bind9 trouble in -PRERELEASE

Jeremy Chadwick freebsd at jdc.parodius.com
Wed Nov 1 21:18:05 UTC 2006


On Wed, Nov 01, 2006 at 11:40:39PM +0300, Nguyen Tam Chinh wrote:
> I don't think that I can do a stack backtrace when named left no core 
> after that.

No core is being left either due to limits (limit/ulimit) being
imposed on the user or group 'bind', or because chroot precautions
are being used.

The way to work around this, assuming the box isn't being used by
end-users for application development (thus needing their own
corefiles for their apps), is to do the following in sysctl.conf:

kern.sugid_coredump=1
kern.corefile=/some/absolute/path/%N.%P.core

You can adjust these in realtime with sysctl as well.

Be sure to specify an absolute path that the 'bind' account has
write access to, and is big enough to fit a decent-sized coredump.
If you have a large /var filesystem, /var/tmp is an OK place.

-- 
| Jeremy Chadwick                                 jdc at parodius.com |
| Parodius Networking                        http://www.parodius.com/ |
| UNIX Systems Administrator                   Mountain View, CA, USA |
| Making life hard for others since 1977.               PGP: 4BD6C0CB |



More information about the freebsd-stable mailing list