Getting stack trace from runaway apache

In our HP-UX 11 production environment, we use apache 2 servers as
front-ends for BEA Weblogic servers via BEA's weblogic plugin.Under some
conditions, an apache thread can go into a runaway condition, consuming
90% of a cpu and spending 30% of its time in system calls. BEA wants a
stack trace to help with their investigation.

We are trying to find out if there is a way of generating a stack trace
for the thread. Kill -3 does not seem to work. Googling around, the only
lead indicates we have to set up a debuggable server and use gdb to
attach to it and examine the stack. This solution is problematic in that
the condition is rare and only happens in production, and we are loathe
to run a debuggable apache in production.

Is there an easy way to generate a stack trace for the httpd process
hosting the runaway thread?