I have been dealing with memory leaks on cRIO for a looong time now. When cRIO runs out of memory, in some cases RT Watchdog does not trigger, the program halts but cRIO does not restart. I tried to develop a Linux watchdog that gets launched and killed by LabVIEW via System Exec, but since LabVIEW is set not to have permissions to launch processes, I need to use it on startup and LabVIEW needs to communicate with watchdog over file where timestamp is saved - if it does not change, LabVIEW must be dead and cRIO is restarted.
There is a major problem with doing it. I follow basic rules for setting the process as startup through SSH terminal. I copied bash script to /etc/init.d, ran these commands:
chmod +x /etc/init.d/test.sh
chmod 777 /etc/init.d/test.sh
/usr/sbin/update-rc.d -f test.sh defaults 99 0
Symbolic links to process and privileges are added, but no process is running on startup. The weirdest thing is that we have another bash script which tweaks NTP configuration, it launches ntpd process as a result, and this one works! Furthermore, this test.sh works perfectly when run directly or in the background with nohup command. No success even when calling the script with nohup from the NTP script that works. I enclose both scripts, I would be glad if you had any ideas.
Thomas
