After finding out that my MySQL backups from bH were corrupt (as well as all of my backups from all of my domains for the last few months), and after taking steps to ensure that this sort of thing doesn’t happen again, I was finally able to fix blogHarrisburg today. Due to the loss of the database backups, I won’t be able to “fix” the existing mangled posts. New posts, however, should work just fine. If you want the technical explanation, read on — if not, rest assured that your local new media source will resume normal functionality right about …NOW!
The database backups were mangled because of a fortunately-unusual and very technical, complex problem known as “I dropped my external drive off a four-foot-high table in the NOC — again — and my hard drive didn’t like it — again.” This was both a disaster and a blessing. A disaster because I’ve now lost all my documents from 2006 (the one set for which I didn’t have multiple backups). A blessing because I now have an excuse to purchase a leaner, meaner, 2.5″ 250gb external mini-drive to supplement my Macbook Pro’s somewhat tiny 80gb internal drive.
I’ve also used this as an excuse to bring back one of my old PCs from the dead as an SFTP server. Those of you who know me well or have ever my Executive House apartment know that I used to have no less than seven PCs running at any given moment. My audio server, my movie server, a Linux sandbox, a Linux desktop, an OS/2 Warp 4.0 machine, my main desktop (a then-state-of-the-art Athlon XP 2000 with a Radeon 9700 Pro), and my fileserver. Once I moved to my house, which unfortunately does not include free electricity, I cut back to just my laptop. I’ve now assembled
one of my old ultra-low-power C3-based motherboards into what basically amounts to a 1tb site backup holder. The electric draw will be noticeable, but not as big as running a standard desktop 24/7. I bought a second one of these to help mitigate the frequent power outages that slam my area. Cheap and generic, sure. But I’ve had one guarding my VOIP modem, DSL modem, and wireless router now for about six months, and it gives them about 5.5 hours of runtime (which is a big deal, as my alarm system runs through my VOIP).
Having a 24/7 SFTP server on my static IP allows me to set up automated daily MySQL backups and weekly whole-site backups. Every Monday, at some point during the wee hours of the morning, the sites will all go offline for about ten minutes while HyperVM generates a disk image of my virtual machine. I’ve capped the inbound speed to the SFTP server at a paltry 50k, so there won’t be any interference with site performance while it uploads the backup to me. Going forward, I should have several months’ worth (why do something if you don’t do it overkill?) of MySQL and whole-disk backups on hand at any given moment.
As for the website content, it turns out that when I upgraded to WordPress 2.5, it replaced the newer version of MagpieRSS that I was using with an older one. This was unbeknownst to me, of course. Since most version upgrades of WP wind up tweaking the database, it’s not possible to just move back down a version using the same database. Thus, without a working database backup, I would’ve had to install an old version of WP and started bH over from scratch. I chose to wait and see if a fix was posted. And yesterday, one was. As seems to always be the case, the solution was mind-boggingly simple and required approximately three seconds out of my life to apply. I re-installed the new version of MagpieRSS and am now sitting here at my Ikea-based desk with my fingers eagerly crossed.
Everything should be back to normal. If it isn’t, I’ll fix it tomorrow. Right now I’m heading up to S&B’s to buy new summer clothes, then I’m out to Duke’s to enjoy what’s left of this sun.