sometimes, especially when you're not restoring a current working disk on a server, you may want to restore files to the backup server.
```
cd /tmp
tar xvjf /var/wbel-home.tar.bz2
ls -lR home
```
Restore the backup stored on the backup server (VM) to the WBEL machine. Restoring the backup on the original host it came from is a common technique to recover from a failed disk or similar hardware mishap.
```
cd /tmp
ssh IPADDR "cat /var/wbel-home.tar.bz2" | tar xjvf -
ls -lR home
```
Include your command history and the results of your two restores in your lab.txt file.

LVM Snapshots

Remove logical volume created in an earlier lab if it still exists.
```
umount /iso
lvremove /dev/cit470/my_isos
vgdisplay -v
```

Create new logical volume, leaving room in VG for snapshots

lvcreate -n users --size 4G cit470          # Create a 4GB users volume
mke2fs -v -j /dev/cit470/users              # Make an ext3 filesystem
mkdir /users                                # Create a mount point
vim /etc/fstab                              # Cfg /users to mount on boot
mount /users                                # Mount /users

Create test files under /users.

cd /users
tar xvjf /var/backup/home_*
mv home/student .
rm -rf home
dd if=/dev/zero of=/users/2MB_File bs=2048 count=1024

Load the Snapshot kernel module
```
modprobe dm-snapshot
```

Create a snapshot of the users volume.

lvcreate -s -L 100M -n users_snap /dev/cit470/users

Check the logical volume status information.
```
lvs
```
Mount the snapshot and examine its contents. How do they compare to the contents of the original volume?
```
mkdir /users/snap
mount /dev/cit470/users_snap /users/snap
ls -l /users/snap
```
Remove a file from /users, simulating an accidental deletion.
```
rm /users/2MB_File
ls -l /users/2MB* /users/snap/2MB*
```
Check disk usage of both logical volumes.
```
lvs
```
Recover the file from the snapshot.
```
cp -p /users/snap/2MB_File /users
```

Remove the snapshot

umount /users/snap
lvremove /dev/cit470/users_snap

Implement a cron job to create a new snapshot of /users every hour. The snapshot should include a timestamp in its name. The cron job should also remove snapshots older than 24 hours. Include the crontab line and the script in your lab submission.