httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Matthew Gray <>
Subject Re: Children Dying-- 0.7.2c Linux
Date Thu, 08 Jun 1995 22:32:00 GMT
> > Running 0.7.2c under Linux 1.2.6 seems to having children dying off for 
> > no reason.  This seems to be occurring inside of the accept.  What are 
> > other people using for mutex_lock_on and off?  This only seems to occur 
> > when there are multiple simultaneous connections.
> Hmm. Do you have multiple processors ?

Nope.  Does one need to bother with a mutex lock at all in single 
processor systems?  (I tried using a flock on the log files as a mutex 
lock, but that didn't help).

I'm going to try to figure out more specifically under what conditions 
the children are dying.  Has anyone else seen this at all?

It certainly could be linux not handling multiple simultaneous accepts on 
the same socket well.  The behavrior I'm observing looks just like what 
Ryan Watkins is seeing under Solaris.  I'll try the sa_client to 
sockaddr_in change that Rob suggested.


View raw message