We have a strange issue with our rails app deployed on a linux box with
a nginx in front and a mongrel cluster behind.
Mostly the app runs straight, fast enough - everything is fine.
But sometimes, a request needs up to 60 seconds to finish. Meanwhile
top shows no busy mongel or nginx processes. Lsof also shows no open
blocking file handels. It seems that nothing happens while the request
We played with other versions of nginx and some configuration options
but without success.
These error messages from thenginx.log maybe indicate that the problem
has nothing to do with the nginx:
2007/07/06 20:20:22 [info] 23329#0: *28 client 220.127.116.11 closed
2007/07/06 20:20:53 [error] 23332#0: *11 upstream timed out (110:
Connection timed out) while reading response header from upstream,
client: 18.104.22.168, server: www.groops.de, URL: “/foobar/”, upstream:
“http://22.214.171.124:3000/foobar/”, host: “www.groops.de”, referrer:
Maybe somebody here made similar experiences and could give us a hint.
That would be great!
ruby 1.8.6, rails 1.2.3, mongrel 1.0.1, nginx 0.5.26, Suse Linux