Removing a reverse DNS zone forces replication to stop by error
In listener.log you might see entries like this:
31.01.19 11:39:09.412 LISTENER ( PROCESS ) : updating 'zoneName=168.192.in-addr.arpa,cn=dns,dc=domain,dc=com' command m 31.01.19 11:39:09.557 LISTENER ( ERROR ) : dntree_del_id: delete failed: subordinate objects must be deleted first 31.01.19 11:39:09.557 LISTENER ( ERROR ) : cache_update_master_entry: storing master entry in database failed 31.01.19 11:39:09.557 LISTENER ( ERROR ) : cache.c:429:cache_update_master_entry mdb_txn_commit: failed: MDB_PAGE_NOTFOUND: Requested page not found (-30797) 31.01.19 11:39:09.557 LISTENER ( WARN ) : received signal 6
This is a (currently unresolved) bug.
As a workaround as long this bug is unresolved you will have to re-join your servers which will flush the local listener cache and therefore fix the error in the local database.
Note: In case your join is too slow you might be able to speed it up based on this article.