Well another day, another problem to solve.

I have problem like this:

For a while everything works fine. Higher the load, higher the 
probability for this to happen.
What happends is that in some point new children are not being created 
except ONE.
so users who have active connections and have 4-5 threads up...they stay 
alive and everything works, but domains that have low usage...have their 
children killed off and end up with only one child serving whole vhost, 
so if usage grows users get lot's of "timed out" problems since one 
child can't serve them all.

Making a restart (as seeon on the graphs, at 00:00 I do logrotate), 
fixes it.

I've added some graphs too.
http://peruser.xservu.com/bugs/

This seems to be most illustrating: 
http://peruser.xservu.com/bugs/p2_jama2.png

I have this problem for about 2-3 weeks now. It happened out of nowhere. 
2 things what have changed since: 1) I compiled mod_limitipcon ( to cut 
thouse guys who try to rip off whole webpage and thus making 
couplehundred connections on the same time so all my children are full) 
and 2) we had dual gateway network (2 different subnets) and now we have 
one.

mod_limitipcon have been commented out of the conf for a week now and it 
doesn't change a thing.

Any suggestions where to look would be great.
I probably need to turn on debug mode to make things clearer, but if 
light bulb goes on for someone let me now.
_______________________________________________
Peruser mailing list
[email protected]
http://www.telana.com/mailman/listinfo/peruser

Reply via email to