Lag, not only in Eve
As many of you already noticed, I had a couple outages in the last days. Those crashes mostly occurred in the late evening hours (20:00+ GMT) with very high memory usages which forced the server to start swapping and ending up in killing apache processes cause they weren’t responding anymore. Thankfully I just had to kill and restart the xen domain to get everything back online. But 2 crashes in a row withing 20 minutes isn’t funny.
I started live debugging slow queries, optimized indizes, optimized querys (over and over again), tried to remove the amount/duration of locking update/replace commands when updating the database in the background, assigned more memory (EveMaps was always running fine with only 2gb ram and has now 4gb available) and the last thing was now optimization the robots.txt cause google and other bots are having fun aswell on such a huge page with so much information all around New E. And yes, I found some situations when some chinese bots tried to request 300+ pages/map/etc within less then a minute.
Well even today I got a 4 hours ago, I suffered some very high load (load 30+) and swapping/iowait situations again. Well, atm I can just optimize and hope that the next evening it’s stable. During the rest of the day it’s always looking promising (load <0.3). I hope I get everything fixed and stable for the weekend, so I can get on holiday without having to worry about a potential crash.
Topic change: Upcoming event
On Thursday DOTLAN EveMaps gets officially 2 years. If there’s interest, I’ll open my Teamspeak 3 Server for some nice chatting in the evening. There will be a dedicated birthday posting with TS3 details coming up later.
Fly safe.