CPU Usage

[Gushi]

That is to say, what stats are useful in debugging this issue?  The
memstats file doesn't seem to help.  Regular stats dumps don't seem to
help.  Running under strace (or more likely, truss) would be a hell of
a capture because it can take DAYS to go horribly wrong.

If I turn on query logging (assuming one abusive resolver customer is
enough to do this) -- is there some tool that can help me analyze
those stats?

If I dump the cache, is there something that can give me an analysis?

Is there a better debug level I should be using?

Should I file a bug report?

Is there some way I can log why (according to top) it's eating 96
percent of the CPU and all the memory the OS will give it? Or why
Nagios is telling me it's not responding?  Or why rndc and/or any
"kill" other than -9 does nothing about half the time?

This is under FreeBSD.  One of the reasons we use it is because it's
the same platform the  ISC uses itself for the F-root, and figure it
would help to get a good perspective.  (i.e. "if it's good enough for
the ISC...)