We have a big rails app (openstreetmap.org) which has memory leaks.
It’s running with the script/process scripts under lighttpd. What
we’re doing is restarting (killing) the processes every couple of
hours to stop them swapping and using spawner -r to magically restart
We’d like to restart the threads after 500 requests, or something
nicer and not lose the process queue. Has anyone done this or have
any advice on the best way?
(Yes, we’d love to have the time to find the leaks but are
firefighting right now)