I'm liking this community less and less each day. When this sort of comment on this sort of article is the top comment something is wrong. For some reason I don't seem to be able to vote that comment down either.
My comment was just that; a comment. What would be worth voting down would be the original /article/ for making such a statement, rather than an observation that the statement was made.
I must agree with you, however, that I cannot figure out why my flippant observation got voted up, but this is not the fault of the comment.
Fair enough. I've seen a lot of quite innocent but flippant comments absolutely hammered before so it stands out that this was the top one. I thought your statement tended to agree with the original one but maybe that isn't the case. What ever the case I'm seeing an increasing amount of groupthink creaping into hacker news and it isn't pleasant.
I wasn't very clear, but my comment was more trying to highlight that political correctness is the enemy of potent, memorable communication. The article was not suggesting or presenting anything particularly interesting, but seemed to have been modded up because of its potency. Perhaps this is also what happened to my comment!
On your topic, however, I agree that it would be sad to see a Slashdot-esque consistency to opinion on here, but I consider it an inevitability. When it takes over, yet another site will become our refuge :)
He does know things. For Windows users, Cygwin is a decent alternative to fucking yourself, in case that's the situation on your work computer (nevermind bootability):
Create a .bat file that calls other more granular bat files (so that you can backup certain things more frequently) and then schedule a task to run nightly.
And for those of us for whom "D'oh, I deleted that!", or "Nooooo....I overwrote that file yesterday!" is more of a danger than "My house just burned down", I strongly recommend rdiff-backup:
It's short, sweet, incremental - and, because it uses diffs, not just hardlinks and new files, it doesn't use an extra 500MB each time you add 2k to the end of a 500MB dataset.
This question doesn't seem to be addressed on the rdiff-backup site, and is the main advantage of the posted solution: How much time do you waste getting a system with a dead main disk back online?
Regardless of your backup strategy, you're going to have to replace the physical disk. If the disk you are backing up to is compatible with the broken disk, you 1) already have a disk to put in your computer, and 2) you can already boot off it. So the time wasted is the absolute minimum.
Even for people using rdiff, it's probably best to have a ready-to-go backup as described in the article. His recommendation of having a bootable backup is still valid, and is not limited to the "My house just burned down" case. That's what the third disk is for :-)
It optionally encrypts data, though. I don't do that, because encrypting backups seems like playing with fire to me. What if you lose the private key? Now you have two problems.
You should use a passphrase, not a private key. Just use a passphrase that you're not going to forget, or write it down and put it in a safe deposit box. (The only people you are trying to protect your data from are people that work for Amazon. They don't have access to your house, so writing down the password isn't a security problem.)
If you want to have more than one version of your backup, rsync gained support for that somewhere in the past few years.
You want to look at the "--link-dest" option. Essentially you can keep parallel trees of backups but they use hardlinks to share unchanged files. In practice, backing up web/mail/misc servers I find each extra copy I keep takes about 5% of the first copy's space. rdiff-backup is more space efficient, but 5% isn't much and this is simpler.
If you name the backup trees after the day of the week you don't even have to worry about rotating names and purging old trees, they just overwrite.
(note: you lose the ability to just flop the drive back into the original machine. But I wouldn't do that anyway. I'd use the opportunity for a clean install.)
I've been using SpiderOak (http://spideroak.com) to backup my code daily. Upload to their server. Nice simple gui interface. And it's free for < 2GB. And it works for Linux (unlike DropBox ;)
Please tone down the excessive condescension in article headlines. It's not that the headlines are annoying (they are), but rather that it's weird that this would be a way of getting people to click on a link?
So that rsync command wouldn't delete stuff if stuff got deleted on my hard disc? Disk failures are not the only way to lose data. I'm guessing the archive flag of rsync takes care of that?
