Kees Monshouwer

IMHO d_maxchunkrecords(1) is a not too bad workaround for the very few people that are affected by this, but is should be optional as it might hurt for others. Optional...

'published' and 'active' changes for existing keys should also result in a serial increase

Quick observation soa-edit is lower cased.

> ### Expected behaviour > > No error should be generated as no configuration telling the server to send NOTIFY to the NS targets was provided. PowerDNS is notifying NS...

Empty only-notify will save some resources since it will disable the useless resolving. See https://doc.powerdns.com/authoritative/settings.html#only-notify for the exact details.

#12888 is likely related

This is the result of enabled consistent-backends with a disabled query cache. consistent-backends=no will restore the "old" behavior. https://doc.powerdns.com/authoritative/settings.html#setting-consistent-backends

Increasing the serial before rectify creates a race where a notify is sent before the zone is rectified. In backends with limited transaction support, like lmdb, this wil result in...

I like the idea of dnssec and ALIAS. There is however something you cannot ignore. Using stubDoResolve will make ALIAS much more sensitive to exhausting all available backends. There are...

I think documenting the interaction of SOA-EDIT with the automatic soa calculation is sufficient. I can imagine SOA-EDIT can be useful is setups with very few catalog zone updates to...