[solved] What is best disaster recovery backup method?

[bt101]

I understand that a simple power outage can destroy the Archlinux installation.
What is the best method of backup for disaster recovery?

I assume you can put the USB drive in another machine and do a bit copy (dd).
Will that work if you restore it back to the same drive?
Will that work is you restore it back to a different drive (different size)?

If you do some sort of file backup, can that be restored and made bootable?

I don't have a good understanding of what is on the USB drive. I know that on a "regular" PC linux install, you can restore "files" back to the hard drive but you need to do a little more to make it bootable (so the machine can locate the physical spot on the disk to boot). Is it the same thing with the ARM linux installed on the USB drive, or does the plug device boot from internal flash first and then just look for "files" (not physical location) on the USB drive? Of course, this would make a difference on the type of backup restore that you can do. If the plug device doesn't care where things are stored on the USB drive, it would allow a simple file restore to recover.

[slycat]

I've kept all the linux files on one partition and all the other goodies like media on a storage partition. My linux part is <5gb so my method is to backup my linux part to a tarball before I run "pacman -Syu", that way should anything go awry I can un-tar the contents back into the partition. Here's a quick script I made to save time typing it all each week:
$this->bbcode_second_pass_code('', '#!/bin/sh
#
if [ $1 ]; then
# various /etc/rc.d/* stop for samba/transmission/anything non-system running (for peace of mind)
cd /media/Harddrive/Backups
tar -cvpzf POGO$1.tgz --exclude=/media --exclude=/proc --exclude=/lost+found --exclude=/sys --exclude=/dev --exclude=/mnt /
# /etc/rc.d/* start all the same programs
else
echo Usage: pogobackup {MMDDYYYY}
fi')

I just call it (pogobackup 08152011) and normally gives me a ~530mb tgz or smaller if you run pacman -Scc beforehand and once it's done I update. I occasionally visit the folder and delete backups over a month old. I have once had to reinstall so I connected the drive to a working linux box and extracted the files to the drive and made sure to remake the excluded folders (media, mnt, proc, ...). I never had to do anything special afterwards making me believe the plug boots from /dev/sda1 specifically and not judging on UUIDs (proven if you ever had two drives connected at boot up where some people have problems when it finds the wrong drive first). This is the same method I have been using on my other linux box and has never failed me. It may not be the prettiest method but it works for me
If you're fearful of hardware problems, just get a surge-protector or UPS.
Hope that clears stuff up

[kmihelich]

Power outages don't destroy anything, usually. The most you'll likely need to do is fsck the root partition on another computer if you're using ext2. If you're using ext3/4 as a rootfs (depends on plug type) then it will boot right back up.

[chadr]

intermittent power outages caused all sorts of havok on my old late 90s pc based server, but with a ups it would run forever before a hard drive or power supply died. It was one of those inexpensive $50 apc units. Got two of them on craigslist for $20.

[bt101]

$this->bbcode_second_pass_quote('slycat', 'I') never had to do anything special afterwards making me believe the plug boots from /dev/sda1 specifically and not judging on UUIDs

[WarheadsSE]

edit first post, change Subject. bang.

[bt101]

$this->bbcode_second_pass_quote('WarheadsSE', 'e')dit first post, change Subject. bang.