Thanks Gustaf,
I will look into compiling naviserver with the -g option to see if we can generate a core file that can generate a backtrace with gdb.
We had started a standard some time ago to always restart the server after an upgrade because of some oddities some developers were seeing. We believe it was when a developer would instantiate a proc in the shell (for debug purposes) that the issue would arise. After talking to Tony about it we think that you are right and we should just use the "Reload" option during day releases
Your point is well taken about caching. It would be interesting to disable the caching for testing purposes. Is there a parameter that does this or a compile option? Or would we need to disable it for each type of cache by modifying the code?
Thanks for your help,
Marty