Slashdot Mirror
Backing Up is Hard to Do?
Joe Barr writes "NewsForge is running a story this morning on a personal hardware/software backup solution for your Linux desktop (NewsForge is owned by Slashdot's parent OSTG). The solution doesn't require a SCSI controller, or tape drive, or the ability to grok a scripting language or archiving tool to work, either. It's based on point-and-click free software. Plus it includes a dead-parrot joke by Linus Torvalds."
You're joking right Linux is teh bomb all I do is use dd because it's so damn easy on Linux!
Backing up isn't painful, restoring is.
Ha.
If you are a KDE-hater you don't get to use this cool software.
We have portable USB Hard drives now, put the data on it and keep it in a safe place. You can take it any where you need to (oh no my house is on fire! for example).
It's not difficult to drag some files onto it so why do we need a "solution" to a problem already solved?
I like muppets.
Phew, at least we got those out of the way already. Now, let's get some new clichés!
Join the anonymous, help develop the network: http://www.i2p2.de
A painless solution. After getting zlib, development packages for XFree-86, Qt 3, and KDE 3....
oh great... now everyone who post after you is going to get a -1 Redundent... I hope you enjoy screwing with people's karma...
- Your stupidity got you into this mess, why can't it get you out? -Will Rogers
dd if=/dev/hda of=/dev/hdb
how hard was that?
Adults are obsolete children. - Dr. Seuss
I use rsnapshot to automate my backups to another host. Works like a dream, providing multiple virtual point in time copies (just like similar functionality from Network Appliance, etc.).
For the more typing inclined people, create a directory and do this:
rsync -av --delete --no-whole-file /folder-to-backup/ /backupfolder
This is a test. This is a test of the emergency sig system. This has been only a test.
Phew, at least we got those out of the way already. Now, let's get some new clichés!
:)
'new clichés!' eh ??
The best way to create differential backups under Unix is with hardlinked snapshots. Easy Automated Snapshot-Style Backups with Rsync has a good explanation of how to do this. The best part is that restoring is as simple as copying a file. Each snapshot is a folder hierarchy on disk, and you can browse through any snapshot and find files you want.
/tmp using mkzftree if I need to restore something.
One small improvement over rsync (IMO) is to use mkzftree from the zisofs-tools package. It's designed to create compressed ISO filesystems which will be transparently uncompressed when mounted under Linux (and other supporting operating systems; it's a documented ISO extension). mkzftree supports an option for creating hardlinked forest (like cp -al and rsync), with the advantage that the files are compressed, thus saving space. ISO isn't quite as flexible as ext2 for things like hardlinks, so what I do is have DVD-sized disk images formatted as ext2 to store the snapshots. I burn the disk images directly to DVD; each one can hold ten or twenty compressed snapshots (of my data anyway). The disadvantage is that I can't read the files directly (because they're compressed, and the transparent decompression only works with ISO) but it's easy to decompress a file or folder to
It shouldn't be hard to make the transparent decompression code work with other filesystems than ISO, as long as they're mounted read-only. The files are just gzipped with a header block indicating they are compressed.
Yeah, that sounded like a Lain attempt at humor to me.
1. Reach over and plug in USB 120 gig drive.
2. Become root, and go to /root.
3. Type "./backup.sh".
That is a script that goes to all the directories I care about (/root, /etc, /srv/www, /usr/local/share, and my home directory), and basically does this for each drive.
cd $DIR rsync -avz --progress --delete . $MNT/$DIR
where $MNT is where the USB drive mounts.
4. Unmount the drive and unplug it.
This is quick (a few minutes) and easy, and since rsync reads the files from the last backup to figure out what needs to be copied, it should catch it if I develop a bad sector on the USB drive.
I left it out in the above, but the backup script also, before doing the rsyncs, lists my crontab into a file, so that gets backed up.
Someone taught me a cool trick to backup up all files with the highest possible compression ratio and speed: mv * /dev/null
There's Dar (as mentioned in the article) and also bacula for remote backups - go check them out if they're new to you.
AT&ROFLMAO
In Soviet Russia, old people backup Koreans.
The fan uses up all of the oxygen and you suffocate. Many people in Korea have died becuase of this, but the CIA and Fox news don't want you to know.
"...prevents the restoration of individual files or directories from an archive at present"
"I lost the first two archives I created, and I'm still not positive how I did it"
"So go ahead, laugh at me"
Alrighty then
A hard drive crash over the holidays left me scrambling to get back to a productive desktop as quickly as possible. Luckily, I had my /home partition on a separate drive, so I didn't lose precious email, stories, research, and pictures. But it did get me thinking about my lack of preparedness. Where was the back-up system I've talked about for years, but never acquired? This is the tale of how I rectified that glaring omission, and built myself a personal back-up system using inexpensive parts and free software.
/home directory on hdd. Backing up directly to CD would be too slow and too cumbersome for me, so the first thing I needed was some new hardware.
The hardware
My desktop machine includes three IDE drives and an ATAPI CD-ROM drive. I have Debian installed on hda, SUSE on hdc, and my
In the past I've researched tape drives and found that for a decent drive, I would also have to add a SCSI controller. Those two items can be pretty pricey. I opted for a less expensive configuration.
I decided to go with a removable IDE drive, connected via USB. I bought a 3.5-inch hard disk enclosure with USB 2.0 connectivity on eBay. It cost roughly $45, including shipping. With three drives to backup, I needed a large-capacity IDE drive to hold all the data. It turns out I already had one, just waiting for me to use. I raided the stash of goodies I've been hoarding to build a killer MythTV box and found a 250GB Hitachi DeskStar -- just what the doctor ordered. I got it on sale at Fry's Electronics a couple of months ago for $189.
I have the mechanical skills of a three-toed sloth, but I still managed to cobble together the drive and the enclosure, neither of which came with directions. Four screws hold the faceplate on the enclosure, and four more hold the drive in place inside. Even I was able to puzzle it out.
The most difficult part was the stiffness of the IDE cable running between the faceplate and the drive. In hindsight, I recommend connecting the power and data cables from the faceplate to the drive before screwing the drive in place inside the enclosure. I also recommend not forgetting to slide the top of the enclosure back in place before reattaching the faceplate.
I connected the USB cable to the enclosure and the PC and powered on. Using the SUSE partitioning tool, I created an ext3 filesystem and formatted it on the Hitachi drive, using the default maximum start and stop cylinders. That worked, but there was a problem. My great big 250GB drive yielded only 32GB.
One of my OSTG cohorts asked if had clipped the drive for 32GB max, but I had done no such thing. All I did was check to see how the drive was strapped out of the box. It was set to Cable Select, which was fine with me, so I left it like that. His question worried me, though, because I had never heard of a 32GB clip thingie before.
I called Hitachi support to find out what was up with that. Their tech support answered quickly. When I explained what was going on, he agreed that it sounded like it was clipped to limit its capacity. This functionality allows these big honkers to be used on old systems which simply cannot see that much space. Without it, the drive would be completely unusable on those machines.
I asked why in the world they would ship 250GB drives configured for a max of 32GB by default, and he denied that they had. He asked where I got the drive, then suggested that Fry's had "clipped" it for some reason. There are jumper settings to limit the capacity, but my drive had not been jumpered that way. Perhaps Fry's sold me a returned drive that a customer had "clipped", then returned the jumpers to their original position. We'll never know.
The tech told me how it should be jumpered for Cable Select without reducing capacity. I opened the USB enclosure, pulled out the drive, and found it was already jumpered as he described. Undaunted, I pressed on.
On the Hitachi support page for the drive, I found a downloadable tool wh
4.7GB each side. Being in cartridges, they supposedly have a much longer lifetime than DVD-RW. Plenty for weekly full backups. I don't backup the system directories like /usr/bin because they are from an install. I backup /etc/ /home /root /usr/local /var/{various}. No need for anything fancy.
Infuriate left and right
It gets hard when trying to find an affordable system to do it.
Tape drives and media are fairly expensive when you start to get into the capacities that are convienent for backing up. Using another HDD is just as expensive. CD and DVD backup is virtually out of the question for even moderate capacity drives.
LK
"Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
On my Windows machines, I run NTBackup.exe every single night, so that I have a 24 hour old backup, incase the worse happens. It's saved me plently of times.
A largely on-topic question:
I'm using an external USB drive to back up my music collection every couple of weeks. My main system drive is ext3, but my external backup drive is fat32 (for compatability with Windows). I'd like to use rsync to only copy over the new/changed music. But for whatever reason rsync always copies _everything_ over. I'm suspicious that it has something to do with the file system differences...rsync is always complaining about being unable to change permissions on the fat32 drive. Anyone gotten this to work properly?
dd if=/dev/hda3 > gzip -c9 > /nfsshare/hda3backup.gz
which is of course slow and not very flexible, but works...
In Communist China, the government shoots old people backing up Koreans and makes the family pay for the bullet.
He plugs in a USB drive, runs KDar to fill it with stuff.
/bin /usr, etc, which I then burn onto a couple of DVD9-Rs. I can run this to recreate my system.
/home.
Now, when his system borks, how does he restore? Or did he think that far ahead?
I skimmed the article, and nothing about restoring. Your backup is useless if you can't restore it.
Does he have to install and configure linux, X, and KDE just to be able to access KDar?
Forget all this jibberjabber, and emerge or apt-get or type whatever command you use to get Mondo/Mindi. Just perfect for home boxes, and most other use.
Burn yourself a bootable CD that can recreate your box, just like Norton Ghost for Linux. I have it write out the iso files and boot disk for
I run a seperate job to backup
Whats important, is to seperate system from user data when it comes to backups. This also forms my "archiving" system, since old "/home" backups stick around, so if I want to take a look at the version of foo.c I was writing 6 months ago, it's easy enough to find.
As much as I love Mondo/Mindi, it's not the be-all and end-all. AMANDA is a better choice for a corporate (more elaborate) environment. It's a PITA and not worth getting involved with for a simple user box.
I don't need no instructions to know how to rock!!!!
How often do people want to restore the whole archive vs. just a single file? Almost never. This is pretty typical of "backup" solutions and how people implement them, they test the backup but rarely test that you can restore from same.
Everything on OS X works and works well. So quit struggling to get such simple things as backup working on your primitive Linux OS and switch to the greatest OS ever conceived. OS X.
Just do an NFS or SMB mount:
/mnt/backup /mnt/drive /mnt/backup .
mount -t smbfs -o username:password \\10.0.1.111\backup
cd
tar -cvjf
(If I recall the commands correctly.) I use this all the time to make quick snapshots of my Gentoo installation before emerging some bleeding edge package.
I don't know what kind of crack I was on, but I suspect it was decaf.
i don;t know about everyone else, but isn't that one of those things that should have come up pretty early in BETA testing?
"Great backup program.. too bad it can't restore"
Lain is the main person in "Cereal experiments Lain",which is a japenese animated TV show(ANIME for short).
Backups are fun!(2)
(2)Do not eat backup
How many computers are too many?
User: I think my back-up is dead
Linux: No, no.....No, 'e's stunned!
User: STUNNED?!?
Linux: Yeah! You stunned him, just as he was wakin' up! tar's and cpio's stun easily.
User: Um...now look...now look, mate, I've definitely 'ad enough of this. My backup is definitely deceased, and when I started it 2 hours ago, you assured me that its total lack of progress was due to it bein' tired and shagged out following a prolonged mount.
Linux: Well, he's...he's, ah...probably pining for the fjords.
1. External USB drive. (Actually two, so that one can be on the shelf and one plugged in). The drive is bigger than my existing system disk, and partitioned as one big bootable partition.
/media/usb* ; do /boot /home /usr /var /mp3s $DEST
/etc/init.d/postfix stop /etc/init.d/innd stop
2. A cron job that runs 4 times a day that does
for DEST in
if [ -d $DEST/home ] ; then
rsync -aSuvrx --delete /
fi
done
If anything went wrong with the main disks, it would be pretty simple to get grub installed on the USB drive, and whip it out of the external enclosure and into the computer.
Simple, unattended, and easy to recover from. The only thing I'd add to this is that I'm thinking of doing a
before the backup and a similar start afterwards to get the system a little more quiescent.
The next Cmdr Taco duplicate will be ready soon, but subscribers can beat the rush and see it early!
I'm no MS lover, at all, but do the steps involved to get this backup system working seem unbelievably, insanely, over complicated to anyone else? And the backup software doesn't have the ability to restore individual files or folders due to a bug??? That's like a "bug" in my car that makes in not a car.
Backing up my windows machines:
1. Hook up usb One Touch drive
2. Press button on drive.
http://backuppc.sourceforge.net/
A beautiful program for doing network backups. Slick web GUI, file pooling, etc. I have all of my personal PCs (mixed Linux/WinXP) at home backing up to a Mandrake box. You obviously are limited somewhat by your network speed when doing full backups. . .
Ed.
Personally I use the same sort of setup but use rdiff-backup to do the actual backup/restoration. It's really nice because I now have nightly differential backups of my system without consuming a large amount of disk space.
Also, rdiff-backup allows for remote operations. So you can have a central server back up many desktops, with relative ease. It doesn't have a nice GUI, but then again, I'm running it all through a cron job anyhow, so who cares.
Restoration is a breeze because the most recent snapshot is just a copy of the regular files, so you can go through and copy over the data. It's a little more work to get a differential snap shot, but it's nice to have if you need it.
This sig has been temporarily disconnected or is no longer in service
Restore is also straightforward - it can be done in place, or by downloading a zip/tar file.
The Raven
KDAR, if it uses KDE's sftp handler, could be used the same way and that would be much nicer.
I could have sworn that KDE had a synching program but can't remember it's name. That too would be helpful, but I'm too lazy.
Friends don't help friends install M$ junk.
The solution doesn't require a SCSI controller, or tape drive, or the ability to grok a scripting language or archiving tool to work, either.
One worry I have about the backup solution from the article is that there's no redundancy; the author backs up three drives to one large drive. If the big drive fails at the same time one of the backed-up drives fails, he's got a problem. I almost think that if I were in the same situation, I'd switch to RAID-5 inside the box with the whole directory structure on the array (instead of broken up onto different drives the way he has it set now) and back that up to the large external drive.
Not perfect, but I'd feel a little more secure with it set up that way.
"It's a wonderful idea. But it doesn't work." -- Tad Danielewski
What I like about this is if I always have a week ago fall back if I mess something up. Or if the original drive fails I just swap the backup in less than 1 min.
And yes I also do select daily data backups (email, etc.)
Is that some kind of sense that allows you to pick out other KDE users in a crowd?
Backups onto USB removable drives?
Not if you _care_ about your data! The drivers for this stuff seem to be very betaish. Lockups, garbled writes, non-standard implementation.
It's just not worth the hassle...
Try this:
/home/joeuser/etc | afio -o -v -Z -L /home/joeuser/backups/backup.log /dev/st0
mirrordir
cron it out hourly
or
find
Is there a snapshot-capable filesystem on linux that can backup off of the snapshot, so you don't have to stop your production apps?
I mean, after all, who cares if he backs up to DVD or CD or network, or whatever. We all know linux is good at moving data. I usually backup with tar -cz to a tarball on CD, and I can restore from this from a minimum CD boot, and I don't get the idea to brag about it on slashdot either.
Choosing this or that media is a non-problem, as long as you understand the difference between an off-site and an on-site backup, and as long as you have some plan as to how you're going to restore your backup in case of total system failure, and how you're going to be able to pull individual files off of the backup (this guy doesn't, but at least he realizes that).
Now tell me how you're going to backup your system without stopping it. Of course, if it's a home system, you can stop it with less pain than a large server, but still it's a pain in many cases.
I'm writing articles for newsforge, and Linus Torvalds personally replies to my email, but alas I'm not smart enough get Linux to work for me the first time.
I've got a Linksys NSLU2 running the Unslung firmware. The Unslung firmware lets me run the NSLU2 as an rsync server.
On my Powerbook I'm running rsync, configured with rsyncbackup.
I live on the East Coast but travel to the West Coast every couple months. I've got it configured so that every night my powerbook just backs itself up to my NSLU2 sitting under the bed (with a very quiet 350GB Maxtor USB drive) no matter where I am.
I also back up my gf's Win2K and WinXP laptops as well as a FreeBSD box on the West Coast. I'm very happy with the setup.
There's a reason incremental backups have been around for two plus decades, and "update the difference between two drives with rsync" is not "incremental".
If you were going to reply and say "oh, but I only do it every X weeks", well- you'll now loose weeks of work if you loose a file/drive.
Please help metamoderate.
NewsForge is running a story this morning on a personal hardware/software backup solution for your Linux desktop
How exactly do you "backup" the actual hardware?
The biggest problem with backing up a live system is maintaining consistancy during the backup - a backup can take hours, and if the system is changing state during that time, you can have the last half of the backup being inconsistant with the first half.
/usr/a/foo and /usr/z/bar at the same time, but if /usr/a/foo gets backed up, then thirty minutes laster /usr/z/bar gets backed up, that is a thirty minute window in which a change can happen and render the backup invalid.
For example - you might have something that changes both
That is why I suggest, for any system where you cannot take it down to single user mode during the backup, you use LVM to create the filesystems, and leave a bit of extra, uncommitted space in the LVM logical group. Then, you can:
Take system down to single user mode.
Take an LVM snapshot of the volumes to be backed up.
Bring system back up to normal mode.
Mount snapshot (read only, of course).
Back up snapshot.
Release snapshot.
Even should the system change the state of the filesystem, LVM will keep the snapshot'ed state around, allowing it to be backed up with a minimum of fuss.
www.eFax.com are spammers
Lately I've been using Unison to back up to my ipod: http://www.cis.upenn.edu/~bcpierce/unison/ It's cross-platform and works on mounted file systems as well as ssh. Also if I add documents on my ipod, changes are made to the source (if I say so). All in all, it's a great little (free) tool.
I'm sure there are better ways, but here's a paraphrase of my backup.sh. Whole thing took about 25 minutes with testing.
/hd2/backups.zip /collected /hd2/backups.zip /hd2/iso /hd2/iso/date* /hd2/image_$today /hd2/iso /hd2/image_$today
today=$(date +%F)
zip -r -u
cp -v -f -u
rm -r -f
mkisofs -r -J -o
cdrecord -v -speed=16 dev=0,0,0 -data
eject cdrom1
Where "collected" are files copied to the local machine through a series of smb mounts and copied across the network as well as important files on the local system that need backup. And "/collected" and "/hd2" are on different drives.
So every morning I get dated dvd of all my important files and I have them collected on two additional hard drives.
-dameron
"A known bug in the current version (1.3.1) prevents the restoration of individual files or directories from an archive at present, but that may be fixed in the next release. " That sounds like something Veritas wrote.
Copying data to a single IDE drive and calling it "backup" is just pathetic.
He should read the Tao of Backup http://www.taobackup.com/ and be enlightend.
I have an old machine with a removeable IDE drive bay in it and a fair number of ide disks in removeable caddys. I simply backup to that disk (scripts that tar/gzip the data) and when the week is over, shut the machine down and swap in a new disk. I could probably even hot swap it, but I would rather make really sure everything got written to disk. Once nice advantage of this is I can read my backups on any machine with an IDE controller, plus when I need more space I just upgrade to bigger disks and use the smaller disks in machines elsewhere. More expensive than tape, but boatloads faster and easier to verify and restore backups. Plus I don't end up later with a big box of useless tapes when I need more backup space. A nice padded lunchbox sized carrier keeps the disk protected while I take it offsite.
I Am My Own Worst Enemy
I manage the tech/training for student publications at a university. Our server is an Apple Xserve supplied by OIT. It works great but we do not let them back it up because it is prohibitively expensive.
Our solution was to buy a couple 160GB FireWire LaCie hard drives. They have heavy-duty aluminum cases and USB2, FireWire and FireWire 800 interfaces. I use CMS Products' free BounceBack Backup Express software to automatically syncronize the files on the server to the files on the hard disks.
It works great. It mounts the server's drive and updates only the files that have changed. We have two drives and alternate backing them up so there is always one off site (my house...ha) in case the building burns down.
With just one or two boot floppies, I can back up and restore my Linux drives to either: other internal IDE drives, other parititons on the same drive, external USB1 and USB2 drives, burnable CDs, or burnable DVDs.
Heck, it is so fast and reliable, I've been known to backup the drive just before even *trying out* new software or options, and if I don't like it, I just Ghost it back to how it was.
Now, I know it isn't free, or even Linux based, but it is hard to argue with cheap, reliable, and fast backup procedures that just work all the time...
I've read the whole article. My! You'd better be a geek to have to cope with all the little worries..
Getting cheap AND working hardware on E-Bay. My mom will not do it for the sake of her computer.
32GB limitation by jumpers. Not obvious for an end-user.
Booting up *nixes from various drives in order to access the limited drive, then fiddle with partitions. I still don't dare touching my configs for more than OS at a time. Let alone various OSes on various drives.
Compiling KDart?! Compiling what? What do I have to do? "Comp..??" You have to admit, it's not for the dummy kind.
Definitely not "Backup made easy" but "Made not so expensive" since the price tag still reaches 300$ (drive + box from e-bay + screws + shots of valium to calm you down when your machine refuses to boot after all the offence you just did to it).
I bought Linux Hacks. This, Webmin and a remote machine accessible using Samba or sftp does the daily backup just fine.
There are a million easy to use backup programs for Windows, yet I hardly know anyone that backs up thier home machines. I'm all for easy-to-use GUI tools, and if this is what it takes for someone to do backups, then so much the better. But people dont generally do back ups not because it's hard to do, but because it's inconvienent or requires specific user intervention. People will spend hours downloading and burning movies, but they wont spend 5 minutes putting thier email, bookmarks, and data directories onto a CD/DVD
Any backup solution wont really be effective unless it it transparent and continues works without ANY user intervention after the intitial configuration. 'Set it; then forget it' as that guy in the informercials likes to say.
I have a very simple bash script set as a cron job that backs up ~10 important data directories to file server at 5am every morning. But that requires leaving the computers on 24/7 (which I do), but most people dont like doing that. So they either have to 1) schedule thier backups to a time when the computer is likely to be on and risk impacting whatever they have to be working on at the moment, or 2) explicity lauch thier backup program when it's convienent for them (IOW, it never gets done).
I always berate my friends for not doing backups, but before I got off my own lazy ass and put a backup strategy in place (whether that be taking half an hour to write a simple script or downloading some whiz-bang GUI tool), I probably went 6 months without backing up a single byte. Terrible, I know, but I think we've all seen first-hand the barrier to backing up is not programs, but users.
From the kdar home page, it looks like it keeps a catalogue of the files in the archive, but does not keep the catalogue with the archive.
If you loose the disk that was backed up, and the catalogue with it, is the kdar archive file useless ?
I use rsync to keep track of daily changes, and tapes to make backups. Tapes have the advatage of not showing up as a drive than can be destroyed if the system gets hacked.
"I'm not sure how stable usb in linux is" ???
/dev/sd0 device to use with fdisk?? It's been that way for at least 4 years!
why? because YOU'VE never used it?
you are suprised that usb gives you a
you've never heard of the jumper on a disk that sets a hard limit in disk size??
AND YET, YOU CONSIDER YOURSELF EXPERT ENOUGH to write a poorly informed article
I agree -- you've expertly written a poorly informed article. Even getting in all the elbow rubbing elitism including email from Linus. (Ironically telling you how ignorant you are.)
I just simply take and once a month scheduled (or if any major change is made or about to be made) and simply boot a gentoo livecd, mount the drives I want to backup then simply tar compress recursively from /, then cp it to hdb1.
/mnt/gentoo with corresponding mount points mounted. chroot /mnt/gentoo /bin/bash;env-update;source /etc/profile; grub-install #to rewrite the mbr reboot.... hasn't failed me yet.
If anything goes wrong I simply repeat the steps however I extract them to
Bottom line is it's easy to backup, save, and restore. I'm thinking about patening this method as "last known good configuration", or "system restore" seems like the thing to do since microsoft has been patening all the *nix technologies as their own lately.
With rdiff-backup, backup dozens of gigabytes effortlessly and restore as effortlessly at any point in time. Add it in a nightly cron job and you are golden !
From the description : "rdiff-backup backs up one directory to another, possibly over a network. The target directory ends up a copy of the source directory, but extra reverse diffs are stored in a special subdirectory of that target directory, so you can still recover files lost some time ago. The idea is to combine the best features of a mirror and an incremental backup. rdiff-backup also preserves subdirectories, hard links, dev files, permissions, uid/gid ownership, and modification times. Also, rdiff-backup can operate in a bandwidth efficient manner over a pipe, like rsync. Thus you can use rdiff-backup and ssh to securely back a hard drive up to a remote location, and only the differences will be transmitted. Finally, rdiff-backup is easy to use and settings have sensical defaults."
Read more !
A network harddrive is a good solution for a small office. Low Power, fast transfer, large capacities.
In case you have a seperate computer or a seperate drive one can use rsync to relativly easily create backups, its just a few lines of Shell:
/your_directory_to_backup \
/your_directory_to_backup to user@other_host:/backup/current/ and in addition to that keeps all the changes you did to that directory in a seperate 'dated' directory like /backup/2005-01-15, so you can also recover files that you deleted some days ago. Some other posters seem to have missed the '--backup' option, which is why I repost the rsync trick.
rsync -e ssh \
--delete \
--relative \
--archive \
--verbose \
--compress \
--recursive \
--exclude-from list_of_files_you_don't_wanna_backup \
--backup \
--backup-dir=/backup/`date -I` \
user@other_host:/backup/current/
This command mirrors everything in
Disadvantage is that you can't easily restore an exact old state of the directory which you backuped, however you can retrieve all the files very easily.
There are also floating some shell scripts around which add to the above rsync line some vodoo to hardlink the different dated directories, so that you have a normal browsable copy of each and every day while only wasting the space for the changes.
And there are also tools which optimize this whole thing a bit more, by compressing the changes you did to files, like http://www.nongnu.org/rdiff-backup/
However overall I found the plain rsync solution the most pratical, since it doesn't require special tools to access the repo and 'just works' the way I need it.
The best solution for your backup problems is to learn to prioritize. No, you don't need to save your pr0n collection. No, you don't need to save every .jpg anyone's ever sent to you. No, you don't need to save every bad joke e-mail you've ever received. No, you don't need to save... you don't need to save... don't save... don't need.
When I was young (early 20s) I saved everything. Then I had an HD crash. I started over and, several years later, my new HD inherited an unrecoverable problem. I started over and then went through a run of about 4 Western Digital Caviar drives which each lasted about one year.
So... what do I save now? As little as possible. Believe me, when you're going through your tree and say to yourself,"well... I might want to save that just in case..." do yourself a favor and hit DELETE. Anytime the word "I might" or "just in case" comes to mind, hit DELETE.
After you've reprogrammed yourself to have a sane set of backup priorities, you'll find that rsync and tar are more than adequate.
For corporate solutions... well, that's a whole different story. Then you're getting paid to feed someone else's legal paranoia. Just buy more drives.
fast as fast can be. you'll never catch me.
Whoah, backup there for a minute, what was the sexist remark?
...and what, exactly is the need for this? Oh, no effort... got it.
JoloK
I still think that removable media (e.g. tape) is the most effective form of backup.
Under Linux, a tape drive can be used effectively to backup a home network, specially when you have offsite storage (e.g. take the monthly backup to a friend or to your work).
Granted, this is only for 10 or 20 GB worth of data, but I am not even half there yet. This does not apply to guys who have a, let's say extensive, collection of movies, or have a huge set of, ahem, images.
2bits.com, Inc: Drupal, WordPress, and LAMP performance tuning.
Fear the KDar, for it is subtle and quick to anger...
I make daily differential backups (via AMANDA) to a rotating set of 12 tapes. If I accidentally delete /etc/shadow or some other important file, I have nearly two weeks to discover the problem and restore a previous version from tape. Your idea gives you, oh, until about the time that rsync discovers the missing file and dutifully nukes it from your "backup" drive.
What you're doing is certainly better than nothing, but it's not a backup solution by any definition of the term beyond "keeps zero or one copy of the file somewhere else".
Far, far better would be for your script to use dump or tar to create incremental backup files on your USB drive and to rotate them out on a regular basis.
Dewey, what part of this looks like authorities should be involved?
Well, yeah :-P
Over here at slashdot, things can still be new, AND be a cliché!
Join the anonymous, help develop the network: http://www.i2p2.de
I've got an older DLTIII drive that I picked up on eBay for a little over $100 plus tapes. I use it to backup the essential data on a weekly schedule. The capacity is only 15G/30G, so it won't work for the Gigs of movies/music. I also have the OS on a small SCSI drive that is backed up nightly to another drive of the same size. This gives a little fat-finger protection that RAID-1 doesn't offer, but still gives a completely redundant solution in case the primary dies. And, if I goof up something I just mount the second drive and restore from the night before!
Dildonna reports: I am a submissive male slave sissy maid for my wife/mistress. She has asked that I post this story to Slashdot so that I may humiliate myself. In the quest to find something to put in my mistress's beautiful body and provide her with great pleasure, I found the vibrator to end all vibrators. It's a remote controlled, multispeed, quadruple jointed vibrator called the "Squirmy Rooter 3000". My mistress was quite pleased with the promises that the unit's box had written on it. But when I used it on her (she no longer allows me orgasm and I can't fuck her because my 6 inch cock is too small to please her) whe was more than pleased. As it turns out, this vibrator is POWERED BY Windows CE Embedded 2006. There's a small serial port near the speed control and the remote turns out to be bluetooth! Not a bad deal for $69.69. Anyway, mistress is telling me to get off of the computer because she wants to try out her new vibrator in a strap-on session with me. Our neighbor is coming over for a little forced cocksucking this morning. I hate sucking cock, but mistress says it will push me deeper into submission with her. Bye. Sissy Maid Dildonna
I'm still trying to find the right combination for backup/restore on OSX.
:P) as described in this article that I wrote for MacOSXHints a couple of years ago.
/Users using /home. Ghost for Unix perhaps?
/Users. I've thought about modifying this to include an incremental on /sw, and /usr/local, but that's being complicated. If worst came to worst, I could recompile any customizations I've done, it would just take a while.
/Users came out to be ~3GB. It takes FOREVER to complete on an external firewire drive.
/Users to DVD every once in a while, and I have an external firewire drive full of MP3's that I do optical backups of from time to time and mail them off to my brother in law in LA (I'm in St. Louis). Any better suggestions on getting a bootable OS image that is easily restored? Possible something I can dump to a DVD and mail off as I'm doing now?
First off, I don't do the default OSX install. I always slice up the partition (not partition the drive, this is a *bsd type OS people!
Then on a semi-regular basis, I will (should rather) clone everything save
Carbon Copy Cloner. This would be applicable to other *nix's of making a bootable clone minus
Then all that's left to do is an incremental backup of
Now, for those Mac users that know their stuff:
CCC will clone nicely to another disk. I've tried cloning to a disk image when everything minus
WTF? Is this an issue that has been resolved in the last year? I haven't even bothered trying, but knowing that 10.4 is around the corner, and I haven't done a decent OS backup in a while bothers me. I backup
Karma: Chameleon (mostly due to the fact that you come and go).
One where I can completely restore a system by installing from CDRs that were burned with a boot sector and all the files on the system?
One flaw in any hard drive backup system: what happens if your system is cracked?
If someone gets into your system, they do an rm -r *, is your backup drive mounted?
What if they're clever and do a mount all, or find your backup.sh first?
I've seen some people take the first and last step of "inserting the USB cable" and "removing the USB cable". Is there any kind of automated system that would ease this, or is it the Hard drive equivelant of "Remove tape, insert new tape".
USB drives also suffer from problems with catastrophic failure, like a fire in your home.
I wonder if there exist any online backup systems that let you do offsite daily differential backups of your system (or critical files) that would let you download or mail you an image of your harddrive (on DVD-R) along with restore software in case anything went wrong. You could charge directly by bandwidth used. Hmm, interesting idea.
mondo will do a full image of your drives (including making images of ntfs/fat32 drives). You boot off the image you create with mondo, and you can nuke the machine and do a full restore from cd/dvd, or do a partial restore.
backuppc is perl based and works wonders on a network for daily backups. (you can backup the server backuppc is running on too!)
Rsyncing to a mounted drive is very useful, and a good way to do things on machines that are nearby. What about remote co-lo machines, though? I've been puzzling trying to find a method to do incremental backups (i.e. NOT dumping the whole system once a week/month) of a co-located system across the net. I need to preserve permissions and ownerships, and as I said, it must be incremental (using rsyinc this should be easy, at least). The hard part, though, is doing it securely. NFS mounts ... are not very secure at all. Especially way across the internet. Going over ssh is good, but how do you automate it? In order to preserve permissions, rsync needs to run as root on both ends. Ssh keys are one way to do this; but in order to have full automation, you have to have a passwordless ssh-key that grants root access on some machine, and that freaks me out. What else can you do? I suppose I could run the script by hand and put in passphrases and such, but I _will_ forget to do that on a regular basis. There's got to be some sort of solution to this!
wheres the dead parrot joke?
You can read from
I was duped. I went specifically to the article looking for the dead parrot joke. was this a scam to get me to RTFA? Shame on you, you bastards!
The rdiff-backup program is the tool I use.
It can backup a tree to another directory, saving previous revisions as you go. The backup looks like a copy of your tree, plus a meta-information directory for storing internal revision information. It's network aware, using ssh to copy things around. It does the same protocol as rsync, so bandwidth/traffic is minimized to just changes and checksums. You can set up a cron job to delete revisions older than X days.
The authors are working on a new project that does the same thing except the backup files are encrypted.
It's written in Python, so you know it's gotta be good! It's open source.
Drawbacks? No GUI. (That's not much of a drawback, is it?)
We've slashdotted NewsForge! Judging from the resubmission retries for this post, we've slashdotted Slashdot! That parrot ain't the only inert ex-regurgitator pushin' up the daisies!
--
make install -not war
I have been using KNOPPIX CDs for creating backups for a couple of years now.
(Linux/win)
I generally just use partimage, but mc browsed tar archives just fine.
IN the "early" days there was some brain damage with versions of partimage, but those are long gone.
I'm surprised HD vendors don't sell them in pairs, with auto mirroring of a nonspinning backup ready for swapping when they inevitably fail around MTBF (YMMV). Either a second drive, or even integrated as a pair of independent platter sets. They could double their prices per capacity, plus a markup for "extra backup protection", and avoid all the brand-switching anger surrounding the disk failure that's really the admin's fault, if there's no other backup. While justifying their humongous total capacity high-end drives for almost everyone.
Like everyone else, I always demand regular backup - from everyone else. I didn't get in the habit of truly regular backup myself until I got a PalmPilot, where backup is a single button-press away. Combined with the simplicity of using "synced" records, Palm's greatest advantage over other "computing" modes I've used is its nearly implicit backup, which has saved my bacon many times. If HDs included such implicit redundancy for backup, and a simple SW switch for "restore" (use the other mirror unit), we'd all be a lot better off.
--
make install -not war
My backup script is run on a cronjob asI don't run level 0 dumps because I'd rather have nodump flags work without having to specify -h. I don't want the ports tree or the 50 gigs of backups from other machines to get burned to DVD.
dump.sh:
I hereby place the above post in the public domain.
I know, everyone hates Windows. But when parents don't want to consider another option I'm stuck helping them.
My mother wants to be able to back up certain folders containing documents on her Windows laptop without doing anything. Is there something that will automatically do it in the background for her? The best thing I can think of for her as adding on a second hard drive, if there is room in the case, and using some software. But what software?
Looking for easy,
J
I only clicked on that link because it promised a Dead Parrot joke!
Consider your needs. You may not need a fancy backup solution, so don't buy an expensive external USB hard drive if your needs don't demand it. For example, here is how I backup my home computer. My costs? $70 for the DVD burner, then just the costs of blank discs.
1. Every two months I burn my music and movies to DVD+Rs. It takes 5 DVDs to hold everything. This data seldom changes--only when I buy and rip a new CD, or download some episodes of Red vs Blue--so every two months is fine. I keep one set of these backups at work.
2. Each week I tar up and gzip my home directory and burn it to a CD-R. Yes, it fits on one CD--I keep all my music and stuff on a different partition. I used to worry about losing data during the week, then I remembered: I don't have much to lose, except my latest journal entries, browser cache/history, and Gnucash entries. I can afford to lose a week's worth of data, because it's easy to recreate. So backing up more than once a week is overkill.
I have an IMAP email account, otherwise I would need to back up my emails every day.
3. I don't backup my OS. If Mandrake gets borked and won't boot, I just reinstall. It doesn't take long, and with urpmi it's a snap to reinstall my favorite applications. (I have a local copy of PLF so installing my favorite *unsupported* packages is fast too.) Then I just copy over the latest backup of my home directory and I'm back in business.
The last time I had to restore everything it took just a few hours total.
Use iBackup, http://freshmeat.net/projects/ibackup/
Then save to media as needed. Run iBackup using cron or fcron, http://freshmeat.net/projects/fcron/
as you wish.
I first boot knoppix
/mnt/lexus/dd984g.bz2.
/mnt/lexus/dd984g.bz2.* | bzip2 -dc | dd of=/dev/hda1
knoppix screen=800x600 desktop=twm
create a backup
dd if=/dev/hda1 | bzip2 | split --bytes=650m -
created a 350 MB file, weird since my partition is 4GB (vmware default), oh lexus is an smb mount
reboot system with a knoppix CD
cat
my original vmware partition reports 4293,60 with cfdisk, after dumping to the system I get 4G exactly, it used to be a 5.5GB ext3 partition, is it possible to recover that 1.5GB for win?
btw:
# su -
# urpmi kdar
fun!
I have a 20GB iPod, but only about 12GB is used. My $HOME is about 2GB, including a bunch of digital photos, but also a bunch of documents, my email, and other stuff I'd rather not lose.
My solution is simple:
This is a very simple shell script that deletes the backup file already on the iPod, then does a 'tar czf - $HOME' and pipes it into gpg using circular encryption (that is, a passphrase.) The encrypted, compressed tarball (about 1.7GB) is written directly to the iPod. Takes about 20 minutes.
I've used this backup copy to do restores, and it's really as simple as plugging in the iPod, using gpg to descrypt the file, piping that into 'tar xvzf -' to re-create my $HOME. I can move all my stuff back to where it needs to be after that.
(For those who wonder: I always make an encrypted backup file in case my iPod is ever lost or stolen. Sure, the bad guy can probably run something to brute force the passphrase, if that's something he's interested in doing, but it's a tough passphrase. I don't worry about it so much, and it's "only" email and family photos.)
A little more googling might give you a more recent knoppix version (as of 2day it's 3.7 and 3.8 out soon), or there might be another bootable CD linux distro for PPC that contains partimage.
c id=11352968
http://debian.tu-bs.de/knoppix/powerPC/
Once downloaded boot, wait for cute GUI, open a root shell, if it's not there then just open a terminal and run
su
partimage
If you are a little more hard core, use dd; this is how I did it:
http://linux.slashdot.org/comments.pl?sid=135938&
Really, I didn't know you could actually backup disk-to-disk. Wow. And he even messes up that simple operation.
Gamingmuseum.com: Give your 3D accelerator a rest.
mkcdrec is a really neat program that packs up your whole system and makes a recovery disk. Its something any sysop should take a look at.
See the homepage here.
Tim Jones, founder of TOLIS Group will be giving a seminar on backup planning at SCALE 3x on Feb 12th in Los Angeles. Theres over 25 other talks going on as well. If you would like to check out the exhibit hall register with the promo code "free" or for a great discount on ticket use the code "newsp".
Heres some info on Tim's talk:
With Backup windows becoming smaller - because of either the amount of data being backed up or extended work hours being seen in many establishments - backup methods and procedures for reducing the requirement are becoming more important than ever before. Also, with the myriad new regulations concerning archival of data, extended and off-site storage planning becomes more important.
We will discuss many old and new procedures that can be used in most administration environments to alleviate or reduce the backup load.
script. It allows specifying directories to backup and exceptions
using the find(1) syntax. It does incremental backups. It uses
split to break up your file into CD-sized chunks for burning onto
CDs. It can use gzip/bzip2 and afio and/or GPG. But most importantly,
restore WORKS! It is as simple as
This sounds expensive.
IF you do weekly backups thats 52 120GB USB drives a year. IF you buy them for $70 each you are spending over $3600 a year to backup your data. Why don't you just get a DLT tape drive instead? It's more realiable and a lot cheaper.
He knows nothing about backing up data. My new policy: don't take backup advice from someone who has absolutely no datacenter work experience. If you suggest backing up to a hard drive you don't know what a backup is. That's not a backup, thats a dumb implementation of RAID 1.
Life has become easier since I installed a DVD-RAM drive. That makes it as simple as
/dvd /dvd
/dvd /dvd/path_to_file_to_restore original_file_path
mount
cp -r list of dirs to backup
(In reality I use a PERL script that calls a longish rsync command, which speeds up the process since it only needs to update recently modified files, but the above should work just fine if it's command-line simplicity you're after. On the other hand once you have the script written all you have to do is type the name you've given it.)
Restoring is as simple as
mount
cp
tar -lcvf /backup/$HOSTNAME-$DATE.tar /
/). The -l switch tells tar to stay on the same FS, as /backup is NFS mounted to a RAID array. Thus, I just backup the local machine, without having to specify which directories to backup, and which to skip.
/backup/whatever.tar
(I'm the type that creates one big ass partition for
Restoration, I do the lazy way:
mkdir test
cd test
tar -xvf
and then I grab the files (the RAID array usually has plenty of space).
Bored? Why not join a decent mess
mount -oro,remount
dump -Xf - -y $SUBDIRS_OF_INTEREST | ssh backupuser@wherever.offsite.com "cat > backup.$(date +%s).lzodump"
mount -orw,remount
Where X is 0 for full backup, 1 for incremental on everything since the last 0, 2 for incremental on everything since 1 or 0, etc.
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
Obviously with daily backups there will be a lot of identical blocks. That is what I use in my software, every block written to the storage is compared against the existing data, and it is only written if it does not already exist. That way I get rid of large amounts of redundancy, and I just need to create some indexes to be able to reconstruct the original files. The most tricky part is how to find the existing blocks efficiently. The second tricky part is how to eliminate redundancy in the indexes. I think I'm doing quite well in both respects.
Currently I have 404GB of tar files in my storage, however the size of the storage is only 5.5GB. But the best part is, that this blocking is not only good for storing tar files, it also works very well with the Linux file system model. With a very simple driver based on block mapping, I can mount my storage as a file system and access all the tar files directly. I can even mount those tar files using tarfs (though tarfs still seems to be lacking a few important features). There is still a few things I need to change before I'm completely satisfied with my storage format. As soon as I don't see anymore reasons to change it, I will release it under GPL.
Do you care about the security of your wireless mouse?
backup2l does a great job at figuring out which files are new or have been modified for incremental backups. Easy to configure a very lightweight.
The type of backup this person is doing is not likely to allow a complete bare-metal restore of a system.
What I prefer to do is use partimage.
I have a boot rescue disk ( Knoppix ) that has partimage on it.
Steps:
1- Boot of a Knoppix CD
2- use DD to make a backup of the MBR
3- Copy the MBR to a USB drive
4- Use sfpart ( I believe thats the name ) to create a file that can be used as an input file to sfpart that will re-create the partitions.
5- Use partimage to make backups of all my partitions. The image files for each partition is stored on a USB drive.
Now to restore should be easy enough:
1- Boot Knoppix rescue disk.
2- DD my backup of the MBR to my boot disk
3- Use sfpart and my input file to recreate my partitions.
4- Use partimage to restore each partition saved on the USB drive.
5- Remove CD and reboot...
Should have a working system....
Because RAID-1 is an exact mirror, I get a complete, bootable backup copy of my system at the time of the shutdown. Downtime is limited to the few minutes it takes to shut down and swap drives. The lengthy process of mirror rebuilding takes place while the system runs normally. And of course, RAID also protects me against random (single) hard drive failures.
This solves the full image backup problem, leaving only the more frequent partial backups you should also be doing. For this, rsync is your friend. The stuff that changes most often on my system are my IMAP folders, which I periodically (several times per day) rsync to my laptop. Besides backing up my mail server, this gives me copies I can carry around and read when I'm offline.
Tape is obsolete. It's just too slow, expensive, unreliable and small. Hard drives are so cheap, fast and capacious that there's little excuse to not run RAID on any machine that can physically hold more than one hard drive. Unfortunately, this leaves out most laptops.
I use a Certance Travan 40 GB IDE tape drive at home (I bought one for our small office Linux server shortly after because it works so well). The drive comes with Yosemite TapeWare, which has a text UI reminiscent of the Novell NetWare Text utilities.
It's easy to use, fast, allows for easy off-site backups, and restores are a snap. The media is a bit pricey, but I find it's a better solution than backing up 20 GB of data to CD (or even DVD).
Look on ebay for used Travan 20 or Travan 40 units.
So, first, we have this "Copy" system that's being called a "Backup" system. PHOOEY.
m 0204c.htm
Next, he admits that the "Backup" system can't restore files or directories. OMG!!
Anybody who adopts this system of backups better be praying to the hard drive gods and be making regular and appropriate homage.
For a REAL backup method that can stand the test of time, try this:
http://www.samag.com/documents/s=7033/sam0204c/sa
It's a system that compresses each file individually, writes them out to a temp directory, creates an iso and then writes the iso to a CD. This way, single bit errors in the compressed archive don't kill the entire thing, just a single file. This becomes more important as your backups begin to age because perfect playback of bits becomes more difficult as the media they are stored on ages.
Any system that creates a data stream of all files and then compresses it is prone to total loss of data beyond any significant error in the playback.
The order is everything.
Stream all files through the compress algorithm... Very Risky.
Compress each file individually and stream it to the archive... Very Safe.
Write everything to cheap read only media... Very smart.
Depending on the data, you can get several Gig's onto a single CD. High quality CD's should be readable for decades.
We should start referring to processes which run in the background by their correct technical name... paenguins.
Look into the scripting language called "expect".
This will tell you how to:
If you want tape solutions, there are plenty on the Net, since most Linux backup software is tape-oriented. Most of us don't do tape and I can't think of any good reason why anybody should start.
Now that I've settled this, what else do you want to talk about?
Tech Public Policy stuff
A friend of mine managed to short out is (Windows) PC whilst plugging in an external drive of some description. I don't kow if it was power on the pins, static electricty or his leaning on the power cord, but it friend the PC and the external drive!
At work, I tell pople if they must use external drives to back-up data, have two drives and use them alternately. This still doesn' give you protection against subtle corruption problems, what if you fsck up something important but don't realise until you copied over your backup device version?
Xix.
"Everything is adjustable, provided you have the right tools"
Good grief...
/
1.1s/is/his/
1.1s/kow/know/
2.2s/friend/fried
4.4s/pople/people/
"Everything is adjustable, provided you have the right tools"
Bacula it's the same, but it has a GUI and cataloging. Still testing it.
and the editor accepted the article? Dar on the command line doesn't have that problem and I know it restores individual files and directories. I like GUIs, but that kind of loss of functionality isn't worth it.
When I wrote my own how-to piece, it's a good thing I couldn't get kdar to install. Otherwise I might be in that position. Dar also does individual file compression.
At least all 3 of the backup solutions in my article(clone to disk, differential, compress/burn to DVD) actually work as intended.
Depending on the data, you can get several Gig's onto a single CD.
Do you mean DVD-R or are you getting 10:1 or better compression? Seriously, what if you have more than one DVD of info to back up?
I set things up to separate compression from burn on purpose. That way, I can let the compression run overnight and burn all the DVDs sequentially. Since I've got 30G+ to back up, it actually takes that long to do compression. Of course, if one goes that route, one had better have a lot of free drive space.
I back up complete images. That way, I don't have to reinstall everything separately after reinstalling the OS and /home .
Tech Public Policy stuff
Unless you are not too choosy.
With your "bakcup" system I can't get different versions of the same file. I get an snapshot of the full system, which is very useful of course, but does not provide me with the possibility of getting old files at a given moment in time.
You,as many people without enough experience backing up machines do, confuse high resilience with high availability of data.
They are similar and related, but not the same.
IANAL but write like a drunk one.
Ghost is OK, but partimage (http://www.partimage.org/) has more or less the same features, and it is Free Software.
Supports most filesystems: ext2fs/ext3, ReiserFS, XFS, JFS, HPFS, UFS, HFS, FAT16/32, and NTFS. Don't be afraid of the "experimental" support for NTFS, it works great if you follow the advice: defragment the filesystem first (I also disable the swap file and hibernation, to get rid of those not-movable and unnecessary files); then try to make the image, and if it suceeds, you're OK.
I've been using it for over a year, mostly for NTFS (who needs ghosting with Debian anyway?), and so far it has worked as a charm.
-f /path/to/appropriate/external/device/or/file, dump supports creating and spanning volumes.
Yes
THIS THING CAN TURN ON A DIME, MACROSSZERO STYLE ALSO FUCK BETA, ~NYORON
I have some many times have been saved by backups. But it is as important to ensure that the backups are good and you can read them.
There is no such thing as perfect. You might a have an perfect machine but mechanical and electrical have "sloppiness" built in and once you exceed the limit of the device tolerance then you are in big trouble. All manufacture's have "sloppiness" in making all equipment to be cost prohibitive.
A good solution is rotation of the backup media using a gray-code counter scheme. This means (conceptually) you keep a gray code counter of the number of backups performed, and the disk or tape you use for each backup corresponds to the bit that changes with the next increment of the counter. Example: if you have 4 DVD-RAMs in your rotation, labeled A through D, then you would rotate the media in the following sequence: ABACABADABACABAD... (no, it's not an extended version of a Genesis song). Now when you need to restore you have a wide range of ages of backups to choose from -- not just every incremental since the last full backup.
Let's say your hard disk crashes just after you've completed this much of the sequence:
ABACABADABA
Assuming your backups are daily, that means the backups in your stack have the following ages:
A: 0 days
B: 1 day
D: 3 days
C: 7 days
Thus instead of being able to go back only 3 or 4 days, as with incremental backups using the same number of media, you can go back 7 days (there is some fluctuation in this number depending on where you are in the cycle). The advantage increases exponentially with the number of media in your rotation: 8 to 15 days with 5 disks, 16 to 31 days with 6 disks, etc. The gaps get larger as you go back but that's the tradeoff; if your file was corrupted 10 days ago you might have to choose disk E, 15 days old, so you've lost 15 days of work, but at least you were able to get back to a good version at all.
Now you can use rsync to do each backup and still have the advantage of a history that is at least as long as 2^(N-2) backup cycles.
If you were going to reply and say "oh, but I only do it every X weeks", well- you'll now loose weeks of work if you loose a file/drive.
*Not* if you lose a drive, in any case, because the last backup will always contain the most recent version of the drive before the crash. The only exception would be if the disk crash happens *during* the backup (and this shouldn't matter with rsync), but then if you only have a single backup copy you are a fool.
Incidentally, if you only have a single full backup with incrementals, you run exactly that risk because your full backup might have a medium error.
But, I wanted socialized health insurance!