Debugging kdcs and signals

[Greg Hudson]

On Sat, 2011-09-10 at 15:30 -0400, Sam Hartman wrote:
> With the verfto merge, I find that when I'm debugging the KDC, I tend to
> get interrupted by signals a lot more than I used to.

I haven't noticed this yet, and there are no kill() calls in libev or
libverto.  So, it would be helpful to collect more information.

The one case where I've seen something like this is when I'm debugging a
k5test.py or dejagnu test case and I let the script terminate.  For
example, if I use k5test.py's --stop-before=N, attach to the KDC, and
then let the script continue, the client program will typically time
out, causing the script to fail out and kill the KDC.  I work around
this by using --stop-after=N in addition to --stop-before=N.  I would
guess that this isn't the case you're seeing, but I'm mentioning it just
in case.