Using rsnapshot to backup system drive

der_Typ · 27. Februar 2015

Hi,

I wanted to setup a backup for my system (root filesystem) and I thought about using rsnapshot for the task as there is already a plugin for that. But as I installed and opened the plugin it was only possible to select "Shares" as source and target. What is the reason behind this? In my opinion it makes no sense to backup one share to another share on the same system.

I thought I could get around the manual configuration of a rsnapshot job.
Is there an approved alternative for a complete (file, not image based) system backup? The backup Plugin offers some third party imaging tools like Clonezilla, but I want to be able to restore parts of the system or restore to a smaller partition so i would like to have a file based backup.

I am using a raid1 and lvm for my system drive.

subzero79 · 27. Februar 2015

I use btrfs as OS FS. I snapshot the rootfs daily and send it incrementally to the data drives. The incremental sends only a small portion of the rootfs compared to the one day before. The daily data transfer is no more than 30mb, the whole snapshot-send-receive op takes 4-5 minutes.

der_Typ · 27. Februar 2015

I actually want to get something very similar (only difference would be lvm snapshots I guess). But I am not sure whats the best way to configure the incremental backup because all backup plugins seem to depend on OMVs share-system and I would prefer an option that could be monitored via OMV.
Could you post a little bit more details about your current configuration (programs/scripts used, maybe configs, excludes,...)?

subzero79 · 28. Februar 2015

I can. But there is no gui involved. I'll post it later.

der_Typ · 28. Februar 2015

Thank you!

I was hoping for a solution integrated in OMV but I have no problem using the shell

subzero79 · 28. Februar 2015

I followed this guide to put the FS in a subvolume.

http://blog.kourim.net/installing-debian-on-btrfs-subvolume

This assumes your OS file sytem is already btrfs. I have to do a conversion with systemrescuecd from ext4 to btrfs. Also is recommended that you use the backport kernel 3.16 from debian for daily use. Kernel 3.2 burdens like 3 and half years of bugs in btrfs development.

After the system boots the new btrfs FS, you can mount the drive (rootfs) again to another location where you gonna perform snapshot, and send operations operations, like /mnt/rootfs, so besides my rootfs fstab entry I use this one:

/dev/sda1 /mnt/rootfs btrfs noatime,nonempty,compress=lzo 0 0

This is the script that runs daily on my cron.

Bash

#!/bin/bash


cd /mnt/rootfs


DATE=$(date +%Y%m%d-%H:%M)


btrfs subvolume snapshot -r @ @_snapshot_$DATE 
sync
#fileN is the day before snapshot name
fileN=$(ls /mnt/rootfs/ | tail -n -2 | awk 'NR==1{print $1}')
btrfs send -v -p $fileN @_snapshot_$DATE  | btrfs receive /media/46e48dd3-8f8a-42f4-a503-0084368be19e/rootbackup/
sync

Alles anzeigen

a simple system rollback is just going to /mnt/rootfs and do mv @ @_old && mv @_snapshot_dateXX @ you reboot and you're back in the date for system of course, not data.

If your OS drives dies, then you can boot systemrescuecd and perform a format to the new drive and do send|receive to new btrfs drive. You need to change the rootfs mount line to the new uuid and also install a new grub

Jetzt mitmachen!