Bug #46268 for bind9-public: filter-aaaa system test failing in maintenance branches

The system test used its local root server NS as the NS for its test zones, so changing the way root server addresses are cached caused a change in the way the additional section worked for the test zones. Changing the NS names for the test zones to something not maching ./NS fixes the test.

---

rt46268 contains a correct fix, though for consistency, would it not be better (and perhaps easier?) to cherry-pick parts of commit 03be5a6b4e6 (Mukund's performance improvements which were not backported to 9.11 and older branches)? E.g. like this: $ git checkout 03be5a6b4e6 bin/tests/system/filter-aaaa/ns*/*.db{,.in} $ git commit This way the filter-aaaa system test will also start passing again, but in case we introduce any further changes to it on master, they will be easier to backport. It is just a thought, though; rt46268 is good to go as it is if you like it more. PS. While v9_10 and v9_11 fail the filter-aaaa system test consistently, v9_9 failures are intermittent on Robie (and I cannot reproduce them at all on my machine). Do you know why this is?

---

> rt46268 contains a correct fix, though for consistency, would it not be > better (and perhaps easier?) to cherry-pick parts of commit 03be5a6b4e6 > (Mukund's performance improvements which were not backported to 9.11 and > older branches)? E.g. like this: > > $ git checkout 03be5a6b4e6 bin/tests/system/filter-aaaa/ns*/*.db{,.in} > $ git commit Ah I hadn't noticed those changes. Yes, that should work equally well. > PS. While v9_10 and v9_11 fail the filter-aaaa system test consistently, > v9_9 failures are intermittent on Robie (and I cannot reproduce them at > all on my machine). Do you know why this is? Nope.

---