On Mon, 4 Feb 2002, Randy Arabie wrote:

> Date: Mon, 4 Feb 2002 12:01:11 -0700 (MST)
> From: Randy Arabie <rrarabie@arabie.org>
> To: "William R. Mussatto" <mussatto@csz.com>
> Cc: Dan Nelson <dnelson@allantgroup.com>,
>     Manuel Bouyer <bouyer@antioche.eu.org>, netbsd-help@netbsd.org,
>     mysql@lists.mysql.com
> Subject: Re: sysctl & process limitations
> 
> On Mon, 4 Feb 2002, William R. Mussatto wrote:
> 
> > Have you checked on the ulimit for the user that is mysql?
> 
> I've checked a lot of things.  
> 
> Which would apply:
> 
> 	kern.maxfiles
> 	proc.[mysqld_pid].rlimit.descriptors.soft
> 	proc.[shell_pid].rlimit.descriptors.soft
> 
> Obviously, the process rlimits cannot exceed the limits set in the kernel.
> 
> But, between the other two, which one takes precedence...the shell from 
> which the command was invoked or the actual mysqld process?
> 
> -- 
> 
> Cheers!
> 
> Randy
> 
Someone in the netbsd.org will have to answer since we are a linux shop 
and I'm not sure its the same.  The only reason I mentioned this is we 
ran into a problem w/apache durring the code red worm attacks (its not 
vulnerable, but you do get a bunch of children started.

Sincerely,

William Mussatto, Senior Systems Engineer
CyberStrategies, Inc
ph. 909-920-9154 ext. 27