Problems with Bind + DLZ - stops working

Daniel fangorn at o2.pl
Fri Jul 8 11:25:12 UTC 2005


Hello!

I have recently encountered a very weird problem with Bind (9.2.3, updated
now to 9.2.5, but the problem occured also soon after the upgrade) patched
to support DLZ, with MySQL as backend. It just stopped working, it didn't
resolve any hostnames, even for itself. But the process was running and
the logs did not show anything. What's even more concerning, both master
and slave have shown such behavior. 

The quick solution appears to be commenting out the dlz-mysql
configuration in named.conf, reloading (killall -HUP named), then
uncomment and reload again. And everything is up and running, at least for
several hours. There is no regularity regarding time of failures,
restarting the servers does not help at all (I have to do the
comment/uncomment trick even after the reboot). 

Both servers run slackware, local mysql databases contain the dns_data.
I thought this might be some kind of DoS, but tcpdump did not show
anything suspicious. I guess this might also be MySQL problem, but I'm not
really sure how come both servers, both databases (no relation between
them) should stop working all of a sudden. Was planning to run the daemons
with strace, but I'm not sure I'll be able to analyze that massive amount
of data. I also ran mysql repair, in order to be sure everything is ok
with the data.

There were no recent changes to the system (despite adding new records to
databases, but that's quite obvious, and harmless I guess).

Anyone has seen such behavior? I'll be greatful for some directions what
to look for, as I seem to be lost. Luckily, now everything works fine, but
I'm affraid this weekend might be a tough time for me :-/.

Best regards,
Daniel



More information about the bind-users mailing list