Hello,

This is probably not really specific to NetBSD, but I would hate to blame
BIND right off. :)

One of our DNS servers occasionally has a problem where 'named' grows 
continously until it hits the process limits and then core dumps.
On a process that was restarted a couple of hours ago, here is what I
am seeing:
14:31:36
  PID USERNAME PRI NICE   SIZE   RES STATE   TIME   WCPU    CPU COMMAND
21143 root       2    0   114M  114M sleep   2:14  2.59%  2.59% named

if I do 'ndc dumpdb' the result is:
Jul 10 14:32:24 host named[21143]: dumping nameserver data
Jul 10 14:32:25 host named[21143]: finished dumping nameserver data
...
-rw-r--r--  1 root     cvs    3492393 Jul 10 14:32 named_dump.db

So, I have a 114 Megabyte process with slightly over 3 Megabytes of
data.

And a few minutes later:
14:37:45
  PID USERNAME PRI NICE   SIZE   RES STATE   TIME   WCPU    CPU COMMAND
21143 root       2    0   117M  117M sleep   2:21  0.05%  0.05% named

Hmm, 3 Megabytes in 6 minutes.

This machine is running NetBSD/i386 1.4.2_ALPHA from Jan 21.  I haven't
had an opportunity to update it to the latest on the release branch, but
I haven't seen any commits related to BIND either.

This machine is only servicing one domain with around 5 subdomains.  It
has dynamic DNS enabled on one subdomain (and the reverse) which may be
the problem.

Any ideas for debugging this?  I can get a core dump at any point, or attach
to the process with gdb I suppose, but I don't know what to look for.

-Andrew
-- 
-----------------------------------------------------------------
Andrew Gillham                            | NetBSD ist Affengeil.
gillham@whirlpool.com                     | Nachts ist es kaelter
I speak for myself, not for my employer.  | als draussen.