Oh noes, I'z opening the school email can of worms again! I got the idea for this post from the frequent outages I experienced from my school's mail server in the past month or so. There hasn't been one in the last few days, which is good, but doesn't guarantee against more downing.

So anyway, coming back to the computer after a meal break to find the email page's tab headed with "Problem Loading Page..." started to become routine. "School email server's down again? Oh well, check again in the morning." In a truly reliable system, that would be, "School email server's down?! That never happens!!!"

Actually, the latter quote is basically what I think when Gmail goes down. I haven't had an outage since early last year, and before that they were quite infrequent and quickly resolved. Lately, my school's system has been going down frequently (though not so much any more) and for long periods. Think several hours. It usually failed at night, when there were no techs around to fix it.

As far as I know, there's only one server powering the entire thing, front-end and back-end. If it croaks, nobody can do anything. (I actually never tried sending mail while the interface was down for fear it wouldn't get there.) Gmail has thousands of servers, and techs around 24/7 to fix problems. Believe it or not, I'm not really complaining, not this time. I understand that my school's tech team is only a few people, and that they can't be expected to keep a constant vigil over the servers.

I do have a suspicion, though. I started setting Firefox to auto-reload the page every two minutes using Tab Mix Plus just before the onslaught. I eventually wrote myself a Greasemonkey script (any of my classmates could find this useful, and I'm thinking about adding more features) to do the same thing, only automatically so it didn't have to be reset every time I closed and restarted the browser. It had the same two-minute interval, and the server crashes continued.

Eventually I just updated the script to reload every three minutes instead, and the outages seem to have abated. My guess is that I inadvertently either caused a mini-DoS attack on the server or tripped a rate-limiter that was temporarily blocking my IP address. I am more suspicious of the latter possibility, since it is unlikely that one request every 120 seconds would overwhelm a Web server.

If I'm right about the rate-limiter, let me just point out that Google doesn't limit Gmail requests. I have a ping going once a minute from Gmail Manager, a browser extension that checks my mail and pops up notifications. If it was the unlikely DoS alternative, I think it's just time to get a new server, or move some accounts off to another machine (if it's a shared server).

My point is that one server just doesn't cut it these days. Hardware failures and software crashes are quite common. If one server can't handle the load, add a backup. That's what I would do.

Originally published at Technobabbles
Content copyright © 2006-2012 by dgw / voyagerfan5761