Jelte Fennema-Nio

Sounds like you're using a libpq from before bbf9c282ce92272ed7bf6771daf0f9efa209e61b But even after that libpq will complain. Just with a more helpful message. Better handling is in my patchset here: https://www.postgresql.org/message-id/flat/CAGECzQScQ3N-Ykv2j4NDyDtrPPc3FpRoa%3DLZ-2Uj2ocA4zr%3D4Q%40mail.gmail.com#cd9e8407820d492e8f677ee6a67c21ce

Slightly changing the output from pg_dump should allow you to work arround this issue. With something like this you should be able to automate that (untested command). ``` pg_dump |...

@philip-harvey can you explain in what other ways it is a breaking change for you?

I'll try to take a look at this soon, thank you for your work on this.

Could you run `citus_indent` on your PR? It's hard for me to see what code is actually changed because some tabs have been changed to spaces. Instructions to run it...

FYI I pushed a commit to your branch to reindent it.

The build error on PG16 seems to be because you included `#include "distributed/connection_management.h"` after `pg_versioncompat.h` in `maintenanced.c`. (this is arguably a bug in in how we define has_createdb_privilege in `pg_versioncompat.h`,...

I haven't looked at it in more detail yet, but at least the flaky test detection is failing because this test randomly fails: ```diff +++ /__w/citus/citus/src/test/regress/results/multi_maintenance_multiple_databases.out.modified 2024-02-02 13:40:41.496394020 +0000 @@...

Yes if you go to the overview and scroll to the bottom you will see all the artifacts produced and also the diff of the failures inline: https://github.com/citusdata/citus/actions/runs/7812230917/attempts/2#summary-21309102210

The flaky tests you might be able to reproduce locally by running the test in question multiple times (this runs it 200 times): ``` src/test/regress/citus_tests/run_test.py multi_maintenance_multiple_databases --repeat 200 ```