Whenever <FONT FACE=monospace>ircd</font> gets a new connection, does it fork a new server process? ie, when you've got 40 users connected, do you have 40 <FONT FACE=monospace>ircd</font> processes?<br><br>Also, is <FONT FACE=monospace>ircd</font> running under inetd?<br><br>If it's "yes" to both questions, then you may have hit a built in <FONT FACE=monospace>inetd</font> default. To prevent runaway processes bringing your system to a grinding halt, <FONT FACE=monospace>inetd</font> limits the number of instances of a daemon that may be started over a set time period. The default if for <b>40</b> connections over a 60 second period.<br><br>You can change this. The following is taken from the <FONT FACE=monospace>inetd</font> man page on one of our Unix boxes here. (No Linux boxes to hand with man pages installed at the moment...)<br><FONT FACE=monospace><br> The -i option allows you to set the time interval (in<br> seconds) over which the number of daemons started by inetd<br> will be counted. The -n option allows you to specify the<br> number of instances of the same daemon that may be started<br> during the interval specified by -i. The defaults are 40<br> instances in 60 seconds.<br><br> An example of the use of -n and -i is:<br><br> inetd -i 30 -n 50<br><br> which specifies that up to 50 instances of the same daemon<br> may be started in a 30 second period before inetd decides<br> that something is wrong and makes the daemon unavailable.<br> The daemon will become available again after a fixed period<br> of ten minutes or if SIGHUP is sent to inetd. These options<br> are likely to be needed only when many connections are being<br> opened and closed.<br></font><br>Note - changing these values may result in opening up your box to DoS attacks. Maybe you should look at running <FONT FACE=monospace>ircd</font> outside of <FONT FACE=monospace>inetd</font> if possible.