In the last episode (Nov 27), Chris Elsworth said:
> Most of the time, MySQL 4.0.16 runs absolutely fine on my FreeBSD 4.9
> system. However, today, unexplicably, it's run into the same problem
> 3 times now; a couple of the threads suddenly start eating 100% CPU
> for no good reason while doing a SELECT. They'll sit there until I
> kill them, which results in a lot of hair tearing as it comes back up
> checking all the tables.
> I haven't been able to produce a core despite sending it a SIGABRT,
> but I'll try recompiling with -g (do the supplied binaries have this?)
You can use the gcore command to create a coredump of a running program
without killing it, and the pstack command (in ports) is handy for
printing the stacks of all threads in a running process without even
coring it. Once you get a stack trace, send it to
freebsd-threads@stripped and/or here.
> Pretty much all I can find around that might contribute relates to
> the poor threading implementation on FreeBSD and the name resolving
> issue. I should have worked around both of them; I've built with
> linuxthreads, and I have skip-name-resolve in my.cnf.
You should only need one or the other, actually.