Do you make local backups of your blog?

Just this evening, the server that this blog is on had a weird hardware failure. It just froze and the blog was inaccessible. The rest that followed was some sort of reality check and yet another lesson in making backups of backups.

I sent a request to the data center techs to do a hard reboot from their end. After the first reboot, the server would not respond as if it was hanging or something. A second hard reboot made it respond and I was able to login to SSH and restarted all services (Apache, mySQL, Bind).

However, the websites are still not running despite the fact that apache and mysql were fine. I was watching the processes and there were no instances of any HTTP which I initially thought was broken or the configurations went bonkers.

After an hour of checking, viewing config files and restarting services, the problem still remained. While looking at the list of accounts, I discovered some of the accounts had zero disk usage as though they don’t have anything stored in them. I realized several of the domain accounts were on the secondary drive and that drive is practically missing or was not loaded by the system.

I already expected the worst while praying for the best — the server could have suffered an irrecoverable hard drive failure. I immediately checked for the internal backups and found out that not all accounts had the latest copy. This blog specifically did not have a complete backup. Apparently, the huge account size (2.6GB) prevented the automated backup service to complete the daily backups. Same thing with the weekly backup.

I still have the monthly backups and I could start from there but that would mean that I’ll lose several weeks of entries or at most about a month.

I ran some tests and fixes on the missing drive. Luckily, another graceful reboot was all I needed for the system to detect the missing drive and load it. All that checking and diagnosis as well as coordinating with the DC engineers took me over 2 hours to finally resolve. This was fairly quick compared to other problems before where it’s usually taken some 4 to 6 hours (and there was a time about 48 hours due to a drive replacement and an OS reload).

It reminded me that despite the multiple automated backups, one should also consider regular local backups.

For WordPress blogs, this would mean downloading the Themes and Plugins folder, the Uploads inside WP-Content and a full DB backup (there’s a plugin for this one) and download. Doing this once every week-end should do as well as downloading your daily posts using feedreader clients such as FeedDemon might spare you some headache when the time comes your backups fail on you.