BackupPC

How to backup your stuff without really crying

Bill Childers

http://wildbill.nulldevice.net/presentations/bkuppreso

Why Backup?

To protect against data loss

Hardware failure
User error
Software error
Disaster recovery

Archival purposes

Conventional Backup Methods

Tar

Tar is powerful, but can be arcane to use (old school unix mojo)
Does not scale very well from an admin perspective
Backs up everything, every time (not very efficent)
Does not work well on slow network links

Rsync

Rsync is more efficent than tar, thanks to rsync algorithm
Also does not scale very well from admin perspective

Conventional Backup Methods (Cont.)

Conventional "Backup" Programs

Programs like Amanda, Bacula, Netvault, Backup Exec, Netbackup...
Work well, most of the time
When they blow up, they blow up spectacularly
Not very storage efficient, as they backup every (changed) file on every machine

All backup methods typically require the admin to be the one to restore files to a machine

Enter BackupPC

Provides a web-based interface to manage part of the backup process and all the restoration
Leverages familiar open source tools (rsync, tar, smb) underneath
"Single-instance" backup, combined with compression = very efficient use of space

Uses hard links on the filesystem to point one file to multiple machines

Can offload all the restoration tasks/issues to the user, not the admin
Bare metal restoration (x86) is possible using Knoppix as a boot medium
Will send emails to admins and users detailing backup status (or lack of status)

How to get BackupPC

Source -- http://backuppc.sourceforge.net
Debian-based distros -- apt-get install backuppc (read the docs for passwords)

Configuration of BackupPC

BackupPC assumes you want to backup the entire system
Default paths to backup and/or exclude specified in systemwide conf file
You can create a mini-conf file for individual hosts that don't match the overall scheme
UNIX hosts backup methods:

tar over ssh
rsync over ssh (preferred)
smb

Windows hosts backup methods:

smb over admin share (C$)
rsync over ssh (via cygwin)

Configuration of BackupPC (cont'd)

/etc/backuppc/config.pl

$Conf{BackupFilesExclude} = ['/proc', '/sys', '/tmp', '/media', '/Volumes', 'dvdrip'],
$Conf{XferMethod} = 'rsync';
$Conf{RsyncArgs} = ... '--checksum-seed=32761',

Uncomment for HUGE performance increase

/var/lib/backuppc/hostname/config.pl

$Conf{RsyncClientPath} = '/opt/csw/bin/rsync';
$Conf{RsyncShareName} = ['/export/backup', '/space1/bill', '/space2/htdocs'];

Cool features for client machines

BackupPC server will "ping" its configured clients

If a client is detected on the network, it will only be backed up during "offhours"
If a client is NOT detected on the network for a certain period of time, and then is detected, a backup is immediately initiated.

So, what does BackupPC look like?

BackupPC: Host Summary

BackupPC: Restoring a file

BackupPC: Restoring a file (cont'd)

BackupPC: Restoring a file (cont'd)

BackupPC: User's Interface

Knoppix Bare Metal Restore

Fix disk (if necessary)
Boot Knoppix
Create Filesystems (if necessary)
Mount F/S Read/Write
Exchange SSH keys from BackupPC to the client
Restore the entire system from the last backup, but point it to the "mounted" path of the new filesystem
Use grub-install to repair your bootloader if running linux

Hack #53 in Knoppix Hacks by Kyle Rankin, available at your local bookstore

Demo / Q & A