We were on the receiving end of the slashdot effect last Tuesday. The timeline went something like this:
09:59 – I finally check my email, and click up Roger’s web page. Recognizing Slashdot material when I see it, I email him back, warning him that I’m going to submit it.
10:36 – I give Roger’s server a quick review to see if anything needs patching or upgrading. I find a couple of minor security updates and install them.
11:10 – Story submitted to Slashdot
11:33 – Story accepted
18:37 – MRTG graph shows our bandwidth utilization maxed at 100%, where it stays for the next 6 hours. Fortunately, it’s after business hours so other clients are mostly unaffected by the slowdown.
00:30 – Bandwidth finally drops below 100%. Roger’s Linux/Apache server never skipped a beat, serving all requests that could get down the pipe to it. I tried hitting the box from my cable modem at home and it was really slow but working. There were a few complaints on Slashdot from people who couldn’t see the site so I’m sure some requests weren’t making it through. Several mirrors popped up pretty quickly.
07:00 – Suprisingly bandwidth usage started climbing again about 4am CST and we are maxed out again this morning. The story has a way to go before it scrolls off Slashdot’s front page.
10:30 – The story finally scrolls off the front page and hits start slowly dropping off.
Hits on Roger’s server remained much higher than normal for another two days but things are mostly back to normal now. Roger received a pile of email during the course of events and will probably be posting the more interesting items on the page over time. The most surprising thing in all of this was how many Slashdot readers didn’t understand CmdrTaco’s ^H^H^H humor.