So you may have noticed (but probably not, as you have ten thousand more interesting things to do than refresh my front page) the site was down for a few hours today (right after I blogged about how uptime would be better).
There's a good reason for that, though. Well...a reason. 'Good' may be pushing it.
The host machine for all my server VMs, and all the servers themselves, are now all running Fedora 17. Yup, because hosting production servers on a pre-release of Fedora has never caused anyone any kind of problems before, right?
They were all on F15 before; in blatant defiance of our official upgrade policies, I upgraded them via yum. The /usr move code in dracut doesn't seem to be working in F15's dracut (even the latest one in updates-testing), so I upgraded them all to F16 and then from F16 to F17.
It went remarkably smoothly, really. Four systems (the host and three servers), and only one real problem - it turns out F17's Apache has probably been completely broken as regards SSL for six weeks. The fact that no-one apparently noticed this before probably indicates I'm the only one who's a big enough idiot to run production servers on F17...
Still, fixed that up with some help from a very helpful person on #httpd, and all is well. So far, anyway. I'm sure I'll wake up tomorrow and everything will have exploded.
Did I back up the server images first, you ask? Heh. Heh, heh heh. Of course I didn't. Come on, anyone dumb enough to do this in the first place is too dumb for backups...