If you are running FC5 on your IRLP node, chances are, you have already experienced some un-explained node lock-ups.  This is very frustrating and can be time consuming depending on the accessibility of your IRLP node. 

It usually always happens upon disconnecting from another node or reflector.  The problem has been mentioned on the IRLP groups but to my knowledge, nobody is actually engaged in tracking down and fixing it.  So, I decided to throw together a simple watchdog for IRLP nodes that are experiencing this issue.  It actually employs the Linux Kernel's software watchdog capability to accomplish the task.  My watchdog utility consists of an executable timer that kicks the watchdog device at a particular interval, and an init script that configures the kernel watchdog upon booting.  If your node is not in an easily accessible location (such as at a repeater site, etc.) or if you're simply tired of having to manually reboot your node to get it to come back, then my watchdog may come in handy on your node.  I recently rewrote the watchdog timer in C for better performance and smaller memory footprint.  It also fixes the kernel watchdog warnings that were being sent to the console.  You can download the latest version of my watchdog for FC5 IRLP nodes here:

http://rob.pectol.com/irlp/kk7av-watchdog.tgz 

After downloading it to your node, enter the directory that contains it and simply do a, "tar -zxvf ./kk7av-watchdog.tgz" and it will unpack to a directory called, "watchdog."  Inside it, you will find 2 files and a README.txt file that explains how to install the watchdog and make it active on your node.

So far, it has successfully recovered my node after several hard lock-ups.  Until they figure out what the problem is and get it fixed, this seems to be a decent work-around.  Enjoy!