It’s been more than a week already after I switch from Apache to lighttpd. Apache just can’t handled the traffic anymore on this little 256 slice. Lighttpd indeed a very fast and lightweight web server. Even with the default 16 spawned FastCGI process on the initial start, I’d still have 100+ Mb left of my 256 Mb RAM compared to 32 MB left when using Apache with MaxClients set to 10. That’s really a big win for this small vps.

For the first few days, things were doing just fine. Hits to one of the sites spiking up just after I did the switch and it’s such a joy watching the speedy new server. But the joy not lasting for long, lately I’d figured up that lighty process randomly died after a few hours serving a request. Some search on Google confirmed that it has some sort of memory leak and quite a few people experiencing the same problem as mine. So while it fast, it’s not stable. No concrete fix were found and everyone just suggesting and guessing the actual problem.

The temporary fix that I did for now is just adding a cron job that would restart the process hourly. At the same time I would try to look at nginx which look’s quite promising as an alternative to lighttpd and apache.