Apache processes occassionaly hanging under mod rails

Rails Deployment
1

Every couple hours I always get a few apache processes that for some
reason never quit properly. When I get enough of them my apache no
longer responds to any web requests and I need to restart it. It looks
something like this (notice the times for the first 3 apache processes):

www-data 19283 16243 0 18:24 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19426 16243 0 18:32 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19437 16243 0 20:31 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19460 16243 0 21:49 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19470 16243 0 21:52 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19483 16243 0 21:54 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19487 16243 0 21:54 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19489 16243 0 21:54 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19493 16243 0 21:54 ? 00:00:00 /usr/sbin/apache2 -k
start
www-data 19501 16243 0 21:55 ? 00:00:00 /usr/sbin/apache2 -k
start

I have killed all of them currently that haven’t been spawned in the
past 15 minutes but I think they are stuck at sending response…last
time i checked under http://localhost/server-status

the only message in my apache error logs are occasionaly:

[Mon Aug 04 21:25:13 2008] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 8 children,
there are 4 idle, and 21 total children
[Mon Aug 04 22:05:28 2008] [info] server seems busy, (you may need to
increase StartServers, or Min/MaxSpareServers), spawning 8 children,
there are 3 idle, and 19 total children

which I have just upped my server settings (i dont think the stalled
processes are related to those error message though):

Timeout 120
MaxKeepAliveRequests 200
KeepAliveTimeout 3

StartServers 50 MinSpareServers 10 MaxSpareServers 20 MaxClients 256 MaxRequestsPerChild 4000

is there any way to find out what the apache processes are stalled on?
I think it might be related to some of my new code because it just
started happening recently but I can’t reproduce it on demand.

i’m running ubuntu hardy 8.04. thanks!

2

Scott,

It is only recently that I have been playing with mod_rails, I

haven’t see such behavior yet.

Did the Rails log shows anything that could be associated to the

problem?

–
Aníbal Rojas

http://anibal.rojas.com.ve

On Aug 5, 9:11 pm, Scott K. [email protected]

3

Aníbal Rojas wrote:

Scott,

It is only recently that I have been playing with mod_rails, I

haven’t see such behavior yet.

Did the Rails log shows anything that could be associated to the

problem?

–
An�bal Rojas
http://hasmanydevelopers.com
http://rubycorner.com
http://anibal.rojas.com.ve

On Aug 5, 9:11�pm, Scott K. [email protected]

there were no errors until after i killed the processes…then i could
see which page received a sigterm before it finished loading…but i
could not reproduce the result. it seemed to happen randomly.

i just switched to nginx + thin server last night and everything is
working great so far…and using less memory!

i think it may be a bug in mod_rails.

4

I only use Apache when some of its mods or PHP is required. Otherwise
I stick to your mix: Nginx + Thin, it is a much lighter and simple
solution. Anyway it would be interesting gathering more information
regarding your problem as mod_rails looks promising.

Anibal

On Aug 8, 8:49 pm, Scott K. [email protected]