I've been looking for the best Linux backup system, and also reading lots of HN comments.
Instead of putting pros and cons of every backup system I'll just list some deal-breakers which would disqualify them.
Also I would like that you, the HN community, would add more deal breakers for these or other backup systems if you know some more and at the same time, if you have data to disprove some of the deal-breakers listed here (benchmarks, info about something being true for older releases but is fixed on newer releases), please share it so that I can edit this list accordingly.
- It has a lot of management overhead and that's a problem if you don't have time for a full time backup administrator.
- It mainly comprises of using tar for backups which is pretty inflexible by modern standards.
- The enterprise web interface is OK but it's had so many bugs it's not funny.
- Backups are very slow.
- Restores are slow and painful to manage.
- I haven't found it to be great when trying to integrate with puppet / automation frameworks.
- Too complex to configure
- Stores catalog separate from backups, need to backup catalog
- Doesn't deduplicate
- Relies on clock accuracy
- Can't resume an interrupted backup
- Retention policy
- Doesn't do encryption
- File level, not block level deduplication
- Really slow for large backups (from a benchmark between obnam and attic)
- To improve performance:
lru-size=1024
upload-queue-size=512
as per: http://listmaster.pepperfish.net/pipermail/obnam-support-obnam.org/2014-June/003086.html
- Client side encryption turns off delta differencing
- Can't purge old backups
- Doesn't encrypt backups (well, there is encbup)
- Slow restore performance on large backups? (Sorry Colin aka cperciva)
- This was a really strong candidate until I read some comments on HN about the slow performance to restore large backups.
- If this has changed in a recent version or someone has benchmarks to prove or disprove it, it would be really valuable.
- Slow restore performance on large backups?
- This was also a really strong candidate until I read some comments on HN about the slow performance to restore large backups.
- If this has changed in a recent version or someone has benchmarks to prove or disprove it, it would be really valuable.
- It doesn't do encrypted backups
- No support for encryption
- Just included here because I knew someone would mention it in the comments. It's Mac OS X only. This list is for Linux server backup systems.
Also Tarsnap scores really high on encryption and deduplication but it has 3 important cons:
- Not having control of the server where your backups are stored
- Bandwith costs make your costs unpredictable
- The so called Colin-Percival-gets-hit-by-a-bus scenario
Attic has some really good comments on HN and good blog posts, doesn't have any particular deal-breaker (for now, if you have one please share with us), so for now is the most promising.
Some HN users have posted the simple script they use. The scripts usually use a combination of
mikhailian's script
FROM=/etc
TO=/var/backups
LINKTO=--link-dest=$TO/`/usr/bin/basename $FROM`.1
OPTS="-a --delete -delete-excluded"
NUMBER_OF_BACKUPS=8
find $TO -maxdepth 1 -type d -name "`basename $FROM`.[0-9]"| sort -rn| while read dir
do
this=`expr match "$dir" '.*\([0-9]\)'`;
let next=($this+1)%$NUMBER_OF_BACKUPS;
basedirname=${dir%.[0-9]}
if [ $next -eq 0 ] ; then
rm -rf $dir
else
mv $dir $basedirname.$next
fi
done
rsync $OPTS $LINKTO $FROM/ $TO/`/usr/bin/basename $FROM.0`
zx2c4's script
zx2c4@thinkpad ~ $ cat Projects/remote-backup.sh
#!/bin/sh
cd "$(readlink -f "$(dirname "$0")")"
if [ $UID -ne 0 ]; then
echo "You must be root."
exit 1
fi
umount() {
if ! /bin/umount "$1"; then
sleep 5
if ! /bin/umount "$1"; then
sleep 10
/bin/umount "$1"
fi
fi
}
unwind() {
echo "[-] ERROR: unwinding and quitting."
sleep 3
trace sync
trace umount /mnt/mybackupserver-backup
trace cryptsetup luksClose mybackupserver-backup || { sleep 5; trace cryptsetup luksClose mybackupserver-backup; }
trace iscsiadm -m node -U all
trace kill %1
exit 1
}
trace() {
echo "[+] $@"
"$@"
}
RSYNC_OPTS="-i -rlptgoXDHxv --delete-excluded --delete --progress $RSYNC_OPTS"
trap unwind INT TERM
trace modprobe libiscsi
trace modprobe scsi_transport_iscsi
trace modprobe iscsi_tcp
iscsid -f &
sleep 1
trace iscsiadm -m discovery -t st -p mybackupserver.somehost.somewere -P 1 -l
sleep 5
trace cryptsetup --key-file /etc/dmcrypt/backup-mybackupserver-key luksOpen /dev/disk/by-uuid/10a126a2-c991-49fc-89bf-8d621a73dd36 mybackupserver-backup || unwind
trace fsck -a /dev/mapper/mybackupserver-backup || unwind
trace mount -v /dev/mapper/mybackupserver-backup /mnt/mybackupserver-backup || unwind
trace rsync $RSYNC_OPTS --exclude=/usr/portage/distfiles --exclude=/home/zx2c4/.cache --exclude=/var/tmp / /mnt/mybackupserver-backup/root || unwind
trace rsync $RSYNC_OPTS /mnt/storage/Archives/ /mnt/mybackupserver-backup/archives || unwind
trace sync
trace umount /mnt/mybackupserver-backup
trace cryptsetup luksClose mybackupserver-backup
trace iscsiadm -m node -U all
trace kill %1
pwenzel suggests
rm -rf backup.3
mv backup.2 backup.3
mv backup.1 backup.2
cp -al backup.0 backup.1
rsync -a --delete source_directory/ backup.0/
I hope that it is interesting and valuable to point out that rsync.net supports, among other things:
... which is to say that an rsync.net cloud storage account is stock/standard OpenSSH, with SFTP/SCP, so borg and restic "just work" using the SFTP transport. Further, we installed and maintain an rclone binary on our platform[1][2] which means we're not just an rclone SFTP target, but you can execute rclone and run it from an rsync.net account (to go fetch from gdrive or S3 or whatever).
And, of course, any other thing that runs over SSH/SFTP/SCP (from rsync to filezilla) will work.
The entire platform is ZFS so you have protection from corruption as well as optional snapshots.[3]
[1] rclone/rclone#3254
[2] https://twitter.com/njcw/status/1144534055817502721
[3] https://www.rsync.net/platform.html