blackhole option anomaly

[Thib D]

Hello,

I am currently rolling a few of my servers from 9.16.25 to 9.18.0 and I'm
having a few issues :

For context, my test setup is basically a primary server transferring all
of its zones through standard AXFR to a secondary BIND server on a
different IP. Both servers are running the same Bind version.

After rolling the 9.18.0 on both servers, I saw these anomalies :


   - the primary stopped sending notify queries to the secondary server at
   each zone reload
   - the secondary server was able to resolve AXFR queries to the primary,
   but :
   - rndc commands like "refresh" or "retransfer" were not interpreted and
   the zone.
   - After the "refresh" timer set in the SOA expired, the zone would not
   update on the secondary either.

Meanwhile there was no log hinting at a misconfiguration, which made the
debugging and understanding quite tricky.

The only fix I found was to comment that particular line in the config :

blackhole { none; };


Correct me If I'm wrong, but there is no mention of any change for this
option, other than this bugfix from the early 9.17 :
https://gitlab.isc.org/isc-projects/bind9/-/issues/1936

Using actual random IP's in the blackhole field also fixed the issue and re
- enables the zone transfers.

After some digging, it looks like the issue appeared in Bind 9.17.19

Hope it helps,
Thanks
Thibaud.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20220215/d080b82a/attachment.htm>