Git Based Backup

Hey Everybody, sorry about the extended absence. I’ve been very busy of late! I’ve recently decided to try out the newest thing in SCM (Source Control Management), Git. Lots of people are getting on the git bandwagon, and I’m not sure if I’m totally on board yet. One of the things that appealed to me, however, is the ease of setting everything up. I’ve decided, after years, to have an actual backup solution for my website. I’ve been left without backups before, and it’s bitten me in the arse recently. As you can probably guess where this is going, yes, I’m doing git based backups of my website – database and all.

This lets me do some interesting things. First of all, I can have nightly backups, and ‘checkpoints’, points where I could easily roll back changes if I screw something up. Another is a site that only is a snapshot of the live site, whether for testing or redundancy (via branches). And lastly, it lets me have a log of recent changes that I make, and look at changes and have the ability to revert changes that I may have done while working on a script. I typically commit and push changes to a script with a detailed message when I make changes, outside of the nightly backup cycle.

At first, I was curious to see how MySQL dumps would do inside backups. So far they seem to do very well, and don’t mind being in SCM. It gets dumped to text, so stuff that doesn’t change… well doesn’t change. I back up data to one spot regularly, and then semi-frequently to another. This is how it’s laid out:

The live website is denoted with a 1, and the attached database is also pictured. This is where the general traffic and changes are, aka, the critical data. Every night, the data gets replicated to computer 2, which is completely in another country. Unfortunately it’s on a slow link, and not 100% reliable, but hey, at least it’s somewhere right? Computer #3 is my laptop, which is where I may or may not make changes to some files and testing, before committing them. On the server (computer #1), changes are pulled and merged into the repository automatically (if they’re non conflicting), before the fresh changes are committed. I’ve been running this nightly for about a half a week for my low traffic sites, and recently moved it to my blog. The scripts that run will let git automatically add new files, and delete deleted files, having a perfect snapshot of the blog. I’m going to continue testing this on more of my websites, before I make it available to the websites that I host, including a control panel type feature for managing snapshots.

It’s also worth noting that the connection between my server and the main git repository is secured by key based SSH authentication. I haven’t had any issues with this so far and has been an excellent solution.

I like to give credit, where credit is due, so I’m going to give mad props to my webhost, who processed my request for a shell within 15 minutes, installed git within 15 minutes, and fixed the git installation in 15 minutes. Their level of support has been truly remarkable. A big thanks to you guys! Check them out (A Small Orange). If you like them, and decide you want to sign up, don’t forget to use the referral link on my blog .