Excessive notify traffic
support ticket #20530 reports excessive NOTIFY traffic after an upgrade from 9.16 to 9.18.
I have reproduced this behaviour on a test VM: all it requires is a non-responsive NOTIFY recipient. It appears that the overall request timeout is not being applied.
There are a number of other similar patterns in
zone.c related to DS checking and NS queries, which presumably have the same problem.
This regression was introduced by !4601 (merged) (use netmgr for dispatch). I have not pinned down exactly the point in that branch where it stops working because there is other breakage that makes it difficult.