This morning's downtime

Posted: Tue May 01, 2007 5:11 am

Q. Why was isohunt down?
A. We were transitioning some things with regards to how we do routing and load-balancing.

Q. What took so long?
A. Well, I'm a jackass and I wanted to try something fancy, it didn't work... and didn't work... and didn't work... so i popped us back to the 'old' config and everything came up immediately.

Q. What benefit does this serve?
A. I can now reboot the *other* router (or this one, as long as the other's up) without causing any downtime or anything more than a tiny blip before the site's back up 100%, all without anybody touching it (and an added benefit, in testing, I found that the *moment* I take a webserver out of the rotation by killing off it's lighttpd process, it's out of the rotation, just that fast, all automagical and stuffs).

Q. Does this make this Squirrel happy?
A. *chitters loudly* (that'd be a yes)

So, at the end of my day, I feel like I've accomplished something (and if you're a canadian, just fyi, there's a dedicated 100mbit port *JUST* for canadian traffic from the peers listed at http://torix.net/peers.php , so if the internet et al dies, you'll still be able to reach us).

Yay for increased reliability and more capacity!!!